Tall buildings using CLT. An integrated design considering moisture induced effects

Catarina Vilaça Silva outubro de 2020 UMinho | 2020 Tall buildings using CLT. An integrated design considering moisture induced effects Universidade do Minho Escola de Engenharia Catarina Vilaça Silva Tall buildings using CLT. An integrated design considering moisture induced effects outubro de 2020 Tese de Doutoramento Trabalho efectuado sob a orientacao de Professor Jorge Manuel Goncalves Branco Professor Paulo Jose Barbosa Lourenco Catarina Vilaça Silva Tall buildings using CLT. An integrated design considering moisture induced effects Universidade do Minho Escola de Engenharia DIREITOS DE AUTOR E CONDIÇÕES DE UTILIZAÇÃO DO TRABALHO POR TERCEIROS Este é um trabalho académico que pode ser utilizado por terceiros desde que respeitadas as regras e boas práticas internacionalmente aceites, no que concerne aos direitos de autor e direitos conexos. Assim, o presente trabalho pode ser utilizado nos termos previstos na licença abaixo indicada. Caso o utilizador necessite de permissão para poder fazer um uso do trabalho em condições não previstas no licenciamento indicado, deverá contactar o autor, através do RepositóriUM da Universidade do Minho. Licença concedida aos utilizadores deste trabalho Atribuição-NãoComercial CC BY-NC https://creativecommons.org/licenses/by-nc/4.0/ iii ACKNOWLEDGES The present Phd research was the most individual and lonely project I had to develop until now. However, I couldn’t make it happen without the precious help of some important persons who allowed this research to be deleloped. This way, I must thank to some important people: - My advisores, Professor Jorge Branco and Professor Paulo B. Loureço, for all the technical support during definition and execution of present research; - Researcher José Xavier, for all the support related with the experiments based on DIC technique; - Professor Gerhard Schickhofer, researcher Andreas Ringhofer and all technical staff from LAB, who welcome me at Institut für Holzbau und Holztechnologie - Graz Technical University (Austria) and gave me all the conditions and support to conclude the tasks developed in there; - Professor Eduarda Luso, who made available the facilities of Instituto Politecnico de Bragança where we perform some experimental tests; - Technical staff of structures laboratory of Civil Engineering Department at University of Minho for the for the crucial role in realizing a great part of experiments of this thesis, specially to my dear husband Marco Jorge, who had all the patience when the work didn’t seem to go well; - My closest family, parents and syster, for all the emotional support; - my sweet daughters, Maria and Inês, who were born when this research was under development making it longer than expected. - At last, I must thanks to Fundação para a Ciencia e Tecnologia that fund present research by means of the scholarship with the reference SFRH / BD / 79972 / 2011, funded by POPH - QREN - Tipology 4.1 - Advanced Training, co-funded by the European Social Fund and MEC national funds. STATEMENT OF INTEGRITY I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the University of Minho. v Edifícios em altura com MLCC. Um projeto integrado considerando os efeitos da humidade. RESUMO Conscientes das vantagens associadas ao uso da Madeira Lamelada Colada Cruzada (MLCC) nas construções localizadas nos grandes centros urbanos e considerando a falta de conhecimento relativo aos efeitos resultantes de variações de humidade nesse mesmo material, a presente tese foi dividida em duas partes complementares. A Parte I é dedicada ao desenvolvimento de uma vasta campanha experimental focada essencialmente na quantificação dos esforços induzidos pela humidade na MLCC e na quantificação da resistência ao arrancamento de parafusos auto perfurantes inseridos em painéis de MLCC. Os esforços induzidos pela variação no teor de humidade foram quantificados considerando um intervalo de humidade relativa entre 30% e 90%, sendo utilizadas três técnicas de medição diferentes, nomeadamente: Correlação Digital de Imagem, sensores de deslocamento (LVDT) e paquímetro. No que respeita á quantificação da capacidade de arrancamento dos parafusos auto perfurantes, foi desenvolvida uma campanha experimental na qual foram efetuados 590 testes de arrancamento. De modo a compreender melhor a relação entre MLCC e os parafusos auto perfurantes, foram tidos em conta três parâmetros principais: (i) mudanças simples e cíclicas no teor de humidade, (ii) número de gaps e (iii) a largura dos mesmos gaps . A parte II apresenta uma proposta para um sistema estrutural para a construção de edifícios de vários pisos, no qual a MLCC é o principal material de cariz estrutural. Este sistema foi apelidado de Urban Timber (UT) system e foi desenhado tendo sempre presente os efeitos causados pelas variações no teor de humidade quer nos elementos individuais de MLCC, nas ligações mecânicas ou nas soluções propostas para as fachadas. Paralelemente, o UT system foi submetido a uma avaliação estrutural e as suas potencialidades arquitetónicas foram exploradas. Palavras-chave: Madeira Lamelada colada cruzada; parafusos auto perfurantes; efeitos induzidos pela humidade; construção de madeira em altura; resistência ao arrancamento. Tall buildings using CLT. An integrated design considering moisture induced effects. ABSTRACT Aware about the advantages of using Cross Laminated Timber (CLT) as a construction material on the urban environments, and considering the lack of knowledge related with the behaviour of CLT regarding moisture induced effects, present thesis was developed in two parts that complement each other. Part I is based on the development of laboratorial experiments focused on the quantification of moisture induced strains on CLT panels and on the quantification of withdrawal capacity of Self Tapping Screws (STS) inserted in CLT panels. Moisture induced strains were quantified considering a range of relative humidity that varies between 30% and 90% and three different measure techniques were used, namely: Digital Image Correlation (DIC), LVDT’s acquisition and Calliper ruler measurements. Regarding quantification of withdrawal capacity of STSs, a large experimental campaign comprising 590 withdrawal tests was carried out. In order to understand deeply the composite model “CLT-STS”, tests performed considered three main parameters: (i) simple and cyclic on moisture changes, (ii) number of gaps and (iii) the width of gaps. Part II present a proposal for a new structural system for multi-story timber buildings, in which CLT is the main structural material. The proposed structural system was called Urban Timber (UT) system and it was designed always taking into account moisture induced effects either on individual structural elements, mechanical connections or on facade solutions. Independent of limitations imposed by moisture induced effects, a structural evaluation of UT system is performed and architectural potentialities are presented individually. Keywords: Cross laminated timber; self-tapping screws; moisture induced effects; multi-storey timber buildings; withdrawal strength. vii CONTENTS ACKNOWLEDGES ........................................................................................................................ iii RESUMO ...................................................................................................................................... v ABSTRACT .................................................................................................................................. vi CONTENTS................................................................................................................................. vii FIGURE INDEX ........................................................................................................................... xiii TABLE INDEX ........................................................................................................................... xxiv LIST OF ABBREVIATIONS AND ACRONYMS .............................................................................. xxviii Introduction ............................................................................................................................... 31 PART I : QUANTIFICATION OF MOISTURE INDUCED STRAINS IN CROSS LAMINATED TIMBER AND THEIR EFFECTS ON WITHDRAWAL CAPACITY OF SELF-TAPPING SCREWS ........................................... 39 1 State of the art ................................................................................................................... 41 1.1 Cross laminated timber and moisture changes............................................................ 42 1.1.1 Absorption and sorption phenomenon ................................................................. 42 1.1.2 Effects of hygroscopic behavior of wood on CLT .................................................. 43 1.2 CLT connections ........................................................................................................ 49 1.2.1 The role of connections on timber construction ................................................... 49 1.2.2 Self-tapping screws and withdrawal resistance ..................................................... 51 1.2 Digital image correlation - DIC .................................................................................... 54 1.2.3 Fundamentals .................................................................................................... 54 2 Quantification of moisture induced strains in cross laminated timber .................................. 57 2.1 Variables and techniques involved .............................................................................. 58 2.2 Production and preparation of tests specimens and sampling ..................................... 60 2.3 Test setup and test procedure .................................................................................... 63 2.3.1 DIC technique..................................................................................................... 63 Figure (I) 2:10. Measurements made on different test days for specimens from group A (dimensions in mm). ............................................................................................................................................ 67 Figure (I) 2:11. Measurements made on different test days for specimens from group B (dimensions in mm). ............................................................................................................................................ 67 Figure (I) 2:12. Test setup for measuring shrinkage and swelling by means of LVDTs. At left side drawing of test setup (dimensions in mm), and at right side photos of test setup inside the climatic chamber. ......................................................................................................................................... 69 Figure (I) 2:13. Manual measurements performed by caliper ruler on specimens from group A (dimensions in mm).......................................................................................................................... 70 Figure (I) 2:14. Relative humidity and temperature registered by Fitoclima 1000EC45 and Fitoclima 28000 during 324 days. ................................................................................................................... 71 Figure (I) 2:15. Moisture content registered by control specimens during 324 days........................ 71 Figure (I) 2:16. Boxplots of moisture content obtained for outer layers of all specimens tested considering different test configurations and all test days. .................................................................. 73 Figure (I) 2:17. Boxplots of moisture content obtained for inner layers of all specimens tested considering different test configurations and all test days. .................................................................. 73 Figure (I) 2:18. Boxplots of corrected density (ρ12) obtained for outer layers of all specimens tested considering different test configurations and all test days. .................................................................. 74 Figure (I) 2:19. Boxplots of corrected density (ρ12) obtained for inner layers of all specimens tested considering different test configurations and all test days. .................................................................. 74 Figure (I) 2:20. Mean values of full-field restrained strains distribution measured on F1 of specimens from group A for all four configurations in Y and X directions. ............................................................ 76 Figure (I) 2:21. Mean values of released strains measured on central longitudinal section of all slices (S1-S5) of specimens from group A for all four test configurations. .................................................... 82 Figure (I) 2:22. Mean values of full-field restrained strains distribution measured on specimens from group A_F2 in Y direction, for inner and outer layers, during first DIC mapping. ................................. 84 Figure (I) 2:23. Mean values of full-field restrained strains distribution measured on specimens from group A_F2 in Z direction, for inner and outer layers, during first DIC mapping. ................................. 88 Figure (I) 2:24. Mean values of full-field restrained strains distribution measured on specimens from group A_F3 in X direction, for inner and outer layers, during first DIC mapping. ................................. 89 Figure (I) 2:25. Mean values of full-field restrained strains distribution measured on specimens from group A_F3 in Z direction, for inner and outer layers, during first DIC mapping. ................................. 91 xv Figure (I) 2:26. Mean values of full-field restrained strains distribution measured on specimens from group B in Y direction, for inner and outer layers, during first DIC mapping. ....................................... 93 Figure (I) 2:27. Mean values of full-field restrained strains distribution measured on specimens from group B in Z direction, for inner and outer layers, during first DIC mapping. ....................................... 94 Figure (I) 2:28. Mean values of full-field released strains distribution measured on specimens from group B in Y direction, for inner and outer layers, during second DIC mapping, after cut specimens (AC). ...................................................................................................................................................... 100 Figure (I) 2:29. Linear fittings between  and  performed to obtain best correlation for entire CLT slice. ....................................................................................................................................... 105 Figure (I) 2:30. Linear fittings between  and  performed to obtain a correlation for different CLT layers considering their grain direction. .................................................................................... 106 Figure (I) 2:31. Mean values of released stresses measured on longitudinal section of each slice (S1S5) of specimens from group A for all four test configurations. ........................................................ 107 Figure (I) 2:32. Mean values of released stresses measured on longitudinal section of all five slices of specimens from group A for all four test configurations and all five test days. ................................... 108 Figure (I) 2:33. Mean values of full-field released stresses distribution measured on specimens from group B in Y direction, for inner and outer layers, during second DIC mapping. ................................ 109 Figure (I) 2:34. Mean values of released stresses measured on outer and inner layers of specimens from group B for all four test configurations and all five test days. .................................................... 110 Figure (I) 2:35. Linear shrinkage considering initial dimension (day 0) of test specimens. ............ 112 Figure (I) 2:36. Curves of compressive strains considering initial dimension of each cycle with 30% RH. (a) X direction; (b) Y direction; (c) Z direction. .................................................................................. 114 Figure (I) 2:37. Curves of tensile strains considering initial dimension of each cycle with 90% RH. (a) X direction; (b) Y direction; (c) Z direction. .......................................................................................... 115 Figure (I) 2:38. Curves comparing restrained strains obtained by DIC technique and LVDTs measurements. (a) X direction; (b) Y direction; (c) Z direction. ......................................................... 117 Figure (I) 2:39. Mean values of mean released strains measured on specimens from group A in Y direction, for inner and outer layers, using digital caliper ruler and considering measurements taken on all five slices. .................................................................................................................................. 118 Figure (I) 2:40. Released strains obtained for inner layers for all four configurations and for all five slices.............................................................................................................................................. 120 Figure (I) 2:41. Released strains obtained for outer layers for all four configurations and for all five slices.............................................................................................................................................. 121 Figure (I) 3:1. Different test configurations and sampling used for different groups and test days (dimensions in mm)........................................................................................................................ 128 Figure (I) 3:2. CLT panels produced at Laboratory. (a) Four different types of boards needed to produce CLT panels for different test configurations; (b) different configurations of CLT panels. ..................... 129 Figure (I) 3:3. Similar density distribution between different configurations. Graphs were plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). .......................................... 130 Figure (I) 3:4. Procedure to build CLT panels in laboratory. (a) assembling line and glue used to bond layers (PURBOND® HB110); (b) hydraulic pressing machine (LANGZAUNER) bonding two CLT layers; (c) gaps with four millimeters protected to the next bonding procedure; (d) glue line (MINDA equipment) with glue dispenser and timber boards with glue before assembly panel; (e) pressing the third layer; (f) final CLT panels before saw the external gaps. (Note: dimensions are in mm) ......................................... 131 Figure (I) 3:5. Procedure to finish test specimens. (a) test specimens cut from CLT panels; (b) predrilling equipment; (c) STS used to perform withdrawal tests (Rapid® Vollgewinde from Schmid with diameter of 8mm and length of 180mm) and the penetration of STS through entire specimen thickness until avoids the tip effect. ................................................................................................................ 132 Figure (I) 3:6. Similar density distribution for different MC groups. Graphs were plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). ........................................................... 133 Figure (I) 3:7. Relation between temperature and relative humidity for reaching different moisture contents for spruce, suggested by Hartl & Ramberger (1985). ......................................................... 134 Figure (I) 3:8. 2D drawing of test setup. ...................................................................................... 135 Figure (I) 3:9. Photos of test setup. (a) lateral view; (b) front view................................................. 136 Figure (I) 3:10. Equipment involved in test procedure. (a) test machine; (b) and (c) device to ensure the application of the withdrawal force along the screw axis and hold screw head; (d) required free area around the screw axis. .................................................................................................................... 136 Figure (I) 3:11. Moisture distribution for group with 12% of moisture content. Graph was plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). .......................................... 138 Figure (I) 3:12. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_12% and GAP0_12%; (b) REF_12% and GAP4_12%. ...................................................................................................................................................... 141 xvii Figure (I) 3:13. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_12% and GAP_FL_12%; (b) REF_12% and GAP_ML_12%; (c) REF_12% and GAP_OL_12%; and (d) REF_12% and GAP_3L_12%. .................... 141 Figure (I) 3:14. Visual inspection of failures for tests performed with 12% moisture content. (a) typical failure for configurations REF_12%, GAP4_ML_12% and configurations belonging to GAP0_12%; (b) typical failure for remaining configurations belonging to GAP4_12%. ................................................ 141 Figure (I) 3:15. Mean test curves for all configurations with 12% of moisture content, expressing the expected yielding failure mode. ....................................................................................................... 142 Figure (I) 3:16. Moisture distribution for group with 8% of moisture content. ................................ 143 Figure (I) 3:17. Cracking resulted from conditioning period. (a) crack through the STS location; (b) opening of GAP0 on an exterior layer; (c) lateral view of a crack in an outer layer; (d) and (e) lateral view of delamination and crack in middle layer. ...................................................................................... 143 Figure (I) 3:18. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_8% and GAP0_8%; (b) REF_8% and GAP4_8%. ... 145 Figure (I) 3:19. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_8% and GAP_FL_8%; (b) REF_8% and GAP_ML_8%; (c) REF_8% and GAP_OL_8%; and (d) REF_8% and GAP_3L_8%. .................................................... 146 Figure (I) 3:20. Visual failures of tests performed with 8% of moisture content. (a) Visual failure for configurations REF, GAP0 and GAP4_ML; (b) Visual failure for remaining configurations GAP4......... 146 Figure (I) 3:21. Mean test curves for all configurations with 8% of moisture content, expressing the expected yielding failure mode. ....................................................................................................... 147 Figure (I) 3:22. Moisture distribution for group with 18% of moisture content. .............................. 148 Figure (I) 3:23. Damages caused by high MC levels. (a) deformations on CLT shape caused by timber swelling depending on grain direction; (b) closing of GAP4 after conditioning period. ........................ 148 Figure (I) 3:24. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_18% and GAP0_18%; (b) REF_18% and GAP4_18%. ...................................................................................................................................................... 148 Figure (I) 3:25. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_18% and GAP_FL_18%; (b) REF_18% and GAP_ML_18%; (c) REF_18% and GAP_OL_18%; (d) REF_18% and GAP_3L_18%; and (e) REF_18% and GL_18%. ........................................................................................................................................ 149 Figure (I) 3:26. Visual STS failures of tests performed with 18% of moisture content. (a) Visual failure on top of specimen; (b) Visual failure on bottom of specimen. ......................................................... 149 Figure (I) 3:27. Mean test curves for all configurations with 8% of moisture content, expressing a yielding failure mode....................................................................................................................... 151 Figure (I) 3:28. Sketch explaining the reason why density is higher in gap location. ...................... 152 Figure (I) 3:29. GAP0_ML configurations cut to correct the density close to STS location (dimensions in mm). .......................................................................................................................................... 152 Figure (I) 3:30. Relation between density of reference concerning the entire specimen (12) and corrected density considering a small sample of CLT close to STS location (12, !""). ................. 152 Figure (I) 3:31. Linear fittings performed between REF configuration and remaining configurations: (a) configurations with GAP0; (b) configurations with GAP4. kgap and R2values are presented in tables bellow respective graphs. ................................................................................................................ 156 Figure (I) 3:32. Graphs of linear fittings between different moisture levels for all configurations: (a) REF configuration; (b) GAP0 configurations; and (c) GAP4 configurations. kMC and R2 values related with same linear fittings are presented bellow respective graphs. ............................................................ 160 Figure (I) 3:33. Correlation between ,*" and , for: (a) test groups with 8%, 12%, and 18% of moisture content; (b) configurations REF, GAP0 and GAP4. .................................................. 163 Figure (I) 3:34. Correlation between mean values obtained for ,*" and ,. a) mean values for all tested configurations considering MC level and gap width; b) mean values for GAP_FL configurations; c) mean values for GAP_ML configurations; d) mean values for GAP_OL configurations; e) mean values for GAP_3L configurations. ..................................................................................... 164 Figure (I) 4:1. Specimens conditioned inside the climatic chamber. ............................................. 169 Figure (I) 4:2. Cycle performed by climatic chamber in which specimens were conditioned. ......... 169 Figure (I) 4:3. Different test configurations and sampling used for different groups and test days. . 171 Figure (I) 4:4. CLT and GL beams produced at Rusticasa for different test configurations. ............ 172 Figure (I) 4:5. Procedure to build CLT and GL beams using Rusticasa facilities. (a) rectifying timber boards; (b) timber boards passing through glue dispenser; (c) adhesive 1247 from AkzoNobel ; (d) glue dispenser; (e) and (f) CLT and GL beams being mounted in the press structure; (g) and (h) pressing device. ........................................................................................................................................... 173 Figure (I) 4:6. Similar density distribution for different test days. Graphs were plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). ........................................................... 173 Figure (I) 4:7. 2D drawing of the test setup. ................................................................................. 175 xix Figure (I) 4:8. Moisture distribution for group tested at day 0. ...................................................... 176 Figure (I) 4:9. Specimens defined as outliers due to the location of knots on screw path for tests performed on day 0. ....................................................................................................................... 178 Figure (I) 4:10. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_D0 and GAP0_D0; (b) REF_D0 and GAP4_D0. .. 179 Figure (I) 4:11. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_D0 and GAP_FL_D0; (b) REF_D0 and GAP_ML_D0; (c) REF_D0 and GAP_OL_D0; (d) REF_D0 and GAP_3L_D0; and (e) REF_D0 and GL_D0. ............. 180 Figure (I) 4:12. Mean test curves for all configurations tested on day 0, expressing the expected yielding failure mode. (a) nine CLT configurations; (b) REF_D0 and GL_D0. ................................................. 180 Figure (I) 4:13. Moisture distribution for group tested on day 324. ............................................... 181 Figure (I) 4:14. Visual inspection of damages caused by RH cycle on CLT and GL specimens. ..... 182 Figure (I) 4:15. Specimens defined as outliers due to the location of knots on screw path for tests performed on day 324. ................................................................................................................... 184 Figure (I) 4:16. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_D324 and GAP0_D324; (b) REF_D324 and GAP4_D324. .................................................................................................................................. 185 Figure (I) 4:17. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_D324 and GAP_FL_D324; (b) REF_D324 and GAP_ML_D324; (c) REF_D324 and GAP_OL_D324; (d) REF_D324 and GAP_3L_D324; and (e) REF_D324 and GL_D324. .............................................................................................................. 185 Figure (I) 4:18. Mean test curves for all configurations tested on day 324, expressing the expected yielding failure mode. (a) nine CLT configurations; (b) REF_D324 and GL_D324. ............................ 186 Figure (I) 4:19. Relation between density of reference concerning the entire specimen (12) and corrected density considering a small sample of CLT near screw location (12, !""). ................... 186 Figure (I) 4:20. Linear regressions performed between REF configuration and specimens with: (a) gaps with 0 mm; (b) gaps with 4 mm; (c) GL. kgap and R2values are presented in tables bellow respective graphs. ........................................................................................................................................... 190 Figure (I) 4:21. Graphs of linear fittings between different moisture levels for all configurations: (a) REF configuration; (b) GAP0 configurations; and (c) GAP4 configurations. kMC and R2 values related with same linear fittings are presented bellow respective graphs. ............................................................ 192 Figure (I) 4:22. Correlation between ,*" and , for: (a) test performed on DAY 0 and DAY 324; (b) configurations REF, GAP0 and GAP4. ......................................................................... 195 Figure (I) 4:23. Correlation between mean values obtained for ,*" and ,. a) mean values for all configurations tested on DAY 0 and DAY 324; b) mean values for GAP_FL configurations; c) mean values for GAP_ML configurations; d) mean values for GAP_OL configurations; e) mean values for GAP_3L configurations. .................................................................................................................. 195 Figure (I) 5:1. Distribution of test configuration between two different test groups. ........................ 201 Figure (I) 5:2. Similar density distribution for different MC levels. Graphs were plotted based on Normal distribution and Rernard score method - (i – 0,3) / (n + 0,4). .......................................................... 202 Figure (I) 5:3. Moisture distribution for group with 14% of moisture content. Graph was plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). .......................................... 204 Figure (I) 5:4. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: REF_14%, GAP0_14% and GAP4_14%. ........................... 206 Figure (I) 5:5. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_14% and GAP_ML_14%; (b) REF_14% and GAP_BL_14%. ................................................................................................................................ 206 Figure (I) 5:6. Visual inspection of failures for tests performed with 12% moisture content. (a) and (b) typical failure for configuration REF_ML_14%; (c) and (d) typical failure for configuration REF_BL_14%; (e) typical failure for configuration GAP0_BL_14%; (f) typical failure for configuration GAP4_ML_14%. ...................................................................................................................................................... 206 Figure (I) 5:7. Mean test curves for all configurations with 14% of moisture content, expressing the expected yielding failure mode. ....................................................................................................... 207 Figure (I) 5:8. Moisture distribution for group with 25% of moisture content. Graph was plotted based on Normal distribution and Rernard score method - (i – 0,3) / (n + 0,4). ......................................... 209 Figure (I) 5:9. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: REF_25%, GAP0_25% and GAP4_25%. ........................... 209 Figure (I) 5:10. Box chart with notched boxes of , and scatter plot of density of reference and moisture content for the following groups: (a) REF_25% and GAP_ML_25%; (b) REF_25% and GAP_BL_25%. ................................................................................................................................ 209 Figure (I) 5:11. Mean test curves for all configurations with 25% of moisture content, expressing the expected yielding failure mode. ....................................................................................................... 210 xxi Figure (I) 5:12. Linear fittings performed between REF configuration and remaining configurations: (a) configurations with GAP0; (b) configurations with GAP4. k+* and R2values are presented in tables bellow respective graphs. ................................................................................................................ 213 Figure (I) 5:13. Graphs of linear fittings between different MC levels for all configurations: (a) REF configurations; (b) GAP0 configurations; and (c) GAP4 configurations. kMC and R2 values related with same linear fittings are presented bellow respective graphs. ............................................................ 216 Figure (I) 5:14. Correlation between ,*" and , for: (a) test groups with MC=14% and MC=25%; (b) configurations REF, GAP0 and GAP4. ......................................................................... 218 Figure (I) 5:15. Correlation between mean values obtained for ,*" and ,. a) mean values for all tested configurations considering MC level and gap width; b) mean values for GAP_ML configurations; c) mean values for GAP_BL configurations. .............................................................. 219 Figure (II) 1:1. Scheme explaining how tall timber buildings fit in the predicted urban context and how they answer to the three dimensions of sustainability. ..................................................................... 232 Figure (II) 1:2. Legal limitations imposed by building codes regarding number of floors for timber buildings (Bo et al. , 2014) (Green and Eric Karsh, 2012) (Östman and Källsner, 2011). .................. 236 Figure (II) 1:3. Wood building initiatives all around the world. ....................................................... 237 Figure (II) 1:4. Crosswise lamination of CLT panels. .................................................................... 239 Figure (II) 1:5. Sawing pattern of the outer boards of a log. Image adapted from (Augustin, 2008). ...................................................................................................................................................... 239 Figure (II) 1:6. Three different spatial configurations based on plate properties suggested by (Bejder, 2012). (a) Enclosed box, (b) dissolved box and (c) ‘floating’ structure. ............................................. 248 Figure (II) 1:7. Chronological evolution of large timber buildings/structures. ................................ 249 Figure (II) 1:8. Stadhaus, 24 Murray Grove, London – UK. (a) external view of building; (b) excessive compartmentalization of structural solution. .................................................................................... 254 Figure (II) 1:9. Bridport building, Hackney London, UK. (a) External view; (b) CLT load bearing structure. ...................................................................................................................................................... 254 Figure (II) 1:10. Forté, in Melbourne’s Docklands, Australia. (a) 3D external rendering; (b) Picture of works on site. ................................................................................................................................. 254 Figure (II) 1:11. Via Cenni, Milan, Italy. (a) external 3D rendering (b) vertical cross section. ......... 254 Figure (II) 1:12. Limnologen project, Växjö, Sweden. (a) External view of two of four buildings; (b) floor elements (Serrano, 2009). .............................................................................................................. 255 Figure (II) 1:13. LifeCycle Tower ONE. (a) 3D digital external rendering of the LCT ONE in daylight; (b) Hybrid structural system. ................................................................................................................ 255 Figure (II) 1:14. Wagramerstrasse timber building. (a) 3D external rendering; (b) Picture of works on site. ................................................................................................................................................ 256 Figure (II) 1:15. Wood Innovation Design Center. (a) External view of building; (b) picture of inner space. ............................................................................................................................................ 256 Figure (II) 1:16. Brock Commons. (a) External view of building; (b) hybrid structural system. ....... 256 Figure (II) 1:17. Barents House Project. (a) 3D external rendering; (b) 3D structural rendering. .... 256 Figure (II) 1:18. FFTT system (a) 30 storey timber building model proposed by mgb Architecture + Design; (b) Hybrid timber-steel solution. .......................................................................................... 256 Figure (II) 6:1. Simplified Scheme of the UT system identifying main timber structural elements. . 260 Figure (II) 6:2. Detail of how to hide lamination lines on CLT walls. .............................................. 261 Figure (II) 6:3. Simplified Scheme of UT system identifying main timber structural elements. Dimensions in meters. .................................................................................................................... 262 Figure (II) 6:4. Double timber beams applied in constructed buildings: at right Tamedia Building and at left Sky believes in better building. .............................................................................................. 265 Figure (II) 6:5. Connection between CLT floor panels and double glulam beams: at left before at right after the evaluation of the UT system made by Dias (2017). ................................................ 268 Figure (II) 6:6. Connection between double glulam beams and CLT walls by means of an external steel connector. ............................................................................................................................. 270 Figure (II) 6:7. Connection between double glulam beams and CLT walls by means of an internal steel connector. ............................................................................................................................. 271 Figure (II) 6:8. Connection between double glulam beams and CLT walls by means of an internal hardwood connector. ..................................................................................................................... 272 Figure (II) 6:9. Connection between reinforced concrete wall/foundation and CLT walls proposed before evaluation of the UT system. .............................................................................................. 274 Figure (II) 6:10. Connection between reinforced concrete wall/foundation and CLT walls after Dias (2017) evaluation of the UT system. ............................................................................................. 274 Figure (II) 6:11. Acoustic solution that hides all structural system. ............................................... 277 Figure (II) 6:12. Acoustic solution that exhibits part of structural system. ..................................... 277 Figure (II) 6:13. Comparison between cellular construction and the UT system, regarding building shape possibilities. ......................................................................................................................... 280 xxiii Figure (II) 6:14. Architectural possibilities offered by UT system. ................................................. 281 Figure (II) 6:15. Sequencial process for design, pre-fabrication and erection of a CLT building based on UT system. ................................................................................................................................ 283 Figure (II) 7:1. Façade solution based on rain screen concept, in which water from the rain run a direct path directly to the ground. ............................................................................................................. 296 Figure (II) 7:2. Façade solution based on rain screen concept, in which water from the rain must be drained from façade plane and from external balconies. .................................................................. 296 Figure (II) 7:3. Detailed solution in which the UT system proposes a continuous rain screen system. ...................................................................................................................................................... 298 Figure (II) 7:4. Façade solution based on a simple glazed curtain wall. ........................................ 299 Figure (II) 7:5. Façade solution based on a double glazed curtain wall. ........................................ 299 Figure (II) 7:6. Detailed solution in which UT system proposes a glazed curtain wall. .................... 300 Figure (II) 7:7. Detailed solution in which UT system proposes a continuous rain screen system with external balconies. .......................................................................................................................... 303 5 AI3 Dynamic modulus of elasticity, in MPa J  velocity of sound propagation, in m/s 1 Mass of test piece, in grams J Volume of test piece, in cm 3   , B2@B Maximum withdrawal resistance obtained by mechanical test, in N/mm 2 K 1 Maximum withdrawal load given by test machine, in A Screw diameter, in mm C  Length of penetration, in mm 1  Mass of specimen before drying, in g 1 L Mass of specimen after drying, in g   , M Density with 12% of moisture content for each specimen, in kg/m 3   , M Density with a different moisture content for each specimen, in kg/m 3   ,  , M Corrected withdrawal resistance for each test specimen, in N/mm 2   , B2@B , M Withdrawal resistance for each specimen resulted from test machine, in N/mm  2 Density of reference (mean value for  NO obtained with entire data of tests performed), in kg/m 3   , M Density of each specimen with a moisture content of 12%. Ƞ   Effect of number and width of gaps in withdrawal resistance of STSs P  Effect of each gap added in the withdrawal resistance Ƞ  Effect of simple MC changes on the withdrawal resistance of STSs P  Effect of each percentage unit of MC added/subtracted in the withdrawal resistance 0 Angle between screw axis and grain direction. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 31 INTRODUCTION The continuous increase of the urban density all around the world forced cities to grow bigger, making tall buildings an ordinary typology in developed cities. The urban population keeps growing, foreseeing that until 2050 the world population living in cities will reach 70% (Green & Eric Karsh 2012). This reality will reinforce the demand for large building solutions and it is necessary to be aware that urban density becomes an increasingly important component when addressing climate change. So, it is necessary to reformulate the current construction practice in order to answer the higher standard of living and higher energy efficiency required today. Unfortunately, tall buildings are generally linked to large negative impacts on environment, raising the need to look for new environment friendly solutions. Tall timber buildings are a concept that emerged connected with this need, stressing the wood sustainable profile as the key factor to reduce the negative environmental impact of the construction sector. In some countries where timber has a social character, such as Sweden, German and Japan, timber is positively appreciated as a building material (Stehn & Bergström 2002). However, this subject can face serious barriers in countries where wood culture does not exist. Regardless of these difficulties, tall timber buildings are an exciting and current topic, expecting that their qualitative advantages overcome remaining socio-cultural barriers. Sustainable profile linked to timber as a construction material can be a strong ally to recent European environmental policies, namely EU’s 20-20-20 plan and 2050 Energy Roadmap, as well as the 2030 Agenda for Sustainable Development. Indeed, timber is a natural material, renewable, recyclable and able to capture carbon dioxide (CO 2 ). This last feature is the most important once, similarly to forests, harvested timber can be used for carbon storage. However, in order to guarantee the wanted positive environmental effects, the carbon emissions during manufacturing of wood products should be reduced and these materials should have significant long lifespans assuring a useful carbon sequestration (Esbjörnsson et al., 2014). Using timber as a structural material in construction sector is one solution that fits on longlasting carbon storage, assuring that CO 2 stored will only be emitted into the atmosphere upon combustion or decay of timber. Wood stores more carbon than the equivalent CO 2 emitted by the harvesting, processing, transport and fabrication (Cambiaso & Pietrasanta, 2014). While each ton of solid wood panels sequester around 1.6 tons of CO 2 , the production of one ton of steel and cement releases 1.5 and 1.1 tons of carbon, respectively. (Stehn & Bergström, 2002). The economic sustainability of timber construction is closely related with savings provided by technical facilities, construction speed and long-term savings. Firstly, material production is industrialized and P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 32 controlled by a computerized numerical control (CNC) system which ensures the high quality of construction elements, reduces the amount of material waste and makes the execution of connections easier. Secondly, the simplicity of construction associated to timber (easy handling and prefabrication) allows a significant reduction of construction time, simplifies the building site yard, requires reduced teams and increases the on-site safety. Lastly, dependent on construction system chosen, significant longterm savings related with energy consumption, operation and maintenance can be more relevant than costs resulted directly from the erection of building (Lehmann, 2012). Despite the large number of advantages associated to the use of timber as a structural material, there are some important barriers to be overcome. During the last century, wood was qualified as an inflammable material with low durability, fact that resulted on stagnation of timber as a construction material and allowed steel and reinforced concrete to completely domain construction sector. In some cultural contexts, the negative connotation imposed to timber was so ingrained that the idea of building a large-scale multi-story timber building is not well accepted, even nowadays (Langenbach, 2008). However, taking a look into the past, it is possible to find interesting examples of tall timber buildings/structures, erected with archaic technologies that are still standing. Cross laminated timber (CLT) is an engineered timber product that has been largely associated to the concepts of tall timber buildings due to its excellent mechanical properties, relatively low mass and easy application. CLT constructions are often cited as a great sustainable solution due to their capacity for storing a large amount of carbon dioxide (Green & Eric Karsh 2012, Omland & Tonning 2009). Analyzing the progressive increase of tall timber constructions, it can be observed that CLT is a transversal material, being the leading material in the majority of the constructed buildings as well as in the proposals for new construction systems. There are examples of monolithic construction systems that are fully based on CLT as structural material. Other buildings combine CLT with concrete cores or with linear elements made from steel and other wood-based products. In this thesis, CLT will be presented as a structural material and its advantages and disadvantages will be listed. Furthermore, demonstration buildings in which CLT is applied will be described as well as their construction systems. The present research addresses multi-story timber buildings, but also discusses the innovative role that CLT can play in this type of construction. The initial motivation for this research was to give some contribute to the implementation of timber construction in urban context through the development of a new CLT based construction system thought to be applied in multi-story buildings, in which some important parameters would be analyzed: architectural versatility, either for the organization of interior Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 33 space, or for the shape or exterior appearance of the building; cost competitiveness; functional requirements such as structural behavior, fire protection and placement of building services. As a result, some case studies would be developed in order to demonstrate the versatility of developed system. However, as the research progressed, the author felt that proposing a system based on a material and structure rather different from traditional systems brings new challenges. In fact, it became clear why architects who are not familiar to the system do not risk to propose a solution based on a timber structural system. Here, it was decided to perform an experimental campaign that would allow the author to indepth characterize a new exciting: Cross Laminated Timber (CLT). Despite the advantages associated with timber construction, many architects and engineers do not feel confident suggesting timber in projects where other materials are more traditional. These professionals suffer a lack of knowledge and experience in timber construction. To empower architects, the knowledge must be expanded in material properties, production and construction as well as in building physics and economy of timber construction. This way architects would have a stronger voice in the decision making for sustainable solutions to urban building construction (Esbjörnsson et al., 2014) . After some in-depth research on the subject of CLT characterization, a gap in the knowledge regarding the hygroscopic behavior of CLT as well as the behavior of screwed connections subjected to moisture induced effects was detected. At that moment, the research was divided into two different parts. One would be dedicated to the development of an experimental campaign focused in the evaluation of hygroscopic behavior of CLT elements and in the evaluation of moisture induced effects on withdrawal capacity of self-tapping screws inserted in CLT elements. The other part would be dedicated to the development of a structural system using CLT as main structural material and designed to shape multistory timber buildings. Joining these two subjects in the same doctoral thesis may seems not evident, however, it must be said that without the laboratorial tests performed by the author, the structural system proposed would not be the same. That is because the system was developed taking special attention to the moisture induced effects. The position of structural elements, the building sequence, the connection between elements and façade composition were defined considering the behavior of CLT regarding changes on its moisture content. Wood hygroscopic behavior varies according to the wood species, but it can also vary between solid wood and timber engineering materials. Cross lamination of CLT can restrict moisture induced movements, reducing swelling and shrinkage movements when comparing with solid timber or Glulam. Besides the changes on geometry, moisture variations can also lead to changes on timber mechanical properties, P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 34 such as shear strength and modulus and, consequently, changes on the load-carrying capacity of timber elements. Moisture gradients are another important moisture effect which may affect the stress of wood. This effect results of the slow moisture diffusion in wood when humidity load is variable or different from initial equilibrium. As previously mentioned, when wood is exposed to variable humidity conditions it absorbs and releases moisture from / to the air. However wood needs long time periods to reach equilibrium for different levels of relative humidity and, depending on timber size, this can take several weeks or even months (Time, 1998). This means that timber structures are affected by climatic variations (fast climatic changes), which do not let wood reach equilibrium. As result, moisture gradients are induced in wood sections and hence internal stresses arise. Moisture gradients induce important differences in shrinkage and swelling of wood, which will develop so-called moisture induced-stresses (MIS), due to constrained volume variations. Regarding connections, in Europe, self-tapping screws (STS) are the solution most used, essentially because, the results are satisfactory and the system is extremely simple to apply, without the need of predrilled holes. STS are able to combine axial and lateral loads and also allow their withdrawal. STS are positively known as an easy and economical solution and recommended by manufacturers for most joint details. Therefore, the interest in obtaining further knowledge about STS’s performance has been growing. The advantages associated to this kind of fasteners allowed an easy implementation in the field of timber construction, but further research is still needed. Actually it is essential to develop an adequate design formula, considering factors related with CLT specificities, such as gaps in unglued cross boards and other sawn grooves (necessary in case of vacuum press production procedure), and other parameters not considered in building codes, such as moisture induced effects or service classes with moisture contents above the fiber saturation point. So, considering the relation between the two parts of the present thesis, the research question is: How to consider moisture induced effects when a CLT structural system for a multi-story building is being designed? Figure 1 presents the goals considered in the present thesis and also indicates in which way the different goals are related with each other. Different goals are presented here by numbers, even if it does not mean that goals were reached in such sequence. In fact, most of tasks were being developed simultaneously. Bibliographic research was a permanent action during the development of the work, and as the results of experimental campaigns were obtained while the Urban Timber (UT) system was being developed. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 35 Figure 1. Scheme relating main goals involved in the present thesis. The first goal (G1) is dedicated to bibliographic research on potentialities of multi-story CLT buildings. Analyzed the available publications, the role of timber in modern cities was presented: the main barriers to multi-story timber construction were identified, solutions were gathered and the sustainable profile of timber construction was discussed. Then, the role of CLT on multi-story timber buildings was defined and main advantages and disadvantages were pointed out. At last, most important demonstration buildings and innovative proposals were described. All these issues are presented in chapter 1 of Part II. The second goal (G2) emerged from the need to acquire knowledge about the behavior of CLT regarding moisture induced effects and it is presented in chapter 1 of Part I. After a research about all characteristics of CLT as a structural material, it was observed that the information about the behavior of CLT and its connections to moisture change was scarce. So, an in-depth research was performed in this field: hygroscopic behavior of timber was explained in order to introduce all phenomena that occur in timber when it is submitted to moisture change; existing information about the behavior of CLT in particular, regarding moisture changes was collected; the role of connections on timber construction was explained, as well as the significance of withdrawal capacity of self-tapping screws. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 36 The third goal (G3) is discussed in chapter 2 of part I and it is focused on the quantification of moisture induced strains in CLT elements, when it is submitted to Relative Humidity (RH) changes in the environment, considering different sealing conditions: radial, tangential and longitudinal direction. Three different measuring techniques were used, namely: Digital image correlation (DIC), LVDT’s and caliper ruler. Test specimens were submitted to RH cycles in which RH varies between 30% and 90% in order to simulate aging effect. The fourth goal (G4) is dedicated to the quantification of the effect of moisture changes on the withdrawal capacity of self-tapping screws inserted in the main face and side face of CLT elements. Performed tests considered several configurations regarding location and width of gaps. Test specimens were submitted to simple moisture changes as well to cyclic changes. Using the obtained results, the influence of gaps location and width was modelled using a predicting model develop by Uibel and Blaß (2007). Tests that considered simple moisture changes were performed at Institut für Holzbau und Holztechnologie - Graz Technical University (Austria) and tests in which STS was inserted in the side face of CLT panels were developed at Civil Engineering Department of University of Minho and at Instituto Politécnico de Bragança (IPB). The experimental campaigns related with the withdrawal capacity of STS’s are presented in chapters 3, 4 and 5 of part I. The fifth goal (G5) and sixth goal (G6) are treated in chapter 2 of part II. G5 is dedicated to the development of a new structural system for multi-story timber buildings, in which CLT is the main structural material. The proposed structural system was called Urban Timber (UT) system and ended up to be a hybrid structural system, once it was concluded that beams should be shaped by Glulam elements, instead of CLT. The structural design of the UT system was made in a partnership with (Dias, 2017), who developed the structural evaluation of the UT system, designing a 3D model in RFEM-5 software , a finite element software by Dlubal. However, the UT system was developed considering not just structural behavior but also architectural issues and adaptive design, production of structural elements, and construction sequence (G6). In the present thesis, the architectural potentialities are presented individually and no demonstration building was developed. Silva (2014) used the potentialities to present a building solution as the main goal of her master thesis. The author resorted to the UT system to propose a solution for a multi-story timber building in which architectural possibilities and adaptive design was explored. At last, goal 7 (G7) is dedicated to apply the knowledge obtained by the conclusion of part I of present thesis. All the decisions taken related with location of structural elements, design of connections and Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 37 development of façade configurations in which moisture induced effects were considered are explained in chapter 3 of Part II. This is where Part I and Part II intercept making this thesis make sense. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 38 Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 39 PART I : QUANTIFICATION OF MOISTURE INDUCED STRAINS IN CROSS LAMINATED TIMBER AND THEIR EFFECTS ON WITHDRAWAL CAPACITY OF SELF-TAPPING SCREWS P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 46 were only based in two points it should be included an end effect to obtain the maximum stress (T UVR = 1.30T U[V\ ). Figure (I) 1:2. Test specimen, before and after sawing into eleven slices (J. Jönsson, 2004). ] U[V\ = ∆ _ _ = ` O − ` N ` N (1:1) T U[V\ = ] U[V\ ∙  ( c ) (1:2) ] U[V\ Mean Strain ` N Distance between two measure points before cutting specimen into slices ` O Distance between two measure points after cutting specimen into slices T U[V\ Mean stress (MPa)  ( c ) Modulus of Elasticity as a function of moisture content Different groups of specimens were divided and subjected to different actions: seasoned in constant humidity, subjected to a single climate change, exposed to cyclic climate change, and exposed to natural outdoors under shelter. Sub-groups of each group were tested in different times during climatic exposure (see Table (I) 1:1). The results showed that even without RH changes it was found maximum tension and compressions stresses in the order of 0,2 MPa. For the specimens with moisture induced gradient the tensile stresses were two times higher (sometimes higher than characteristic tensile strength perpendicular to the grain), and compressive were three times higher. The cycle exposure did not present any cumulative effect. In a posterior study, Jönsson and Thelandersson (2003) realized an analysis of the effect of moisture gradients on tensile strength perpendicular to the grain in glulam , using the same experimental program. The study was based on the combination between initial moisture induced stresses and stresses from external loading during tension tests. Their main conclusions indicate that specimens subjected to a drying process present a marginal reduction of tension capacity, because moisture induces compression stresses, while specimens subjected to a swelling process present a significant reduction of tension Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 47 capacity since moisture gradients induce tension stresses. Similarly to previous investigation performed by J. Jönsson (2004), cyclic exposures did not present significant differences in comparison with single climate changes. Table (I) 1:1. Relative humidity changes used on experimental program performed by Jönsson (2004). GROUP RH [%] CLIMATE EXPOSURE TOTAL NUMBER OF SPECIMENS NUMBER OF SPECIMENS TESTED DAYS OF TESTING A 40% 40% 14 _ _ 80% 80% 14 _ _ B 40% 80% 14 2 1,3,5,6,11,24,38 80% 40% 14 2 1,3,5,6,11,24,38 C 60% Cyclic: 40% and 80% (seven days interval) 30 2 3,5,10,17,24,31,38,45,52,59,66 ,73,80,87,94,101 D 60% Natural climate 52 2 0,3,11,20,33,42,62,69,76,90,10 4,118,132,146,161,175,188,20 3,221,230,244,258,272,287,30 4,317 Angst and Malo (2012b) also developed an experimental work with glulam which provided important data to understand the effect of climate variations on timber cross section. Their main goal was to extend the basis for a numerical model. For that, based on Jönsson (2004), they quantified moisture induced strains (restrained and released and moisture induced stresses perpendicular to the grain direction for glulam specimens subjected to an single climate changes (Table (I) 1:2). Angst and Malo (2012a) developed another study in this field in order to understand the effect of moisture induced stresses on the behavior of self-tapping screws used as reinforcement elements in glulam structures. The climatic conditions and swelling/drying procedures used were the same described at (Angst & Malo, 2012a). After first seasoning, a self-tapping screw was inserted in the center of each glulam specimen. The RH levels were selected to represent typical conditions for sheltered, unheated structures in Nordic countries. The application of single instead cyclic climate changes was selected, because annual changes in RH are more significant than daily changes (Angst & Malo, 2012b). Combining experimental and numerical results they could conclude that screw reinforcement can lead to a reduction of between 30-70% of tensile stresses perpendicular to the grain located in the center of a glulam cross section. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 48 Table (I) 1:2. Test program developed by Angst and Malo (2012b). TEST SERIES (MEASUREMENTS) SEASONED IN RH EXPOSED TO RH TOTAL NUMBER OF SPECIMENS PARALLEL SPECIMENS DAYS OF WETTING/DRYING UNTIL TESTING MOISTURE INDUCED STRAINS AND MOISTURE CONTENT 50% 90% 20 5 5,12,21,38 90% 50% 20 5 5,12,21,38 MODULUS OF ELASTICITY 50% 90% 12 3 5,12,21,38 90% 50% 12 3 5,12,21,38 HYDRO EXPANSION COEFFICIENT 50% 90% 3 3 At increasing time intervals (hours to days) 90% 50% 3 3 At increasing time intervals (hours to days) Relatively to studies focused on effects of moisture induced stresses on CLT, the sources are scarce. Some work in this field was developed by Gereke (2009), who measured linear swelling in three direction of a CLT panel (Figure (I) 1:3), between relative humidifies of 35% and 85%. Considering the same RH range, Gereke (2009) also realized comparative measurements of internal stresses of CLT specimens with three layers, in which the material of middle layer varies. The measurements were made manually by means of a linear gage, which can lead with significant inaccuracy. Author concluded that the free swelling and shrinkage of adjacent layers differs by a factor of 10 (radial/longitudinal) to 20 (tangential/longitudinal) resulting in serious structural damages and shape distortions which may reduce the material serviceability. Often MIS exceeds the tensile strength of timber perpendicular to the grain, leading to cracks (either on the surface of timber or in the central part of timber sections), shape distortions and reduction of load bearing capacity (by splitting failure). The present thesis presents an experimental work performed with three layered CLT panels with the same objectives of the described work, but presenting some new parameters, such as: the digital image correlation (DIC), which will be adequately described ahead in the text and the conditioning of moisture flow by means of insulation of different layers. (a) (b) (c) Figure (I) 1:3. Procedure to determine the internal stress state of laminated wood panels: (a) removal of the edges, (b) cutting into strips, (c) release of strain by sawing in the glue lines (Gereke, 2009). Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 49 1.2 CLT connections 1.2.1 The role of connections on timber construction In a general way, connections are key points for all kinds of timber structure. This statement of timber structures stands for CLT, being the connections essential elements to guarantee the integrity of the structure and to provide strength, stiffness, stability and ductility. For example, the structural efficiency of the floor acting as a diaphragm and the walls resisting to lateral loads in a CLT structures are dependent on fastening connection details. The efficient design and fabrication of connections often determines the level of success of timber buildings. This is an issue that has to be carefully studied in the field of multi-storey heavy timber structures, either for buildings entirely made of timber or buildings composed by hybrid structures. To get an idea, according to Wells (2011), in the case of a connection between walls and floor subjected to high loads, generally pointed out as a weak link in a CLT structure, a connection made of screws or nail arrays is crucial to ensure safety. One important advantage of the CLT structures is the high accuracy guarantee by the computer numerical control (CNC) technology, which allows sophisticated connection systems. Moreover, the good dimensional stability recognized to CLT helps to ensure an accurate execution of connection. CLT systems accept different kinds of fasteners that are required for different functions. There are fasteners to connect different CLT elements in different contexts (roof/wall, wall/floor, and inter-storey connections – see Figure (I) 1:4) and there are fasteners suitable to connect CLT to other wood-based materials or to steel or concrete. Self-tapping screws are the solution typically recommended by manufacturers and are essentially used to connect panels to panels in floors and floor to walls. Other efficient solutions to connect panel elements are: wood screws rivets, bolts and dowels. When high loads are involved connections resort to split rings, shear plates and tooth plates. Some innovative glued-in rods connection systems have been emerged associated to high degree of prefabrication of CLT, examples are: Geka connectors and KNAPP © (FPInnovations2, 2011). P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 50 (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) Figure (I) 1:4. Examples of connection between CLT elements. (a) connection between floor panels with self-tapping screws, (b) connection between wall panels with self-tapping screws, (c) connection between wall panels with screws and steel plates, (d) connection between floor and walls elements with self-tapping screws in a platform construction system, (e) connection between floor and walls elements with screws and steel plates in a platform construction system, (f) connection used to assemble balconies with self-tapping screws, (g) connection used to assemble balconies with screws and steel plates, (h) connection between floor and walls elements with self-tapping screws in a balloon construction system, (i) connection between floor and walls elements with screws and steel plates in a balloon construction system, (j) connection between roof and wall elements with selftapping screws, (k) connection between roof and wall elements with screws and steel plates. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 51 In Europe self-tapping screws (STS) are the solution most required, essentially because the results are satisfactory and the system is extremely simple dismissing predrilled holes. They are able to combine axial and lateral loads and also allow its withdrawal. In a general way, the application of inclined screws in timber-to-timber connections, arranged under an angle of 45° between screw axis and member axis, have a higher load-carrying capacity compared to common shear connections due to the high withdrawal capacity of the self-tapping screws. Further, the additional use of steel plates with special holes, incorporating the geometry of the screw head, leads to a further increase in load-carrying capacity and stiffness. The ultimate load of these joints is mainly limited by the withdrawal capacity, the tensile capacity of the screw and the friction between the steel plate and the timber member (Krenn & Schickhofer, 2009). 1.2.2 Self-tapping screws and withdrawal resistance Despite the large variety of fasteners and types of connections compatible with CLT construction, nowadays STSs are widely used. They are positively known as an easy and economical solution and recommended by manufacturers for most joint details. Therefore, the interest in obtaining further knowledge about STSs performance has been growing. The advantages associated to this kind of fasteners allowed an easy implementation in the field of timber construction, but further research is still needed. Actually it is essential to develop an adequate design formula, considering factors related with CLT specificities, such as gaps in unglued cross boards and other sawn grooves (necessary in case of vacuum press procedure), and some other parameters not considered in building codes procedures, such as moisture induced effects or service classes with moisture contents above the fiber saturation point. Further, there is the question related to the angle between the screw and the grain direction. When compared with standard screws, STS present some important advantages, such as: (i) the special shape of the thread region allows a high load transmission into the surrounding wood; (ii) generally, they are hardened after rolling the thread, increasing the yield moment, the torsional strength and the steel tensile capacity; (iii) the stiffness of the connection increases while the danger of “slipping” decreases (Frese & Blaß, 2009). Furthermore, STS are characterized by a high load-carrying capacity when axially stressed, essentially due to the combination of two characteristics: (i) the long thread lengths and (ii) the hardened steel with tensile strengths up to 1200N/mm 2 (A Ringhofer et al., 2015). The characteristic withdrawal resistance of a fastened timber connection is an essential parameter to be considered, especially if the screws are inserted at an angle (Q) to the timber grain. In this case, the P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 52 load-carrying capacity of screws loaded in withdrawal becomes more important than the load-carrying capacity of screws loaded perpendicular to their axis. Eurocode 5 (EN 1995-1-1, 2004) specifies the methodology to determine characteristic withdrawal resistance (f ef,g ) for the composite model “timberscrew”, which is based on a relation between screw penetration depth (l ij ), screw nominal diameter (d), timber characteristic density (ρ g ) and the angle between the screw and grain direction (Q) (see equation (1:3). In recent years, several studies have been performed aiming to improve this standardized proposal, considering advances on screws technology and timber products as well as introducing new parameters to the equation. Recent publications are focused on the study of the slenderness of screws (Ellingsbø & Malo, 2012), the angle between screws and grain direction (Bejtka & Blaß, 2002) (Krenn & Schickhofer, 2009) (Grabner, 2013), the moisture content (Andreas Ringhofer et al., 2014) (Abukari et al., 2012) (A Ringhofer et al., 2015) and temperature (Pirnbacher et al., 2009). Considering the specific case of CLT, some important researches were developed looking for a withdrawal equation that considers CLT specificities. Uibel & Blaß (2007) performed an extensive test program to analyze withdrawal resistance of self-tapping screws inserted, either in plane side or in narrow side, in CLT plates. As a result, they suggest a withdrawal equation which combines the following parameters: nominal or outer diameter (d) of the screw, effective pointside penetration length (_ [l ), angle (Q) between screw axis and grain direction and CLT density (ρ) (see equation (1:4)). In the present study, this prediction equation is adjusted in order to include new variables related with the changes in MC and the existence of gaps. Muñoz et al. (2010) developed an experimental study in which the withdrawal resistance of a CLT wallto-floor connection using self-tapping screws, was tested. Test results were compared using various withdrawal equations concluding that most equations tend to over-estimate the withdrawal resistance, which leads to the need of revising the proposed equations, especially those proposed by design standards. The comparison used the equations proposed by Canadian and European standards (CSA O86-09 (Canadian Standard Association) (equation (1:5) and equation(1:6)), Eurocode 5 (equation (1:3)), the model proposed by Uibel & Blaß (2007) (equation (1:4)) and the equation proposed by the screws manufacturer Würth (equation (1:7)). Ringhofer et al. (2014) collected, from different sources, data related with the effect of changes in MC on withdrawal resistance of STS inserted either in solid timber, CLT and Glulam (GL). Authors used data Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 53 collected to develop a simple bilinear model approach for a MC range between 8% and 20% (equation (1:8)). The same bi-linear model is applied in the present paper adding new variables: the number and width of gaps. Aware about the differences between solid timber and laminated timber products, Ringhofer et al. (2015) developed a stochastic model, verified by laboratorial test results, in which they treat withdrawal resistance as dependent on the density and on the number of layers penetrated by the screw.  VR , m = 0 . 52 ∙  o p . q ∙ _ [l o p . N ∙  m p . r (1:3)  VR , m Withdrawal resistance of self-tapping screws in CLT, in N;  Thread diameter in mm; _ [l Effective penetration length of the threaded part (mm);  m Characteristic density, in kg/m 3 . s VR , t , uv[w = 0 . 44 ∙  p . r ∙ _ [l p . y ∙  p . zq 1 . 25 ∙ ! O ] + | O ] (1:4) s VR , t , uv[w Predicted values for withdrawal resistance of self-tapping screws in CLT, in N;  Nominal or outer diameter of the screw, in mm; _ [l Effective pointside penetration length, in mm;  For joints in the plane side of CLT: density of CLT (whole cross section), in kg/m³; For edge joints in CLT: density of the relevant layer(s), in kg/m³; ] Angle between screw axis and grain direction. CSA O86-09 – Withdrawal resistance for wood screws (clause 10.11.5.2) } ~ = 68 ∙  p . rO ∙ • N . zz ∙ ` u‚ ∙ 1 . 25 , in N (1:5)  Screw diameter, in mm; • Mean relative density of main member; ` u‚ Threaded length penetration in the main member, in mm. CSA O86-09 – Withdrawal resistance for lag screws (clause10.6.5 – table 10.6.5.1) } ~ =  ∙ ` u‚ ∙ 1 . 25 , in N (1:6)  Screw diameter, in mm;  Tabulated value for the withdrawal resistance of lag screws, in kN/mm; ` u‚ Threaded length penetration in the main member, in mm. Würth – Stress in the screw shaft direction (clause 3.3.2) s VR , m = 10 ∙ _ [l ∙  , in N (1:7)  Screw diameter, in mm; _ [l Point-side penetration length of the threaded part minus one screw diameter, in mm. Ƞ † = ‡ 1 . 00 1 . 00 − ˆ U‰ ∙ ( c Š % Œ − 12 ) !" • 8% ≤ • ≤ 12% 12% ≤ • ≤ 20% (1:8) For solid timber, ˆ U‰ is 0.031 or 0.036, when Q is 90° and 0°, respectively. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 54 1.3 Digital image correlation - DIC 1.3.1 Fundamentals During last decade, contact free measurement techniques, such as two-dimensional digital image correlation (2D DIC), are raising fans in the field of experimental solid mechanics. Its success is essentially related with few attractive advantages: first, the simplicity of test setup that requires just a fixed camera, a white light or natural light source and a computer program to process the images; second, test specimens just need to receive a black and white surface painting with a random gray intensity distribution, which is typically obtained by aerosol spray or airbrush painting; third, 2D DIC method is compatible with most of digital image acquisition devices; and fourth, this technique provides full-field measurements for displacements and strains. However, such a simple method also entails some disadvantages: first, the 2D DIC method are totally dependent on the quality of imaging system; and second, the accuracy of strain measurements is lower than the interferometric techniques (Pan et al., 2009). The great argument to fight this fact is that DIC has the ability to perform measurements over an entire region of interest (ROI). It provides full-field displacements and strains by comparing a set, with a minimum of two images, acquired in different test stages. The first step for images computation is defining ROI at initial (reference) image, which is divided into evenly spaced virtual grids (facets). Motion registered between images acquired before and after deformation is measured by tracking the same points (or pixels) between images. However, to compute properly the displacements, each point of interest is centered at a square reference facet, which will be used to track the location of same point. The unique pattern of each facet will ensure that each point is accurately located once its neighborhood is also verified. The facet should be large enough to distinguish itself from the other facets and allow enough strain precision, however if it is too big, the spatial resolution of the data field is reduced (Crammond et al., 2013). According to (Pan et al., 2008b) and (Lecompte et al., 2006), there is a trade-off between using large and small facet sizes: facet size must be properly selected according to random intensity distributions of speckled patterns. To identify reference and deformed facets as being the same a displacement mapping function has to be chosen. The most commonly used are first and second-order shape functions once both consider the expected shape change of reference facet instead of simple rigid body translations. Then, a correlation criterion must be chosen and solved with regard to the desired deformation vector (p) using a proper algorithm to optimize correlation coefficient (e.g. NewtonRaphson or Levenberg-Marquardt methods). Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 55 According to literature (Pan et al., 2009) (Pan & Li, 2011), the zero normalized sum of squared differences (ZNSSD) is the correlation criterion that offers the most robust noise-proof performance and is insensitive to the offset and linear scale in illumination lighting (Equation ((1:9)). In summary, the accuracy of displacement measurements performed by DIC can be influenced by several variables: sub-pixel optimization algorithm, subset shape function, subset interpolation scheme, image noise as well as camera lens distortion (Pan et al., 2008a). / •‘’’“ ( * ) = ” Š  (  • ,  • ) −  U – ∑ Š  ˜  • ,  • ™ −  U Œ O š š − + ˜  › • ,  › • ™ − + U – ∑ Š + ˜  › • ,  › • ™ − + U Œ O š (1:9) where, Ω is the facet domain,  (  • ,  • ) is the pixel gray level at location (  • ,  • ) in the reference image, + (  ′ • ,  ′ • ) is the pixel gray level at location (  ′ • ,  ′ • ) in the deformed image, and  U is the mean gray level value over the subset in the reference image, and + U is the mean gray level value over the facet in the deformed image. In the case of timber research, this technique has been used to perform measurements on material exposed to humidity changes in order to quantify shrinkage or swelling coefficients (Angst & Malo, 2012b) (Girma Kifetew et al., 1997) (Grima Kifetew, 1996) and moisture induced stresses (Angst & Malo, 2012a) (Gereke, 2009) (J. Jönsson, 2004) (Johan Jönsson & Svensson, 2004). Other studies focus on the behavior of timber connections or on the development of timber cracking for elements submitted to mechanical actions. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 62 Figure (I) 2:8. Test campaign for group A and group B. DAY 56 90%RH DAY 161 30%RH 16 TESTS 16 TESTS 16 TESTS 16 TESTS C1_B C2_B C3_B C4_B C1_A C2_A C3_A 14 DAY 182 90%RH DAY 213 65%RH C4_A SAMPLING 10 16 TESTS DAY 35 30%RH 10 10 10 10 10 12 2 2 222 2 2 222 222 22 2 2 222 222 22 2 2 222 222 22 2 2 222 GROUP AGROUP B LVDTs 2 TESTS 2 DAY 0 65%RH 4 TESTS 2 2 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 63 2.3 Test setup and test procedure 2.3.1 DIC technique The test setup used for measurements performed with DIC technique was one of the main concerns of this experimental campaign. As required by the technique, it was mandatory to ensure that the object is planar, parallel to the camera and keeps a constant distance from the camera for all test moments. However, as the camera and specimen had to be placed at test setup every time image capture was required, it was not simple to guarantee that the relation between both was always the same. Figure (I) 2:9 depicts the steel structure that was designed for this propose. It is composed by a steel base and two perforated L-section steel columns in which a horizontal Rexroth aluminum profile can be fixed at the most favorable height. It is in the Rexroth profile that camera and lights are fixed, every test day, in the most similar position possible. Despite the care to keep the camera always at the same position, the zoom position was recorded in order to reconstruct the test setup with the same reference. Once these two positions (camera and zoom) were fixed, replacing specimens was guaranteed by the axis drawn in the specimen surface, which should overlap the camera axis (see section 2.2). Furthermore, the conversion factor between pixels and millimeters was done for each tested specimen individually using a graph paper (5 x 50 mm) glued at surface of specimens. Image acquisition was performed by a 8-bit Charged-Coupled Device (CCD) Baumer Optronic FWX20 digital camera equipped with a Nikon AF Micro-Nikkor 200 mm f/4D lens (Table (I) 2:1) using the GOM ARAMIS ® DIC-2D v6.02 software. Camera and specimen were positioned in the test setup with a distance between them that varies dependent on geometry of specimens, leading to different conversion factors (Table (I) 2:1). Lens aperture was fixed to f /11 improving the depth of field and shutter time was set to 5 ms. Two lighting sources (LEDMHL10) were used to guarantee the adequate lighting of the specimen surface. Once acquisition of each reference image was done, the region of interest (ROI) was defined and subdivided by means of a virtual grid (facets). To define the facet size (3375 x 3375 pixels), three main parameters were considered, namely: the size of ROI, the optical system and the quality of speckled pattern (average of speckle size). The facet step was set to 2197 x 2197 pixels. Then, the in-plan displacements were computed and the full-field strain distribution determined. In terms of evaluation of P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 64 displacements, typical resolution is in the range of 0,01-0,02 pixel. Evaluation of strains was done considering a typical resolution that varies between 0,01-0,03%. DIC was used to compare the behavior of different configurations designed to understand the effect of moisture flow on CLT hygroscopic behavior. This choice was essentially based on two points: first, DIC offers full field measurements and second, the technique requires a simple testing apparatus making possible to test a large quantity of specimens. Figure (I) 2:9. Test layout: at left side pictures of steel structure during the test and at right side drawing of the same steel structure (dimensions in mm). Top view Front view Rexroth aluminum profile Lights Camara Perforated L-section columns CLT specimen Metal base CLT specimen Lights Rexroth aluminum profile Perforated L-section columns Metal base d Distance between specimen and lens 1500 500 50 400 50 50 500 79,5 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 65 Table (I) 2:1. Components of the optical system and measurement parameters. CCD camera Model Baumer Optronic FWX20 (8 bits, 1624 x 1236 pixels, 4.4 µm/pixel) Shutter time 5 ms Acquisition frequency 1 Hz Lens Model Nikon AF Micro-Nikkor 200 mm f/4D Aperture f/11 Lighting LEDMHL10 (color temperature: 6000 K) Working distance Group A F1 966 mm F2 and F3 884 mm Group B 1024 mm Conversion factor Group A F1 0,127 mm/pixel F2 and F3 0,096 mm/pixel Group B 0,097 mm/pixel Project parameter – facet Facet size 3375 x 3375 pixel Step size 2197 x 2197 pixel Project parameter – Strain Computation size 5 x 5 facets Validity code 55% Strain compotation method Total Image recording Acquisition frequency 1 Hz As mentioned before, all test specimens were pictured before being submitted to the RH cycles (day 0) and then pictured again for the remaining test days (day 35, day 56, day 161, day 182 and day 324). Similarly to some published work (Angst & Malo (2012), Gereke (2009) and Jönsson (2004)) two different DIC maps were done at each test day: first DIC map was done considering the effect of humidity loads on the unchanged specimens, measuring restrained strains (] v[t‚ ), while second DIC map were done P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 66 considering the effect of humidity loads on the sliced specimens, measuring released strains (] v[• ) caused by cross lamination either through the panel surface (Figure (I) 2:10) or between CLT layers (Figure (I) 2:11). Restrained and released strains were calculated as given in following equations: ] v[t‚ = ` N − ` p ` p (2:1) ` p Initial length taken at Day 0, in mm; ` N Length taken at remaining test days to the unchanged specimen, in mm. ] v[• = ` O − ` N ` N (2:2) ` O Length taken at remaining test days after cutting specimens into slices (group A) or after separation of CLT layers (Group B), in mm. Regarding group A, the first DIC map measured compressive/tensile restrained strains developed over the specimen surfaces at all three directions (X – longitudinal in outer layers (OL) and tangential in inner layers (IL), Y - tangential in outer layers and longitudinal in inner layers and Z - radial) of CLT panels. Immediately afterwards, specimens were sliced into 7 pieces (of which just 5 were inside the defined ROI) and pictured again just in their main face (F1) in order to measure compressive/tensile released strains (Figure (I) 2:10). This procedure allowed to evaluate the role of crosswise lamination over the cross section of CLT panels. The specimens were sliced through Y direction, in which higher deformations were expected. As depicted in Figure (I) 2:11, two different DIC maps were also performed for specimens of group B. The first DIC map was made to analyze compressive/tensile restrained strains distribution on lateral face of CLT slices (Y and Z directions), while the second DIC map analyses the compressive/tensile released strains distribution on separated layers of CLT slices. Here, the effect of crosswise lamination on hygroscopic behavior of different CLT layers was evaluated. Values obtained for ] v[• were used to quantify stresses through the surface CLT panels (T žŸ ) as well as to quantify stresses for outer and inner layers (T Ÿ and T ¡Ÿ ). Stresses were calculated as follows according to equation (2:3). Once image acquisition was finished, all timber pieces were measured and weighted, before and after oven dried, in order to obtain density and moisture content levels for all tested specimens. This way, it was possible to evaluate density distribution and the moisture content variation as the RH varies. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 67 Figure (I) 2:10. Measurements made on different test days for specimens from group A (dimensions in mm). Figure (I) 2:11. Measurements made on different test days for specimens from group B (dimensions in mm). 24 134 18 18 93 DAY 0 TEST DAY 18 93 134 13418 18 18 GROUP A 18 93 134 13418 18 18 1 CORRELATION ST 2 CORRELATION ND 24242424 24 29 29 29 12418 18 24 93 12418 18 DAY 0 TEST DAY 24 93 12418 18 GROUP B 1 CORRELATION ST 2 CORRELATION ND FFFF DIC MAP DIC MAP DIC MAP DIC MAP P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 68 T U[V\ = ] v[• , U[V\ ∙  ( † ) (2:3) T U[V\ Mean value of stress, in MPa; ] v[• , U[V\ Mean value of released strain;  ( † ) Modulus of elasticity as a function of moisture content present in the specimen at measuring time, in MPa. 2.3.2 LVDTs acquisition To complement the measurements performed by means of DIC technique, during the same RH cycles, another technique was used to quantify moisture induced movements. In order to obtain a continuous measurement of timber shrinkage/swelling, two CLT specimens from group A and C1 were instrumented with LVDTs in order to register timber movements on all studied directions (X, Y and Z). Taking the work performed by Gereke (2009) as reference, a stainless steel structure to fix LVDTs and CLT specimens was developed (Figure (I) 2:12) in such a way that only movements on measurement directions were allowed. Each CLT specimen was perforated with four hex head lag screws until the depth of first layer. This way, it was expected to obtain measurements close to those obtained with surface measurements made by DIC technique. These screws had a M3 threaded hole in their heads in which LVDTs and fixing elements were curled up. Unguided D5W submersible LVDT displacement transducers from RDP group were used. D5/100WRA LVDTs with a range of +/- 2,5 mm and a linearity error of <ą 0,5/ą 0,25 (% F.S.) were used to perform measurements in X and Y directions, while D5/200WRA LVDTs with a range of ą 5 mm and a linearity error of < ą 0,5/ą 0,25/ą 0,1 (% F.S.) were used to perform measurements in Z direction. Here, F.S if full scale or the full measurement range. Timber movements were registered by a data acquisition hardware from National Instruments while instruments were controlled by LabVIEW. Measurements were registered every hour during the period of RH cycles (182 days). During the final period of stabilization (until day 324) specimens were moved to another climatic chamber preventing the continuation of measurements with LVDTs acquisition system. Obtained data was used to calculate compressive/tensile strains, taking the initial dimension of the specimen as reference. This initial dimension was considered as the dimension at day 0, in order to evaluate compressive strains during all RH cycles. However, the initial dimension was also considered as the dimension at the beginning of each different humidity period in order to compared different Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 69 drying/wetting periods. Compressive/tensile strains were calculated as indicated in previous section by equation (2:1). Figure (I) 2:12. Test setup for measuring shrinkage and swelling by means of LVDTs. At left side drawing of test setup (dimensions in mm), and at right side photos of test setup inside the climatic chamber. 2.3.3 Caliper ruler Manual measurements were performed by means of a digital caliper ruler and used to perform a third analysis for specimens from group A. This technique was used just to take measurements that could not be taken by the other techniques used (DIC or LVDTs). Once specimens of group A were sliced, the caliper ruler was used to measure the total length of different layers on each slice. Then, the same measurements were done after the separation of different layers (Figure (I) 2:13). Registered values were used to calculate released strains for inner and outer layers individually and to evaluate the differences between outer and inner slices. LVDT_173715 LVDT_173714LVDT_173710 LVDT_173711LVDT_173712 LVDT_173713 LVDT_173712 LVDT_173710 LVDT_173711 Top view Front view 250 170 194 20 93 500 170 50 1825 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 70 Figure (I) 2:13. Manual measurements performed by caliper ruler on specimens from group A (dimensions in mm). 2.3.4 Controlling the environment of climatic chamber and timber moisture content during RH cycles During all RH cycles the environment in climatic chamber and moisture content of timber elements were monitored. Figure (I) 2:14 depicts relative humidity and temperature registered during RH cycles and shows that for periods with RH=90% defined environment was easy/fast to reach. This means that during wetting periods the RH remains constant. Differently, drying periods present difficulties to reach the defined environment (RH=30%), which keeps constant for less than half of the entire drying period. This is related with the slow release of moisture from timber to the environment. A set of ten samples with the same dimensions of the test specimens were placed in the climatic chamber during the entire conditioning period in order to control moisture content changes on CLT elements. The results obtained are depicted in Figure (I) 2:15 and Table (I) 2:2, showing that the percentage of moisture content responds to the drying and wetting cycles, however it does not reach the equilibrium for any of the cycles. For drying cycles, moisture content present values between 9,0% and 10,5% while for wetting cycles, moisture content is between 20,1% and 22,3%. TEST DAY GROUP A 292929 160 24 24 93 160 24242424 24242424 3 CORRELATION RD DIC MAP Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 71 Figure (I) 2:14. Relative humidity and temperature registered by Fitoclima 1000EC45 and Fitoclima 28000 during 324 days. Figure (I) 2:15. Moisture content registered by control specimens during 324 days. Table (I) 2:2. Descriptive statistics of results obtained by control specimens for moisture content during the RH cycles. DAY 0 35 56 77 98 119 140 161 182 324 MC [%] MAX . 14,8 9,5 22,3 9,8 21,5 10,1 21,5 10,5 21,2 13,7 MEAN 14,5 9,3 21,7 9,7 21,0 9,9 20,9 9,8 20,6 13,5 MIN , 13,9 9,0 20,8 9,5 20,2 9,6 20,2 9,4 20,1 13,2 COV 0,018 0,015 0,019 0,011 0,018 0,017 0,017 0,030 0,016 0,015 2.4 Test Results As mentioned before, displacements, strain distribution, moisture content and density were measured at six different moments: before initiating RH cycles (day 0); at the end of first and last drying and wetting cycles (days 35, 56, 161 and 182); and at the end of final stabilization period (day 324). In a preliminary observation of results, an important phenomenon was noticed regarding different moisture flow conditions. Analyzing the variation of moisture content levels for different test moments, it is possible to understand that, for configurations in which longitudinal direction is sealed (C2 and C4), the range of MC variation is shorter. This behavior is a consequence of slower water absorption and desorption. Table (I) 2:3, Figure (I) 2:16 and Figure (I) 2:17 present mean values and CoV of moisture content registered for outer and inner layers for all four test configurations and for all test days. Slower water absorption/desorption was registered for middle layers of C2 and C4, in which MC increases/decreases in a range of 3,4 – 4,5% and 2,8 – 3,1%, respectively. In contrast, C1 and C3 present higher moisture content variations, presenting increases/decreases in a range of 6,3 – 8,9% and 10 20 30 40 50 60 70 80 90 Relative Humidity [%] Temeprature [C ] o Temperature Relative Humidity Fitoclima 1000EC45 Fitoclima 28000 0 35 161 182 324 56 75 96 114 136 Time [days] 0 35 75 161 182 324 96 114 13656 7,5 10,0 12,5 15,0 17,5 20,0 22,5 MC [%] Fitoclima 1000EC45 Fitoclima 28000 Time [days] P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 78 - Results obtained for tangential direction (outer layers) suggest that there is a tendency for restrained strains to reduce as the number of cycles increase. So, restriction of timber movements caused by crosswise lamination can be more effective after successive RH cycles, probably due to some exhaustion of timber fibers; - The long duration (21 days) of each drying/wetting period can also be a cause for the observed behavior of CLT. Large drying/wetting periods allow a uniform distribution of moisture content through the CLT cross section and consequently less moisture gradients are generated; - Successive RH cycles can cause warping/twisting on CLT, which can result on unrealistic strains distribution, once the DIC method takes the measurements just on the surface of CLT panels. Regarding the observed behavior for longitudinal direction, the explanation should also be related with the same causes: crosswise lamination; possible fatigue of timber fibers caused by successive RH cycles; and with warping/twisting of CLT panels. It is possible that, due to a combination of these three factors, during the last drying and wetting periods, as the middle layers shrinks or swells on tangential direction, outer layers present the opposite tendency on longitudinal direction. Graphs depicted in Table (I) 2:7 show the full field restrained strain distribution in F1 of some test specimens from group A regarding Y direction (tangential) for C1 and C2 (remaining graphs are depicted in Annex 1:1). Observing color fill graphs, it is possible to identify easily compressive (cold colors) and tensile (warm colors) strains associated with drying and wetting cycles, respectively. It is also possible to observe through Y direction, especially for configurations that were not sealed at the main face of panel (C1 and C4), that there are differences between strains measured for early and latewood. Table (I) 2:8 depicts some color fill graphs obtained for restrained strains measured through X direction of main face of CLT specimens with C1 and C2 (remaining graphs are depicted in Annex 1:2). Surprisingly, for configurations in which main face is not sealed (C1 and C4), strains distribution is not symmetric, presenting in some cases opposite tendencies between right and left borders. The only explanation found for this tendency is the influence of tangential moisture movements of inner layers which restricts movements of outer layers on longitudinal direction. For configurations C2 and C3 this opposite tendency is not so clear, probably due to some influence of elastic membrane on timber movements. As described in 2.3.1 a second DIC mapping was performed. Specimens were sliced in Y direction and pictured again in order to take released strains distribution (] v[• ). Table (I) 2:9 presents the mean released strain values taken from all five slices for all four configurations and for all test days, while Figure Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 79 (I) 2:21 depicts the graphs of same mean values. Mean values were taken from the central longitudinal section of each slice in order to avoid any sawing effects, considering a central area of around 4 x 74 facets. Results show that for test days 0, 35, 161and 324, released strains are close to zero and tensile strains are predominant independent of the test day/moisture content. So, contrary to what was observed for restrained strains, a clear relation between drying/wetting periods and compressive/tensile strains was not observed. The effect of high moisture level at production of CLT panels seems not to induce significant released strains, once ] v[•_U[V\_p_«N_® = 0,000. Obtained released strains present significantly higher values for first wetting period (day 56), in which mean tensile strains are around 0,019 for all four tested configurations. This phenomenon shows that the effect of crosswise lamination is more evident during the first wetting period: despite the tensile restrained strains observed during the first DIC mapping, after sliced, timber still tends to swell substantially. Despite being low, compressive strains were registered just at the end of second wetting period (day 182), which presents the opposite tendency of that observed for day 56. So, similarly to what was observed for restrained strains, there is a clear effect of successive RH cycles on hygroscopic behavior of CLT panels. This effect is easily observed in color fill graphs depicted in Table (I) 2:10. C1 and C2 present here the typical behavior observed for all four configurations but the remaining graphs are depicted in Annex 1:3. All four configurations present similar values meaning that there are no causing effects related with different moisture flow conditions. Differently, comparing released strains obtained in day 0 and 324, it can be concluded that successive RH cycles cause a slightly increase of tensile strains. This tendency proves the existence of a restriction of timber movements caused by crosswise lamination once compressive restrained strains observed at first DIC map became tensile released strains at second DIC map. 2.4.1.2 Group A_F2 and F3 (CLT panels with dimensions of 170x170x93 mm) DIC map performed to specimens from group A included the three different faces of CLT panels. F1 was analyzed in the previous section and herein F2 and F3 (see Figure (I) 2:10 for indication of faces) will be also analyzed. F2 is one of the lateral faces of CLT panel in which outer layers are subjected to movements in tangential direction and inner layer in longitudinal direction. F3 present exactly the opposite arrangement. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 80 Table (I) 2:7. Restrained strains measured by DIC technique for configurations C1 and C2, on F1 of specimens by group A in Y direction, for all five test days, before cut (BC) the specimens. Y DIRECTION ( 0 2@B ) TEST DAY RH [%] C1 C2 SCALE DAY 35 [ RH =30%] A1_C1_F1_BC_35 A2_C2_F1_BC_35 DAY 56 [ RH =90%] A1_C1_F1_BC_56 A2_C2_F1_BC_56 DAY 161 [ RH =30%] A1_C1_F1_BC_161 A2_C2_F1_BC_161 DAY 182 [ RH =90%] A1_C1_F1_BC_182 A2_C2_F1_BC_182 DAY 324 [ RH =65%] A2_C1_F1_BC_324 A1_C2_F1_BC_324 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 81 Table (I) 2:8. Restrained strains measured by DIC technique for configurations C1 and C2, on F1 of specimens by group A in X direction, for all five test day, before cut (BC) the specimens. X DIRECTION ( 0 2@B ) TEST DAY RH [%] C1 C2 SCALE DAY 35 [ RH =30%] A1_C1_F1_BC_35 A2_C2_F1_BC_35 DAY 56 [ RH =90%] A1_C1_F1_BC_56 A2_C2_F1_BC_56 DAY 161 [ RH =30%] A1_C1_F1_BC_161 A2_C2_F1_BC_161 DAY 182 [ RH =90%] A1_C1_F1_BC_182 A2_C2_F1_BC_182 DAY 324 [ RH =65%] A2_C1_F1_BC_324 A1_C2_F1_BC_324 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 82 Table (I) 2:9. Descriptive statistics of full-field released strains measured on F1 of specimens from group A during second DIC mapping, after cut (AC) the specimens. RELEASED STRAINS ( 0 2C ) MAXIMUM MEAN MINIMUM COV DAY 0 35 56 161 182 324 C1 0,006 0,003 0,024 0,010 0,003 0,003 0,000 0,000 0,020 0,000 0,000 0,001 -0,004 -0,002 0,016 -0,004 -0,004 -0,001 1,82 -0,90 0,04 -4,23 0,59 0,26 C2 0,027 0,022 0,004 0,003 0,003 0,001 0,019 0,000 0,000 0,002 -0,028 0,017 -0,003 -0,006 -0,003 1,28 0,03 0,16 -0,08 0,25 C3 0,008 0,025 0,040 0,004 0,020 0,000 0,019 0,001 0,000 0,002 -0,005 0,014 -0,005 -0,003 0,000 -6,28 0,04 1,22 -0,91 0,54 C4 0,003 0,024 0,003 0,002 0,003 0,000 0,019 0,000 - 0,001 0,002 -0,002 0,015 -0,002 -0,004 0,000 -3,88 0,06 0,87 -0,22 0,26 Figure (I) 2:21. Mean values of released strains measured on central longitudinal section of all slices (S1-S5) of specimens from group A for all four test configurations. Regarding F2, Table (I) 2:11 and Figure (I) 2:22 summarize the behavior of all four configurations for inner and outer layers separately. Inner layers present predominant compressive strains during the first RH cycle while during the last RH cycle they present tensile strains for drying periods and compressive strains for wetting periods. It is important to mention here that for the last RH cycle, the results obtained for F1 presented the same tendency in X direction, which also quantify the movements in longitudinal direction. Surprisingly, during the last RH cycle, the outer layers present the opposite tendency of that observed for F1 in Y direction (tangential). This may be related with restrictions of timber movements caused by crosswise lamination. In fact, observing full field strains distribution presented in Table (I) 2:12 it is possible to notice that outer layers suffer tensile and compressive strains through Y direction which are also dependent on growth ring orientation. That is why shear strains present the diagonal symmetry depicted by color fill graphs. On contrast, inner layers present a more homogeneous strains distribution. S1 S2 S3 S4 S5 -0,018 -0,012 -0,006 0,000 0,006 0,012 0,018 0,024 DAY 35 56 161 182 324  strain A_C1_F1_AC_Y A_C2_F1_AC_Y A_C3_F1_AC_Y A_C4_F1_AC_Y Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 83 Table (I) 2:12 presents C1 as example however, remaining graphs obtained for F2 in Y direction are depicted in Annex 1:4. Table (I) 2:10. Typical behavior of released strain distribution observed after CLT specimens being cut (AC) into slices for all six test moments. Y DIRECTION TEST DAY RH [%] C1 C3 S CALE DAY 0 [ RH =65%] A2_C1_F1_AC_0 DAY 35 [ RH =30%] A1_C1_F1_AC_35 A1_C3_F1_AC_35 DAY 56 [ RH =90%] A1_C1_F1_AC_56 A1_C3_F1_AC_56 DAY 161 [ RH =30%] A1_C1_F1_AC_161 A1_C3_F1_AC_161 DAY 182 [ RH =90%] A1_C1_F1_AC_182 A1_C3_F1_AC_182 DAY 324 [ RH =65%] A1_C1_F1_AC_324 A2_C3_F1_AC_324 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 84 Table (I) 2:11. Descriptive statistics of full-field restrained strains measured on specimens from group A_F2 in Y direction, for inner and outer layers, during first DIC mapping, after cut (AC) the specimens. MEAN RESTRAINED STRAINS (ɛ rest_mean ) MAXIMUM MEAN MINIMUM COV F2_ Y DIRECTION INNER LAYERS ( IL ) OUTER LAYERS ( OL ) DAY DAY 35 56 161 182 324 35 56 161 182 324 C1 0,000 0,001 0,017 0,001 0,057 0,011 0,022 0,065 0,050 0,009 -0,002 -0,001 0,009 -0,004 0,001 -0,004 0,003 0,002 -0,003 -0,002 -0,004 -0,004 0,006 -0,005 -0,011 -0,015 -0,039 -0,015 -0,023 -0,009 -0,21 -0,74 0,04 -0,11 1,07 -0,85 3,80 6,95 -0,79 1,03 C2 0,003 0,005 0,011 0,007 0,015 0,018 0,030 0,026 0,029 0,012 -0,002 -0,002 0,009 -0,004 0,000 -0,003 0,000 0,009 -0,002 0,000 -0,010 -0,008 0,006 -0,025 -0,019 -0,014 -0,013 -0,003 -0,020 -0,010 -0,28 -0,44 0,04 -0,25 10,66 -1,12 -3,81 0,69 -4,54 0,87 C3 0,012 0,017 0,026 0,077 0,032 0,016 0,012 0,098 0,010 0,095 -0,002 -0,003 0,009 -0,004 0,001 -0,005 0,001 0,004 -0,002 -0,002 -0,017 -0,033 -0,038 -0,104 -0,012 -0,016 -0,010 -0,018 -0,017 -0,017 -0,59 -1,58 0,45 -1,71 4,44 -0,90 1,23 2,95 -2,75 3,68 C4 0,007 0,006 0,025 0,003 0,013 0,005 0,013 0,026 0,014 0,021 -0,002 -0,003 0,009 -0,004 0,001 -0,004 0,001 0,008 -0,001 0,000 -0,010 -0,015 -0,003 -0,013 -0,009 -0,014 -0,016 0,000 -0,012 -0,010 -0,91 -0,46 0,21 -0,21 2,42 -0,70 1,85 0,38 8,06 -14,82 Figure (I) 2:22. Mean values of full-field restrained strains distribution measured on specimens from group A_F2 in Y direction, for inner and outer layers, during first DIC mapping. -0,018 -0,012 -0,006 0,000 0,006 0,012 0,018 0,024 DAY 35 56 161 182 324  strain A_C1_F2_IL_BC_Y A_C2_F2_IL_BC_Y A_C3_F2_IL_BC_Y A_C4_F2_IL_BC_Y -0,018 -0,012 -0,006 0,000 0,006 0,012 0,018 0,024 DAY 35 56 161 182 324  strain A_C1_F2_OL_AC_Y A_C2_F2_OL_AC_Y A_C3_F2_OL_AC_Y A_C4_F2_OL_AC_Y Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 85 Table (I) 2:12. Typical behavior (C1) of full field restrained strain distribution in Y and Z directions and in shear observed before CLT specimens from group A_F2 for all test moments. RESTRAINED STRAINS FOR A _ F 2 _ C 1 ( 0 2@B ) TEST DAY RH [%] Y Z SHEAR PICTURES SCALE DAY 35 [ RH =30%] DAY 56 [ RH =90%] DAY 161 [ RH =30%] DAY 182 [ RH =90%] DAY 324 [ RH =65%] Note: Red lines at pictures represent the end of ROI considered by DIC measurements. -4,8E-02 -4,2E-02 -3,6E-02 -3,0E-02 -2,4E-02 -1,8E-02 -1,2E-02 -6,0E-03 0,0E+00 6,0E-03 1,2E-02 1,8E-02 2,4E-02 3,0E-02 3,6E-02 4,2E-02 4,8E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 86 Regarding inner layers, different moisture flow conditions do not affect the results and no differences between test configurations were observed. On contrary, outer layers present some differences at last drying period (day 161). Last drying period presents the highest tensile strains observed either for inner layers or outer layers. However, while inner layers present the same mean value for all four configurations ] v[t‚_U[V\_N¯N_«N /«-_¡Ÿ_® = 0,009, outer layers present different ranges: for C1 and C3 0,002 ≤ ] v[t‚_U[V\_N¯N_«N/«¬_ Ÿ_® ≤ 0,004 and for C2 and C4 0,008 ≤ ] v[t‚_U[V\_N¯N_«O/«-_ Ÿ_® ≤ 0,009. This difference should be directly related with sealing of longitudinal direction which delays the desorption process during drying period resulting in a lower ratio between tangential and longitudinal directions (1,0 ≤ ±²³´µ_¶·¶_¸¹/¸º_»¼_½ ±²³´µ_¶·¶_¸¹/¸º_¾¼_½ ≤ 1,1). At the end of stabilization period, the same tendency observed for F1 is also verified for measures taken in F2: outer layers (tangential direction) present predominant compressive strains while inner layers (longitudinal direction) present predominant tensile strains. Despite this tendency, obtained mean strains are very close to zero meaning that successive RH cycles do not causes cumulative strains. Regarding Z direction, inner layers of F2 answer to RH cycles as expected, presenting shrinking/swelling movements for drying/wetting periods, respectively (Table (I) 2:13). The influence of moisture flow conditions is verified again at day 161, in which slow desorption process do not let C2 and C4 to reach compressive strains. Independent of moisture flow conditions, radial direction of inner layers seems to be affected by successive RH cycles specially regarding drying periods. It can be observed in Figure (I) 2:23 that last drying period present lower compressive strains for C1 and C3 while C2 and C4 present tensile strains. Decreases presented by C1 and C3 are of 66,6% and 30,7%, respectively. At the end of stabilization period inner layers tend to be under compressive strains for all test configurations (-0,013 ≤ ] v[t‚_U[V\_¬O-_«No«-_¡Ÿ_• ≤ -0,004). Surprisingly, and similarly to what was observed for Y direction, outer layers tend to present predominant tensile strains at last RH cycle. Observing full field strains distribution for Z direction depicted in Table (I) 2:12, outer layers also present tensile and compressive strains depending on growth rings orientation. However, with the exception of C3, even during drying periods tensile strains tend to be predominant (0,006 ≤ ] v[t‚_U[V\_N¯N_«N/«O/«-_ Ÿ_• ≤ 0,011). Table (I) 2:12 presents C1 as example but, remaining graphs obtained for F2 in Z direction are depicted in Annex 1:5. Configurations that restricts moisture flow in tangential direction (inner layers) are the only configurations that present shrinking movements Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 87 between first wetting and last drying periods for outer layers. This may mean that sealing tangential direction changes the effect of crosswise lamination as the number of cycles increase. Regarding wetting periods, the differences obtained between days 56 and 182 are not so substantial. Higher differences were registered for C3 which presents strains 29,4% lower at day 182. Successive RH cycles seems to reduce the range between drying and wetting periods and at the end of stabilization period strains tend to zero for all test configurations. Strains measured in F3 are summarized in Table (I) 2:14 and Figure (I) 2:24 for all four configurations considering inner and outer layers separately. As expected, the differences observed between inner and outer layers are not so relevant in this CLT lateral face once the layer that present tangential movements (inner layer) is restricted by layers that shrinks/swells in longitudinal direction (outer layers). Observing full field strains distribution, regarding X direction, presented in Table (I) 2:15 this restriction is very obvious once the obtained patterns are much more homogeneous than those obtained for F2_Y (remaining graphs obtained for F3 in X direction are depicted in Annex 1:4). This way, graphs depicted in Figure (I) 2:24 are more in line with what was observed for inner layers from F2, presenting high tensile strains during last drying period, shrinking tendency during last wetting period and swelling tendency during stabilization period. Despite the small differences observed for outer layers of F2, mean restrained strains measured in F2 and F3, for Y and X directions, either for inner or outer layers varies within the same range (-0,005 ≤ ] v[t‚_U[V\_®/¿ ≤ 0,009). Evaluating results obtained for F3 in Z direction the tendency is inverted when comparing with results obtained for F2. Movements in radial direction are not restrained by crosswise lamination so, the range of compressive/tensile strains are just dependent on grain orientation. This way, ranges obtained for F2 (outer layers) and F3 (inner layers) are very close to each other: -0,016 ≤ ] v[t‚_U[V\_žO_ Ÿ_• ≤ 0,017 and -0,009 ≤ ] v[t‚_U[V\_ž¬_¡Ÿ_• ≤ 0,015. Ranges obtained for F2 (inner layers) and F3 (outer layers) are slightly higher for F2: -0,026 ≤ ] v[t‚_U[V\_žO_¡Ÿ_• ≤ 0,023 and -0,016 ≤ ] v[t‚_U[V\_ž¬_ Ÿ_• ≤ 0,018. However the behavior through the RH cycles is similar (Figure (I) 2:23 and Figure (I) 2:25). P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 94 Table (I) 2:18. Descriptive statistics of full-field restrained strains measured on specimens from group B in Z direction, for inner and outer layers, during first DIC mapping. RESTRAINED STRAIN ( 0 2@B ) MAXIMUM MEAN MINIMUM COV Z DIRECTION INNER LAYERS OUTER LAYERS DAY DAY 35 56 161 182 324 35 56 161 182 324 C1 -0,014 0,050 -0,012 0,025 0,013 -0,004 0,082 0,023 0,060 0,024 -0,024 0,041 -0,019 0,014 0,007 -0,021 0,043 -0,019 0,024 0,007 -0,037 0,027 -0,032 0,006 0,001 -0,042 0,018 -0,055 0,005 -0,006 -0,17 0,11 -0,13 0,21 0,23 -0,33 0,27 -0,42 0,43 0,60 C2 -0,006 0,047 -0,006 0,035 0,026 -0,006 0,073 0,000 0,048 0,048 -0,017 0,028 -0,019 0,023 0,012 -0,013 0,038 -0,012 0,020 0,014 -0,028 0,014 -0,037 0,008 0,000 -0,028 0,014 -0,029 0,001 0,001 -0,16 0,12 -0,22 0,17 0,35 -0,30 0,28 -0,38 0,42 0,37 C3 -0,017 0,044 -0,007 0,039 0,609 -0,01 0,08 0,01 0,07 0,04 -0,025 0,029 -0,024 0,023 0,008 -0,02 0,04 -0,02 0,03 0,01 -0,034 0,000 -0,037 0,011 -0,005 -0,04 0,02 -0,05 0,00 -0,01 -0,18 0,15 -0,13 0,11 2,35 -0,30 0,28 -0,44 0,40 0,65 C4 -0,006 0,052 -0,002 0,026 0,012 -0,004 0,066 -0,001 0,041 0,026 -0,009 0,033 -0,007 0,017 0,007 -0,014 0,034 -0,011 0,020 0,007 -0,015 0,014 -0,012 0,006 0,000 -0,026 0,012 -0,037 0,003 -0,001 -0,12 0,12 -0,24 0,17 0,25 -0,27 0,26 -0,36 0,35 0,38 Figure (I) 2:27. Mean values of full-field restrained strains distribution measured on specimens from group B in Z direction, for inner and outer layers, during first DIC mapping. -0,036 -0,024 -0,012 0,000 0,012 0,024 0,036 0,048 DAY 35 56 161 182 324  strain B_C1 B_C2 B_C3 B_C4 -0,036 -0,024 -0,012 0,000 0,012 0,024 0,036 0,048 DAY 35 56 161 182 324  strain B_C1 B_C2 B_C3 B_C4 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 95 The results obtained for inner layers presented substantial differences when comparing C1 with other configurations. In one hand, C4 presented restrained strains that are 63%, 20% and 63% lower than values registered for C1 in days 35, 56 and 161, respectively. In the other hand, results obtained for day 182 stand out, presenting higher restrained strains when comparing with value registered for C1 (Ɛ v[t‚_¡Ÿ_NrO_•_U[V\ = -0,014): 64% for C2 and C3 and 21% for C4. Taking days 35 and 56 as reference, it was observed that restrained compressive and tensile strains tend to decrease at last drying/wetting periods for all tested configurations either for outer and inner layers. However, differently to what was observed for Y direction, influence of moisture flow conditions is not so obvious, once higher decreases were registered for configurations C1 and C4, instead of C2 and C4. Differences registered between first and last wetting periods were the highest, however lower than those observed for Y direction. Regarding outer layers, C2 and C4 still suggest that sealing longitudinal direction delays the absorption/desorption process, presenting decreases of restrained tensile strains of 47% and 41%. However, inexplicably, C1 stands out presenting decreases of restrained tensile strains of 44% and 65%, for outer and inner layers, respectively. Finally, and similarly to results obtained for Y direction, at the end of stabilization period tensile strains are predominant either for inner or outer layers. Table (I) 2:19 presents the ratio between restrained strains obtained for Y and Z directions ¦¤=• Á³Âà ® Á³Âà § for all tested configurations at all test days, for outer and inner layers. Considering outer layers, obtained ratios represent the relation between strains measured for radial and tangential ¦¤ Ÿ=• Á³ÂÃ_»Ä ® Á³ÂÃ_»Ä =ÅƧ directions and for inner layers obtained ratios represent the relation between radial and longitudinal ¦¤¡Ÿ=• Á³ÂÃ_¾Ä ® Á³ÂÃ_¾Ä =ÅŸ§ directions. Table (I) 2:19 also presents the relation between adjacent layers by the ratio between tangential and longitudinal directions ¦¤®=® Á³ÂÃ_»Ä ® Á³ÂÃ_¾Ä =ÆŸ§. As expected, ¤ Ÿ (R/T) presented the lowest values, once it is relating the directions with higher moisture movements. Results obtained for outer layers suggest that DEF is affected by successive RH cycles and not affected, at least not clearly, by moisture flow conditions. Considering all four configurations and days 35, 56 and 161, DEF is in a range of 1.5 - 2.3, while for day 182 a higher difference between radial and tangential directions is verified, obtaining ratios that are around 3.3. At the end of stabilization period C3 stands out with a ratio of 4.4 while the remaining configurations present a range of 1.3 - 2.4. Similarly, the results obtained for DGF (R/L) also suggest that just successive RH cycles influence the relation between radial and longitudinal directions. Lower ratios were registered for days 35, 56 and 161: at days 35 and 161 DGF is in a range of 3,9 – 13,4, while at day 56 DGF is in a range of 2,4 – 3,8. Day P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 96 182 present the higher ratios in a wide range of 24,7-179,6. At the end of stabilization period, DGF tend to decrease presenting a range of 0,7 – 1,3. Regarding the relation between tangential and longitudinal directions (DH) the tendency is exactly the same observed for DEF and DGF , what means that DH (T/L) tends to increase from first to last RH cycle. Days 35 and 161 registered ranges of 2,4 – 3,9 and 4,4 – 7,3, respectively. Day 56 registered a range of 1,7 – 1,9, while day 182 stands out with a range of 12,3 – 50,5. At the end of stabilization period, DH tend to decrease presenting a range of 0,2 – 0,7. As already mentioned, some possible influence of moisture flow conditions on analyzed ratios is not clear. It must be referred that DGF and DH presented values at day 182 for C2 and C3 that are considerably higher than those observed for remaining configurations. However, the reason for this phenomenon is not clear, once what these two configurations have in common is sealing the radial direction. Table (I) 2:20 exhibit the color fill graphs of typical behavior of full field restrained strains registered distribution for Y and Z direction and shear, considering C1 and all test days. C1 is presented as example and remaining full field graphs can be observed in Annex 1:6 and Annex 1:7. Analyzing color fill graphs, it is evident that restrained strains are higher for radial direction (Z), and outer and inner layers are clearly identified. Regarding Y direction, lower compressive/tensile strains are observed for inner layers, fact that clearly restricts the movements of outer layers, especially closer to the outer borders of specimens. Due to the location and orientation of growth rings (see figure at left in Table (I) 2:20), higher strains are located at central part of outer layers. Furthermore, the effects of successive RH cycles are visible at day 182, in which restrained strains observed for middle layers suggest timber shrinkage rather than the expected swelling. Regarding Z direction, outer and middle layers are even easily identified because cross lamination does not influence the timber movements in radial direction. Also due to the location and position of growth rings, but differently of what was observed for Y direction, higher compressive/tensile strains are located at the outer parts of outer layers. Here, the influence of successive RH cycles is also verified at day 182, in which tensile strains are lower than those observed for day 56. Mean released strains measured during second DIC mapping are summarized at Table (I) 2:21 and depicted in Figure (I) 2:28. Obtained mean values considered a central part of each independent layer (an area of around 10 x 107 subsets) in order to avoid any possible effect of sawing action. This DIC map Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 97 just considered strains measured in Y direction once the goal was to quantify strains resulted from separation of different layers. According to mean values presented in Table (I) 2:21, inner layers do not exhibit a relation between drying/wetting periods and compressive/tensile strains. Tensile strains are predominant, which means that independent of moisture content levels CLT inner layers tend to swell after being released. Furthermore, different moisture flow conditions do not affect the results. Obtained released strains are close to zero for majority of test days, however high tensile released strains are registered at last drying period (0,002 ≤ ] v[•_¡Ÿ_N¯N_®_U[V\ ≤ 0,003). Contrary to what was observed for group A_F1, successive RH cycles lead to an increase of released strains, especially during drying periods. C4 presents high tensile strains at day 56 probably due to its small area open to moisture flow (] v[•_«-_¡Ÿ_q¯_®_U[V\ = 0,002). Taking day 0 as reference (] v[•_«N_¡Ÿ_p_®_U[V\ = 0,000), it can be concluded that successive RH cycles do not causes cumulative moisture induced strains, once -0,001 ≤ ] v[•_¡Ÿ_¬O-_®_U[V\ ≤ 0,000. Table (I) 2:19. Ratio between restrained strains obtained for Y and Z directions for inner and outer layers, during first DIC mapping of group B. RATIO DAY CONFIGURATIONS C1 C2 C3 C4 D=¨Ç 123 H 123 ª D GF =¨ ? Fª 35 10,6 5,5 8,6 3,9 56 3,5 2,4 2,4 3,8 161 10,6 13,3 13,4 5,5 182 24,7 179,6 65,0 38,8 324 0,7 1,3 0,9 0,8 D EF =¨ ? ©ª 35 2,3 1,8 2,0 2,1 56 2,0 1,9 2,0 2,0 161 1,5 1,8 1,7 1,8 182 3,3 3,2 3,2 3,7 324 1,4 2,4 4,4 1,3 Δ =¦ H123_EF H123_GF § D H =¨ © Fª 35 3,9 2,4 3,5 2,7 56 1,8 1,7 1,8 1,9 161 7,3 4,7 6,8 4,4 182 12,8 50,5 22,6 12,3 324 0,5 0,7 0,2 0,6 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 98 Table (I) 2:20. Typical behavior (C1) of full field restrained strain distribution in Y and Z directions and in shear observed group B, before cut (BC) the specimens. RESTRAINED STRAINS FOR C1 ( 0 2@B ) TEST DAY RH [%] Y Z SHEAR PICTURES SCALE DAY 35 [ RH =30%] DAY 56 [ RH =90%] DAY 161 [ RH =30%] DAY 182 [ RH =90%] DAY 324 [ RH =65%] Note: Red lines at pictures represent the end of ROI considered by DIC measurements. (dimenstions are in mm) Y Z 160 24 93 -4,8E-02 -4,2E-02 -3,6E-02 -3,0E-02 -2,4E-02 -1,8E-02 -1,2E-02 -6,0E-03 0,0E+00 6,0E-03 1,2E-02 1,8E-02 2,4E-02 3,0E-02 3,6E-02 4,2E-02 4,8E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 99 Regarding outer layers, and similarly to tendency observed for inner layers, the influence of different moisture flow conditions was not observed. However, contrary to what was observed for inner layers, there is a relation between compressive/tensile strains and drying/wetting periods, especially during first RH cycle. Compressive released strains registered at day 35 stand out presenting values between -0,002 and -0,003 while last drying period present values between 0,000 and 0,001. This behavior is more in line of what was observed for group A_F1, in which successive RH cycles suggest the tendency of strain reduction as the number of cycles increase. This way, it seems that during first RH cycle inner layers restrict moisture movements of outer layers, but during last RH cycle it is not observed. Taking day 0 as reference (] v[•_«N_ Ÿ_p_®_U[V\ = 0,001), it can be concluded that reduced tensile strains observed at day 0 become into substantial compressive strains after successive RH cycles (] v[•_ Ÿ_¬O-_®_U[V\ = - 0,002. This value was considered significant once they are very close to those observed for first drying period. Typical behavior of full field released strain distribution in Y direction for second DIC map is depicted at Table (I) 2:22 (remaining graphs of released strains measured for specimens of group B can be seen in Annex 1:8). Analyzing color fill graphs, it is possible to observe that just at last RH cycle, the position and orientation of timber growth rings of outer layers is noticeable. At day 161 compressive strains are located at the side of inner layers while tensile strains are located on the outer part, but the opposite tendency is observed at day 182. Furthermore, regarding inner layers, tensile strains are evenly distributed at day 161 while, regarding compressive strains, the most uniform patterns are observed at days 35 and 324, either for outer or inner layers. 2.4.1.4 Correlation between dynamic and static Modulus of Elasticity Modulus of elasticity (MoE) is a fundamental parameter to calculate internal stresses (σ) presented either in CLT slices or in CTL different layers. So, due to the huge number of test specimens, MoE was measured with a dynamic method ( wS\ ) and then converted to a static MoE ( t‚V ). To obtain the relation between ( wS\ ) and ( t‚V ) a group of 20 CLT slices were static and dynamically tested. First the entire CLT slice were tested and then different layers were tested separately, perpendicular and parallel to the grain for outer and inner layers, respectively. This way, two different geometries were tested, namely CLT slices (170 x 24 x 93 mm) and CLT individual layers (170 x 24 x 29 mm). P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 100 Table (I) 2:21. Descriptive statistics of full-field released strains measured on specimens from group B in Y direction, for inner and outer layers, during second DIC mapping. RELEASED STRAIN ( 0 2C ) MAXIMUM MEAN MINIMUM COV Y DIRECTION INNER LAYERS OUTER LAYERS DAY DAY 0 35 56 161 182 324 0 35 56 161 182 324 C1 0,001 0,002 0,002 0,005 0,002 0,002 0,011 0,001 0,004 0,014 0,003 0,002 0,000 - 0,001 0,000 0,002 0,000 - 0,001 0,001 - 0,003 0,001 0,000 0,000 - 0,002 -0,003 -0,004 -0,002 0,000 -0,002 -0,002 -0,005 -0,008 -0,002 -0,310 -0,003 -0,009 -1,10 -0,58 -0,29 0,15 -1,89 -0,63 6,61 -0,50 0,76 -2,19 2,05 -0,56 C2 0,001 0,003 0,010 0,002 0,004 0,003 0,004 0,005 0,004 0,004 0,000 0,000 0,002 0,000 - 0,001 - 0,003 0,001 0,000 0,001 - 0,002 -0,011 -0,002 -0,001 -0,002 -0,005 -0,009 -0,002 -0,005 -0,016 -0,009 -1,70 11,79 0,24 -0,76 -0,64 -0,60 0,57 6,62 0,79 -0,60 C3 0,001 0,003 0,005 0,004 0,002 0,002 0,005 0,007 0,012 0,014 0,000 0,000 0,003 0,000 0,000 - 0,002 0,000 0,001 - 0,001 - 0,002 -0,002 -0,002 0,000 -0,003 -0,003 -0,033 -0,004 -0,029 -0,007 -0,032 -1,67 1,36 0,15 2,78 -1,26 -1,13 2,48 1,72 -0,97 -0,78 C4 0,003 0,007 0,005 0,003 0,001 0,002 0,007 0,008 0,003 0,005 - 0,001 0,002 0,003 0,000 0,000 - 0,003 0,003 0,000 0,000 - 0,002 -0,005 -0,003 0,000 -0,004 -0,003 -0,008 -0,003 -0,004 -0,004 -0,006 -0,92 1,35 0,14 8,09 -0,81 -0,59 0,77 10,53 -7,68 -0,55 Figure (I) 2:28. Mean values of full-field released strains distribution measured on specimens from group B in Y direction, for inner and outer layers, during second DIC mapping, after cut specimens (AC). DAY 35 56 161 182 324  strain B_C1_IL_AC B_C2_IL_AC B_C3_IL_AC B_C4_IL_AC -0,004 -0,002 0,000 0,002 0,004 0,006 -0,006 DAY 35 56 161 182 324  strain B_C1_OL_AC B_C2_OL_AC B_C3_OL_AC B_C4_OL_AC -0,004 -0,002 0,000 0,002 0,004 0,006 -0,006 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 101 Dynamic MoE was defined by means of ultrasound tests using a Pundit Lab equipment from Proceq with 150 kHz frequency transmission transducers. Ultrasound equipment measures the velocity of sound propagation (È u ), which can be related with  wS\ if the density () of timber element is known (H. S. Sousa et al., 2014). Equation (2:4) shows how  wS\ was obtained.  wS\ = É u O ∙  (2:4)  wS\ Dynamic modulus of elasticity, in MPa; v Ë Velocity of propagation, in m/s; ρ Wood density, in kg/m 3 . As depicted in Figure (I) 2:29 and Figure (I) 2:30, ultrasound acquisitions in CLT slices considered four different measurements: crossing all three layers at the same time (cross 1 & cross 2), crossing two layers per time (cross 3 & cross 4), measuring one layer per time when they are still glued together (glued A+B+C) and measuring one layer per time after separate layers (cut A+B+C). This procedure aimed to guarantee the best correlation between dynamic and static MoE. Due to reduced dimensions of test specimens,  t‚V was obtained by compressive. Also due to reduced dimensions of test specimens, both test setup and procedure (Table (I) 2:23) do not follow strictly any standard. However, parameters like ℎ p (96mm) and test time (180 +/- 60 s) were based on recommendations present in EN 408: 2003. Specimens were compressed by a load cell integrated on a hydraulic system, and a different test procedure was defined for each specimen configuration considering specimen dimensions as well as orientation of timber fibers (Table (I) 2:23). Correlation between  wS\ and  t‚V were made by means of linear fittings, but previously some outliers had to be excluded from the data set. The definition of outliers were made by the analysis of results obtained by the ratio Í´µÂ ÍÎÏà . Results obtained for this relation are detailed in Annex 1:9. Nevertheless, sampling is still between 17 and 37 specimens per each group of tested specimens. Descriptive statistics of values obtained for  wS\ and  t‚V are presented in Table (I) 2:24, while linear obtained fittings are depicted in Figure (I) 2:29 and Figure (I) 2:30. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 102 Table (I) 2:22. Typical behavior (C1) of full field released strain distribution in Y direction observed for group B, after cut (AC) the specimens. Y DIRECTION TEST DAY RH [%] C1 PICTURES C3 PICTURES SCALE DAY 0 [ RH =65%] DAY 35 [ RH =30%] DAY 56 [ RH =90%] DAY 161 [ RH =30%] DAY 182 [ RH =90%] DAY 324 [ RH =65%] Note: Red lines at pictures represent the end of ROI considered by DIC measurements. (dimentions are in mm) Y Z 160 24 93 -4,8E-02 -4,2E-02 -3,6E-02 -3,0E-02 -2,4E-02 -1,8E-02 -1,2E-02 -6,0E-03 0,0E+00 6,0E-03 1,2E-02 1,8E-02 2,4E-02 3,0E-02 3,6E-02 4,2E-02 4,8E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 103 Table (I) 2:23. Test procedures for determination of E ÑÒe for CLT slices and individual CLT layers. SPECIMEN MAXIMUM LOAD [ KN ] VELOCITY [ KN / S ] GRAPH CLT SLICE 3,0 0,10 CLT LAYER PERPENDICULAR TO THE GRAIN 0,8 0,03 PARALLEL TO THE GRAIN 1,1 0,04 Analyzing  wS\ obtained for CLT slices, it is obvious that higher dynamic modulus of elasticity are related with higher propagation velocities and consequently related with measurements that included longitudinal direction (inner layers), namely measurements taken for three layers individually either before cutting (mean = 16043 MPa; CoV=0,47) or after cutting (mean=27342 MPa; CoV=0,23). However, due to high CoV values obtained for measurements taken with the three layers still glued together (0,38 ≤ CoV ≤ 0,47), obtained linear fittings do not exhibit a good correlation with s O values below 0,40 (Figure (I) 2:29). So, best correlation was obtained through the mean obtained by measurements taken from separated layers individually, which presents a reduced CoV value that is closer to that obtained by static tests (/!È wS\ =0,23 and /!È t‚V =0,25). This way, relationship between  wS\ and  t‚V was determined as  t‚V =0,15· wS\ + 497 with a s O =0,64. Regarding correlations performed with CLT individual layers, both linear fittings obtained s O values above 0,51 suggesting relations between  wS\ and  t‚V that are acceptable:  t‚V =0,15· wS\ + 29,8 with a s O =0,51 and :  t‚V =0,12· wS\ + 2532 with a s O =0,74, for outer and inner layers, respectively (Figure (I) 2:30). 2.4.1.5 Calculation of internal stresses As described in section 2.3.1 either group A or B were used to obtain a second DIC map. This was done immediately after cutting specimens into slices (for group A) or after separation of different layers (for group B). Strains obtained with these measurements quantify the restriction of moisture movements forced by cross-lamination. Similarly to work developed by Jönsson (2004), Gereke (2009) and Angst & Malo (2012b), internal stresses were calculated by the relation T U[V\ =] U[V\ ∙ (†) , in which  (†) is 0 50 100 150 200 0 1 2 3 4 Load [kN] Time [s] CLT slices Layer_Parallel to the grain Layer_Perpendicular to the grain P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 110 the configuration that presents a behavior completely out of pattern is C3. It is possible that sealing of tangential and radial directions cause some effects on outer layers, especially at last wetting period in which outer layers present predominant compressive stresses while inner layers present predominant tensile stresses (Figure (I) 2:34 (d)). However, before take final conclusions more research is required, once this phenomenon can be just a result of reduced sampling. (a) (b) (c) (d) (e) Figure (I) 2:34. Mean values of released stresses measured on outer and inner layers of specimens from group B for all four test configurations and all five test days. 2.4.2 LVDTs acquisition As explained at point 2.3.2, constant measurements of moisture induced movements were taken until the end of last wetting period (day 182) of RH cycle using two specimens similar to C1 from group A. Taking initial dimensions of specimens (day 0) as reference, Table (I) 2:27 presents mean compressive OL IL OL -8 -6 -4 -2 0 2 4 STRESS [MPa] B_C1_35_Y_AC B_C2_35_Y_AC B_C3_35_Y_AC B_C4_35_Y_AC -5 0 5 10 15 20 25 STRESS [MPa] B_C1_56_Y_AC B_C2_56_Y_AC B_C3_56_Y_AC B_C4_56_Y_AC OL IL OL -8 0 8 16 24 32 40 STRESS [MPa] B_C1_161_Y_AC B_C2_161_Y_AC B_C3_161_Y_AC B_C4_161_Y_AC OL IL OL -6 -4 -2 0 2 4 6 STRESS [MPa] B_C1_182_Y_AC B_C2_182_Y_AC B_C3_182_Y_AC B_C4_182_Y_AC OL IL OL -4 -3 -2 -1 0 1 2 STRESS [MPa] B_C1_324_Y_AC B_C2_324_Y_AC B_C3_324_Y_AC B_C4_324_Y_AC OL IL OL Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 111 restrained strains registered at the end of each drying/wetting period, while Figure (I) 2:35 depicts the continuous compressive strains registered during the 182 test days. Both, Table (I) 2:27 and Figure (I) 2:35 present obtained results for all three monitored directions (X, Y and Z). Regarding X direction (longitudinal direction of outer layers), just the measurements of one specimen was considered for this analysis, once due to technical issues the other LVDTs does not take the measurements properly. As expected, X direction was the one with lower moisture induced movements. However, it presents predominant negative compressive strains during all the RH cycles, which means that swelling movements tend to be wider than shrinking movements. This way, CLT panels never recover its initial shape keeping it always larger. Furthermore, negative compressive strains (-0,01% to -0,34%) are registered for drying periods, meaning that X direction present exactly the opposite of expected tendency. In other words, X direction present a clear tendency to swell when timber dries. This tendency tends to emphasize as the number of cycles increase and must be related with crosswise lamination. Differently, Y direction present predominant positive compressive strains either for drying or wetting periods. Contrary to what was observed for X direction, shrinking movements tend to be wider than swelling movements. Furthermore, compressive strains tend to increase as the number of cycles also increase either for drying or wetting periods. It can be observed in Figure (I) 2:35 that for drying periods timber does not reach an equilibrium, while for wetting periods timber tends to stabilize few days after RH being changed. At the end of last drying period (day 161) CLT panel is 2,34% smaller in its Y direction and present a reduced moisture content of 9,8%. Surprisingly, at the end of last wetting period CLT panel is 1,09% smaller when comparing with its initial dimensions, despite its moisture content being around 20,6%. Z direction is the one that present the expected relation between positive/negative compressive strains and drying/wetting periods, respectively. Contrary to what was observed for Y direction, specimens show a tendency to stabilize at the end of drying periods, while during wetting periods specimens indicate the tendency to keep swelling (Figure (I) 2:35). For all RH cycles, values registered for wetting periods are clearly more substantial than those registered for drying periods. Furthermore, either shrinkage or swelling movements tend to decrease as the number of cycles increase. Considering initial dimensions of CLT panels, at the end of last wetting period CLT panel is 1,90% larger while at the end of last drying period CLT panel is 0,38% smaller. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 112 Table (I) 2:27. Restrained compressive strains considering initial dimension of specimens at day 0. COMPRESSIVE STRAINS [%] DIRECTION LVDT DAY / RELATIVE HUMIDITY 35 30% 56 90% 75 30% 96 90% 114 30% 136 90% 161 30% 182 90% X 173710 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 173711 -0,01 0,07 -0,13 -0,00 -0,30 -0,04 -0,339 -0,073 MEAN -0,01 0,07 -0,13 -0,00 -0,30 -0,04 -0,34 -0,07 Y 173714 1,21 0,427 1,86 0,70 2,06 0,91 2,38 1,15 173715 1,18 0,24 1,90 0,72 1,77 0,94 2,30 1,04 MEAN 1,19 0,33 1,88 0,71 1,92 0,92 2,34 1,09 Z 173712 0,98 -1,97 0,66 -1,96 0,56 -1,98 0,48 -1,89 173713 0,91 -2,35 0,28 -2,10 0,29 -2,06 0,27 -1,89 MEAN 0,95 -2,16 0,47 -2,03 0,42 -2,02 0,38 -1,90 Figure (I) 2:35. Linear shrinkage considering initial dimension (day 0) of test specimens. Observing compressive/tensile strains separately and taking the dimensions at the beginning of each drying/wetting period as reference, differences between successive RH cycles can be compared easily. Table (I) 2:28 presents percentages of compressive strains obtained at the end of all four drying periods for three analyzed directions, and Figure (I) 2:36 depicts curves of compressive strains registered continuously for the same periods and CLT directions. Analyzing the data, it can be observed that for all directions the first drying period is the less significant, especially for X and Z directions. Y direction present the greatest similarity between different drying periods (1,19% to 1,55%) and is the unique direction in which specimens do not stabilize. Differences between minimum and maximum linear shrinkage are of 0,29%, 0,36% and 1,61% for X, Y and Z directions, respectively. Exhibiting negative compressive strains (-0,01% to -0,30%), X direction tends to stabilize around 5 th day of each drying period. This means that during first five days while CLT panel shrinks in Y direction it swells in X direction. As a consequence of 35 75 114 161 -3 -2 -1 0 1 2 3 Mean Z Mean Y Mean X Linear Shrinkage [%] 56 96 136 182 DAY XY Z Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 113 wider shrinking movements (0,95% to 2,56%), Z direction tends to stabilize later, around in the middle of each drying period. Table (I) 2:29 presents tensile strains (%) obtained at the end of all four wetting periods for three analyzed directions, and Figure (I) 2:37 depicts curves of tensile strains registered continuously for the same periods and CLT directions. Similar to what was observed for drying periods, first wetting period is the less significant for X and Y direction. On contrary, Z direction present the first wetting period as the most significant and successive periods present a descendent tendency. Higher differences observed between successive drying periods for X and Z directions are less significant regarding wetting periods. Differences between minimum and maximum tensile strains are of 0,19%, 0,41% and 0,84% for X, Y and Z directions, respectively. Despite low percentages, X direction present again negative values (-0,08% to -0,27%). During swelling periods Y and Z direction switch their behaviors and Y direction stabilizes around 10 th day of each period, while Z direction swells continuously until the end of wetting period. As expected, Z direction still present higher percentages for tensile strains (2,23% to 3,07%) than Y direction (0,87% to 1,28%). The effect of successive RH cycles on compressive/tensile strains is not well studied and few studies are found regarding this subject. Sousa (2010) performed some experiments with solid maritime pine ( Pinus pinaster, ait ), submitting specimens to three successive drying/wetting cycles and evaluating the shrinkage/swelling coefficients during stabilization periods for radial and tangential directions. Results shown that moisture movements tend to increase as the number of cycles also increases concluding that timber fibers suffer some kind of laxity, changing its hygroscopic behavior. Regarding present study, this tendency was also observed, with the exception of swelling movements in radial direction. Comparing restrained strains obtained with DIC technique and LVDTs measurements for first and last RH cycles some important differences were registered. Table (I) 2:30 presents the mean values of restrained strains obtained by means of both techniques for first and last RH cycles and for all three measured directions. Figure (I) 2:38 depicts the comparative curves. Mean values presented for DIC technique are obtained from average of full-filed measurements considering the entire ROI of F1 for X and Y directions and F2 and F3 for Z direction (Figure (I) 2:10), while mean values presented for LVDTs acquisition are obtained by the linear measurements, according to test setup (Figure (I) 2:12), taken at each test day. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 114 Table (I) 2:28. Compressive strains considering initial dimension of specimens at the beginning of each drying period. COMPRESSIVE STRAINS [%] DIRECTION LVDT DRYING PERIODS 1 st 2 nd 3 rd 4 th X LVDT _173710 n.a. n.a. n.a. n.a. LVDT _173711 -0,01 -0,21 -0,29 -0,30 MEAN -0,01 -0,21 -0,29 -0,30 Y LVDT _173714 1,21 1,44 1,37 1,49 LVDT _173715 1,18 1,66 1,06 1,38 MEAN 1,19 1,55 1,22 1,43 Z LVDT _173712 0,98 2,56 2,43 2,36 LVDT _173713 0,91 2,55 2,30 2,24 MEAN 0,95 2,56 2,37 2,30 (a) (b) (c) Figure (I) 2:36. Curves of compressive strains considering initial dimension of each cycle with 30% RH. (a) X direction; (b) Y direction; (c) Z direction. 0 4 8 12 16 20 24 28 32 36 -0,5 -0,4 -0,3 -0,2 -0,1 0,0 0,1 X direction (from 90% RH to 30% RH) Linear Shrinkage [%] Time [days] 1 st _30% RH_X 2 nd _30% RH_X 3 rd _30% RH_X 4 th _30% RH_X 0 4 8 12 16 20 24 28 32 36 -0,4 0,0 0,4 0,8 1,2 1,6 2,0 Linear Shrinkage [%] Time [days] 1 st _30% RH_X 2 nd _30% RH_X 3 rd _30% RH_X 4 th _30% RH_X 0 4 8 12 16 20 24 28 32 36 -0,6 0,0 0,6 1,2 1,8 2,4 3,0 Linear Shrinkage [%] Time [days] 1 st _30% RH_X 2 nd _30% RH_X 3 rd _30% RH_X 4 th _30% RH_X XY Z Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 115 Table (I) 2:29. Tensile strains considering initial dimension of specimens at the beginning of each wetting period. TENSILE STRAINS [%] DIRECTION LVDT WETTTING PERIODS 1 st 2 nd 3 rd 4 th X LVDT _173710 n.a. n.a. n.a. n.a. LVDT _173711 -0,08 -0,13 -0,25 -0,27 MEAN -0,08 -0,13 -0,25 -0,27 Y LVDT _173714 0,79 1,19 1,18 1,27 LVDT _173715 0,95 1,20 0,86 1,29 MEAN 0,87 1,19 1,02 1,28 Z LVDT _173712 2,92 2,58 2,52 2,34 LVDT _173713 3,22 2,33 2,32 2,12 MEAN 3,07 2,46 2,42 2,23 (a) (b) (c) Figure (I) 2:37. Curves of tensile strains considering initial dimension of each cycle with 90% RH. (a) X direction; (b) Y direction; (c) Z direction. 0 4 8 12 16 20 24 28 32 36 -0,5 -0,4 -0,3 -0,2 -0,1 0,0 0,1 Linear Shrinkage [%] Time [days] 1 st _30% RH_X 2 nd _30% RH_X 3 rd _30% RH_X 4 th _30% RH_X 0 4 8 12 16 20 24 28 32 36 -0,4 0,0 0,4 0,8 1,2 1,6 2,0 Linear Shrinkage [%] 1 st _30% RH_X 2 nd _30% RH_X 3 rd _30% RH_X 4 th _30% RH_X 0 4 8 12 16 20 24 28 32 36 -0,7 0,0 0,7 1,4 2,1 2,8 3,5 Z direction (from 30% RH to 90% RH) Linear Shrinkage [%] Time [days] 1 st _30% RH_X 2 nd _30% RH_X 3 rd _30% RH_X 4 th _30% RH_X XY Z P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 116 Regarding X direction (Figure (I) 2:38 (a)), it is possible to conclude that both techniques suggest that, as a consequence of successive RH cycles, tensile strains tends to be predominant as the number of cycles increase, however DIC technique present wider differences between compressive and tensile strains (- 0,005 to -0,001 for first RH cycle and 0,002 to -0,006 for last RH cycle). Concerning Y direction (Figure (I) 2:38 (b)), it is during the first wetting period that the most important difference happens, once according to LVDTs acquisition outer layers of CLT panels does not reach tensile strains. As a consequence, Y direction measured by LVDTs acquisition never presents tensile strains during the entire test period. When comparing first and last wetting periods, both techniques present similar decreases for tensile/compressive restrained strains. However, when comparing first and last drying periods, DIC technique suggest a decrease of compressive strains (-0,012 to -0,006) while LVDTs acquisition suggest an increase of compressive strains (-0,012 to -0,023). So, the final effect of RH cyclic changes on Y direction is different depending on measurement technique. While DIC technique suggest that the moisture induced movements reduce as the RH cycles increases, LVDTs acquisition suggest that successive RH cycles causes cumulative compressive strains on CLT panel. Finally, observing the behavior of Z direction (Figure (I) 2:38 (c)), higher differences between both techniques are registered for wetting periods. For first and second wetting periods DIC technique present restrained strains that are 56% and 63% lower than those registered by LVDTs acquisition, respectively. This difference is directly related with the way measurements were taken. While DIC technique used external lateral surfaces of CLT panels, LVDTs acquisitions were made perpendicular to the main face of CLT panel. This way, results suggest that timber swelling is higher in central parts of CLT panel than in its borders. Despite these differences both techniques suggest that successive RH cycles tends to nullify compressive strains and keep tensile strains predominant. 2.4.3 Caliper ruler Digital Caliper ruler were used to perform measurements on slices obtained after slicing specimens from group A. Measurements were taken on the length of CLT slices before and after separation of CLT layers. Each specimen was sliced into seven slices and central five slices were measured using a digital caliper ruler. However, the accuracy of this method is dependent on the operator technique as well as on the device quality. This way, obtained values present some contradictory results. Table (I) 2:31 presents mean released strains obtained for outer and inner layers, considering the measurements taken on all five slices of each specimen and Figure (I) 2:39 depicts the same released strains. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 117 Table (I) 2:30. Mean restrained values obtained for X, Y and Z directions using DIC technique and LVDTs acquisitions at first and last RH cycles. MEAN 0 2@B TEST DAY 35 56 161 182 X DIC -0,005 -0,001 0,002 -0,006 LVDT s 0,000 -0,001 0,003 0,001 Y DIC -0,012 0,008 -0,006 0,000 LVDT s -0,012 -0,003 -0,023 -0,011 Z DIC -0,012 0,009 0,001 0,007 LVDT s -0,009 0,022 -0,004 0,019 (a) (b) (c) Figure (I) 2:38. Curves comparing restrained strains obtained by DIC technique and LVDTs measurements. (a) X direction; (b) Y direction; (c) Z direction. 0 40 80 120 160 200 -0,03 -0,02 -0,01 0,00 0,01 0,02 0,03  rest [MPa] DAY LVDT'S_X DIRECTION DIC MEAN_X DIRECTION_F1 LVDT'S_X DIRECTION 0 40 80 120 160 200 -0,03 -0,02 -0,01 0,00 0,01 0,02 0,03  rest [MPa] DAY LVDT'S_Y DIRECTION DIC_MEAN_Y DIRECTION_F1 LVDT'S_Y DIRECTION 0 40 80 120 160 200 -0,03 -0,02 -0,01 0,00 0,01 0,02 0,03  rest [MPa] DAYS (DAYS) LVDT'S_Z DIRECTION DIC_MEAN_Y DIRECTION_F2/F3 LVDT'S_Z DIRECTION P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 118 Table (I) 2:31. Descriptive statistics of mean released strains measured on specimens from group A in Y direction, for inner and outer layers, using digital caliper ruler and considering measurements taken on all five slices. RELEASED STRAIN ( 0 2C ) MAXIMUM MEAN MINIMUM COV INNER LAYERS OUTER LAYERS DAY DAY 0 35 56 161 182 324 0 35 56 161 182 324 C1 0,001 0,001 0,000 0,001 0,002 0,004 0,000 0,000 -0,001 0,002 0,001 0,000 0,000 0,000 -0,001 -0,001 0,000 0,001 -0,001 -0,002 -0,002 -0,002 -0,002 -0,003 -0,001 -0,002 -0,003 -0,004 -0,001 -0,001 -0,002 -0,004 -0,004 -0,004 -0,004 -0,005 -9,51 -2,05 -1,05 -1,63 2,97 1,59 -0,47 -0,61 -0,38 -1,06 -0,83 -0,62 C2 0,001 0,001 0,002 0,001 0,003 0,000 0,000 0,002 0,001 0,000 0,000 0,000 0,000 0,000 0,001 -0,002 -0,001 -0,002 -0,001 -0,003 -0,001 -0,001 -0,005 -0,001 -0,001 -0,007 -0,002 -0,005 -0,007 -0,010 -5,63 1,29 -20,25 2,75 1,27 -0,67 -0,82 -1,00 -1,12 -0,77 C3 0,002 0,001 0,004 0,002 0,002 0,001 -0,001 0,001 0,003 0,003 0,000 0,000 0,000 0,000 0,000 -0,001 -0,002 -0,003 -0,001 -0,001 -0,002 -0,002 -0,002 -0,001 -0,003 -0,003 -0,004 -0,008 -0,004 -0,003 -84,84 -1,77 -4,06 3,61 12,67 -1,37 -0,43 -0,66 -1,30 -1,38 C4 0,001 0,001 0,001 0,001 0,003 0,002 0,001 0,000 0,000 0,003 0,000 0,001 0,000 0,000 0,000 -0,001 -0,001 -0,002 -0,001 -0,003 -0,002 0,000 -0,001 -0,002 -0,002 -0,004 -0,003 -0,004 -0,003 -0,005 -7,81 0,96 5,12 -2,31 3,59 -0,87 -1,05 -0,50 -0,56 -0,77 Figure (I) 2:39. Mean values of mean released strains measured on specimens from group A in Y direction, for inner and outer layers, using digital caliper ruler and considering measurements taken on all five slices. -0,004 -0,003 -0,002 -0,001 0,000 0,001 0,002 C1_AC_IL_Y C2_AC_IL_Y C3_AC_IL_Y C4_AC_IL_Y  strain DAY 35 56 161 182 324 -0,004 -0,003 -0,002 -0,001 0,000 0,001 0,002  strain DAY 35 56 161 182 324 C1_AC_OL_Y C2_AC_OL_Y C3_AC_OL_Y C4_AC_OL_Y Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 119 Despite the reduced strains, it is possible to observe the effect of drying and wetting periods either for inner or outer layers. However, some important differences are observed between test configurations. C1 and C3 present behaviors completely different from those presented by C2 and C4. Fact that could be justified by the effects of different moisture flow conditions. It is important to state here that similar differences were not observed for specimens from group B (in which DIC technique was applied) however, this fact can be related with different dimensions of original specimens. Anyway, results obtained for group B were taken here as a reference to compare the tendencies of all configurations during the RH cycles however, obtained values are not comparable. Considering inner layers, C2 and C4 present compressive/tensile strains for drying/wetting periods during first RH cycle, while C1 and C3 present predominant compressive strains. Despite the differences between configurations, at the end of stabilization period (day 324) all four configurations present predominant tensile strains (0,000 to 0,001). On contrary, results obtained for group B suggest predominant compressive strains for all four configurations at Day 324, while high tensile strains were registered at day 161. Released strains obtained for outer layers present predominant compressive tendency for all four configurations. However, while C2 and C4 present higher compressive strains associated to drying periods and lower compressive strains associated to wetting periods (agreeing with the tendency observed for specimens from group B), C1 and C3 present the opposite tendency during first RH cycle. Finally, at the end of stabilization period, and similarly to results obtained for group B, all configurations present predominant compressive strains (-0,001 to -0,003). As depicted in Figure (I) 2:40 and Figure (I) 2:41 released strains obtained for inner and outer layers were analyzed considering all five slices separately. Observing these figures, it is possible to verify that the expected symmetry centered at slice 3 (S3) was not always registered. However, some tendencies can be pointed out regarding released strains registered for inner layers (Figure (I) 2:40): released strains measured at S3 are closer when comparing drying and wetting periods (green and red lines, respectively); released strains measured for S3 at the end of wetting periods show the lowest tensile/compressive strains; at the end of stabilization period just C4 present compressive strains at S3, while remaining configurations present the highest tensile strains. Some other tendencies can be pointed out regarding outer layers (Figure (I) 2:41): wetting periods present closer results and a more evident symmetry while drying periods present higher differences and symmetry is not verified; at the end of stabilization period compressive strains tends to be higher, however a high variance is observed between different slices. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 126 3.1. Experimental Campaign The experimental campaign described in the present chapter was carried out at Institut für Holzbau und Holztechnologie - Graz Technical University (Austria) - with the background support of Professor Gerhard Schickhofer and PhD Andreas Ringhofer. This task was performed as a training program during a period of six months thanks to financial support of Portuguese foundation for science and technology (FCT) and Erasmus Placement program (University of Minho). 3.1.1 Main goals and parameters involved The performed experiments pretend to understand the influence of three different parameters on the axial withdrawal resistance of self-tapping screws (STS) inserted perpendicular to the main face of three layered CLT panels. The first parameter is related with simple moisture content changes on CLT. For that, it was considered a range limited by moisture content values established by service classes of Eurocode 5. In other words, the moisture content changes respected the limits imposed by service classes 1 and 2, described in point 2.3.1.3. of Eurocode 5. This way, the tests performed considered three different moisture levels, namely: 8% and 12%, respecting the limits of service class 1, and 18%, respecting the environmental conditions considered by service class 2. The decision of respect the limits of service classes 1 and 2 is justified by the warnings of CLT producers, who say that CLT is still considered a material incompatible with service class 3 (KLH-Massivholzplatten, 2011). Second parameter is related to the possible existence of gaps on the STS path through CLT thickness. It is called gaps to the line or space between two boards glued side by side in a CLT panel. To explore this parameter, CLT was produced considering the insertion of STSs through a different number of gaps across all CLT layers. CLT was produced with three layers, which allowed five different gap configurations, namely: reference (REF) – STS is inserted without the presence of gaps; gap in first layer (GAP_FL) – STS is inserted through one gap present in first layer of CLT panel; gap in middle layer (GAP_ML) - STS is inserted through one gap present in middle layer of CLT panel; gap in outer layers (GAP_OL) - STS is inserted through two gaps present in outer layers of CLT panel; gap in three layers (GAP_3L) - STS is inserted through three gaps present in all three layers of CLT panel. The drawings present in Figure (I) 3:1 illustrate these five configurations. Last parameter is related with gap width, which can be 0mm or 4mm. Gaps with 0mm (GAP0) were selected as the reference for the better scenario, once the boards glued side by side are touching each Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 127 other, and gaps with 4mm (GAP4) were selected to simulate the worst scenario, once there is a void between boards. The decision of a maximum width of 4mm was based on the research of Brandner et al. (2013), who presented a summary of main geometrical characteristics of European CLT producers. They concluded that the most common gap width varies between 2mm and 6mm. However, they also refer that producers are looking for improvements for CLT pressing procedures, namely lateral pressing, in order to reduce the width of gaps. So, considering these future improvements, present research fixed a gap width of 4mm as the worst scenario. The combination of these three parameters resulted in nine different test configurations and 270 withdrawal tests. Figure (I) 3:1 presents a summary of experimental campaign and illustrates different test configurations. 3.1.2 Production of specimens and main steps of experimental procedure CLT used to build the test specimens was entirely produced in laboratory in order to obtain specimens free of significant knots and with similar density distribution between MC (moisture content) groups and test configurations. To avoid significant knots, CLT panels with fixed dimensions of 600x400x102mm 3 and three similar layers with a thickness of 34mm each were produced. This way, each produced panel contained six test specimens of 170x170x102mm 3 . The configuration of CLT panels depended on the test configuration (Figure (I) 3:2). Density of timber boards was calculated by equation (3:1), according to ISO 3131:1975, and a random distribution was realized between different configurations. Densities varied between 345kg/m 3 and 576kg/m 3 (CoV=0,09 and mean=462kg/m 3 ) (Table (I) 3:1). In order to save time and avoid unexpected problems, the production of panels was planned in advance following pre-defined steps: 1. Cut (avoiding significant knots), plane and rectify four different types of boards needed to produce different layer configurations (Figure (I) 3:2 (a)); 2. Weight all boards and realize a similar density distribution between all configuration groups and a good distribution, close to a normal one, was obtained (Figure (I) 3:3); 3. Glue and press CLT panels (Figure (I) 3:4 (a) and (b)); 4. Saw the gaps in groups which include it (Figure (I) 3:4 (c) and (f)); 5. Glue and press last layer in configurations with GAP4 in middle layers (Figure (I) 3:4 (d) and (e)). Relatively to production of CLT, it is also important to refer that the timber used was spruce ( Picea abies ), with a nominal strength class C24 according to EN 338 (2009), and different layers were glued with P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 128 PURBOND® HB110 applied by a MINDA gluing equipment (Figure (I) 3:4 (a)). Furthermore, all CLT panels were pressed by a hydraulic pressing LANGZAUNER for 3 hours with a constant pressure of 0.4N/mm 2 (Figure (I) 3:4 (b) and (e)). Figure (I) 3:1. Different test configurations and sampling used for different groups and test days (dimensions in mm). 90 TESTS MC = 8% 20ºC / 29%RH MC =12% 20ºC / 65%RH MC=18% 20ºC / 90%RH 83 83 83 83 102102 85 85 102 102 102 102102 102 83 83 170 170 8585 GAP0_FL GAP0_ML GAP0_OL GAP0_3L GAP4_FL GAP4_ML GAP4_OL GAP4_3L 8383 170 83 83 170 170 170 170 8585 85 85 NO GAPSGAP=0MMGAP=4MM 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 30 10 10 10 10 10 10 10 10 10 10 10 30 30 30 30 30 30 30 30 90 TESTS 90 TESTS REF 102 170 170 SAMPLING Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 129 Type A Type B Type C Type D (a) REF GAP0_FL GAP0_ML GAP0_OL GAP0_3L GAP4_FL GAP4_ML GAP4_OL GAP4_3L (b) Figure (I) 3:2. CLT panels produced at Laboratory. (a) Four different types of boards needed to produce CLT panels for different test configurations; (b) different configurations of CLT panels.  = • È (3:1)  Density of each test piece, in grams per cm 3 • Mass of test piece, in grams È Volume of test piece, in cm 3 Table (I) 3:1. Descriptive statistics of density distribution for all nine different CLT panels. CONF . Nº OF BOARDS MEAN SD COV MINIMUM MEDIAN MAXIMUM P 5 REF 42 473 51,7 0,11 345 473 576 397 GAP0_FL 40 457 41,3 0,09 354 463 532 377 GAP0_ML 40 456 34,5 0,08 378 457 520 395 GAP0_OL 45 465 43,2 0,09 367 466 559 394 GAP0_3L 50 456 44,0 0,10 351 459 558 367 GAP4_FL 40 462 37,4 0,08 390 466 542 395 GAP4_ML 40 462 36,9 0,08 390 462 576 405 GAP4_OL 45 469 41,5 0,09 397 468 560 401 GAP4_3L 50 462 41,1 0,09 369 463 560 396 SD – standard deviation; CoV – coefficient of variation; P5 – 5 th percentile 620 204 34 420 204 34 100 34 620 100 34 420 400 600 102 400 600 102 400 600 102 400 600 102 400 600 102 4mm 400 600 102 4mm 400 600 102 4mm 400 600 102 400 600 102 4mm P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 130 REF GAP0_FL GAP0_ML GAP0_OL GAP0_3L GAP4_FL GAP4_ML GAP4_OL GAP4_3L Figure (I) 3:3. Similar density distribution between different configurations. Graphs were plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). After CLT production, 6 test specimens were cut from each CLT panel (Figure (I) 3:5 (a)). Small timber samples were saved and dried in order to quantify the timber moisture content at that time. After weighted and oven dried, those samples indicated a moisture content of approximately 10%. The goal was to produce CLT with stabilized timber with 12% of moisture content however, due to timing issues, it was not possible to conditioning timber boards for a longer period. So, in order to achieve a moisture content of approximately 12% at screwing time, specimens were conditioned again, in a climatic chamber with environmental conditions of 20ºC and 65% RH, for more 10 days. After that, all specimens were predrilled with a hole of 5mm adequate for screws with a diameter of 8mm (EN 1995-1-1:2004) (Figure (I) 3:5 (b)). The decision of pre-drilling test specimens is justified by the difficulty to ensure the correct insertion of STS through CLT gaps. 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m3] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m3] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 131 (f) (e) (d) (c) (b) (a) Figure (I) 3:4. Procedure to build CLT panels in laboratory. (a) assembling line and glue used to bond layers (PURBOND® HB110); (b) hydraulic pressing machine (LANGZAUNER) bonding two CLT layers; (c) gaps with four millimeters protected to the next bonding procedure; (d) glue line (MINDA equipment) with glue dispenser and timber boards with glue before assembly panel; (e) pressing the third layer; (f) final CLT panels before saw the external gaps. (Note: dimensions are in mm) 400 600 400 600 400 600 400 600 PRESS 3 HOURS PRESS 3 HOURS SECOND LAYER PRESS SAW GAP4 THIRD LAYER PRESS SAW GAP4 102102102 68 4mm P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 132 Still before these 10 days of conditioning, all specimens were again weighted and measured in order to perform another random density distribution between three defined moisture groups. Density distribution exhibited again good results with a density distribution close to a normal one, with median and mean densities close to each other, and with low CoV values between 0,07 and 0,12 (Figure (I) 3:6 and Table (I) 3:2). After ten days of conditioning, specimens were screwed. Moisture content did not reach exactly 12%, but it was close, ≈11%. STSs used were full threaded Rapid® Vollgewinde from Schmid with a diameter of 8mm and length of 180mm (Figure (I) 3:5 (b)). It is important to refer that in order to avoid tip influence in test results, STSs were inserted through the specimen thickness until the tip of STS trespass the entire thickness of specimen (Figure (I) 3:5 (c)). (c) (b) (a) Figure (I) 3:5. Procedure to finish test specimens. (a) test specimens cut from CLT panels; (b) predrilling equipment; (c) STS used to perform withdrawal tests (Rapid® Vollgewinde from Schmid with diameter of 8mm and length of 180mm) and the penetration of STS through entire specimen thickness until avoids the tip effect. 400 600 CUT TEST SPECIMENS PRE DRILLING INSERT SCREW 102 170 170 102 170 170 102 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 133 8% 12% 18% Figure (I) 3:6. Similar density distribution for different MC groups. Graphs were plotted based on Normal distribution and Renard score method - (i – 0,3) / (n + 0,4). Table (I) 3:2. Descriptive statistics of density distribution (ρ) for all nine different configurations between three MC groups. CONF . MC GROUP Nº OF SPECIMENS MEAN SD COV MIN MEDIAN MAX P 5 REF 8% 10 468 57,1 0,12 370 471 549 370 12% 10 472 55,9 0,12 376 471 553 376 18% 10 475 53,7 0,11 388 474 555 388 GAP0_FL 8% 10 453 40,1 0,09 388 459 509 388 12% 10 455 39,2 0,09 393 460 511 393 18% 10 460 40,0 0,09 394 463 519 394 GAP0_ML 8% 10 453 35,8 0,08 393 455 499 393 12% 10 456 35,1 0,08 396 459 500 396 18% 10 459 34,3 0,07 404 463 502 404 GAP0_OL 8% 10 457 41,3 0,09 386 461 514 386 12% 10 461 41,7 0,09 396 463 523 396 18% 10 465 42,3 0,09 401 466 535 401 GAP0_3L 8% 10 453 40,1 0,09 386 460 502 386 12% 10 455 39,2 0,09 392 461 505 392 18% 10 458 40,2 0,09 393 463 509 393 GAP4_FL 8% 10 453 35,8 0,08 394 457 504 394 12% 10 457 36,4 0,08 398 461 510 398 18% 10 461 36,0 0,08 405 465 515 405 GAP4_ML 8% 10 457 35,5 0,08 401 462 511 401 12% 10 461 34,6 0,08 405 466 512 405 18% 10 464 36,1 0,08 408 467 523 408 GAP4_OL 8% 10 462 37,4 0,08 407 466 519 407 12% 10 466 39,4 0,08 409 468 530 409 18% 10 469 38,9 0,08 412 468 533 412 GAP4_3L 8% 10 458 37,7 0,08 400 459 511 400 12% 10 459 38,4 0,08 401 460 512 401 18% 10 461 38,7 0,08 402 463 515 402 SD – standard deviation; CoV – coefficient of variation; P5 – 5 th percentile 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line 300 350 400 450 500 550 600 300 350 400 450 500 550 600 Expected Normal Value  [kg/m 3 ] Expected Value Reference Line P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 134 Finished the production of specimens, it was time to conditioning three groups of specimens in different environmental conditions in order to stabilize CLT with the pre-defined MC levels, namely: 8%, 12% and 18% (Figure (I) 3:1). The first approach to choose the right environmental conditions was based on suggestions of Hartl & Ramberger (1985) (Figure (I) 3:7). However, final decision was grounded on some recent data obtained by Wallner (2012) during his deep research on the evaluation of moisture induced stresses in glulam. Considering a fixed temperature (T) of 20ºC, only relative humidity (RH) levels were changed. Regarding a moisture content of 12%, both authors suggest the same level of relative humidity – 65%. But, based on his recent research, Wallner (2012) suggest different relative humidity levels to reach moisture contents of 8% and 18% (for a temperature of 20ºC): respectively 29% RH and 90% RH. So, three groups of specimens with similar configurations and densities, were conditioned in three different environmental conditions, namely: 20ºC and 29% RH to reach MC=8%; 20ºC and 65% RH to reach MC=12%; and 20ºC and 90% RH to reach MC=18% (Figure (I) 3:1). As the moisture content at screwing time was not 12%, specimens from group that should reach MC=12% were conditioned again for more ten days before being tested. Specimens from group with MC=8% were conditioned in a climatic chamber for a period of twenty days, while specimens from group with MC=18% were conditioned in a climatic room for a period of forty-five days. It is important to refer that, to predict the MC levels of conditioned specimens, a group of control specimens was produced. Those specimens were oven-dried, according to ISO 3130:1975, and their dried weight was used to know when specimens conditioned in different environments reach the expected MC level. Figure (I) 3:7. Relation between temperature and relative humidity for reaching different moisture contents for spruce, suggested by Hartl & Ramberger (1985). Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 135 3.1.3 Test procedure Axial withdrawal test procedure for screws is simple and it is clearly normalized at EN 1382:1999. Figure (I) 2:8 and Figure (I) 3:9 depict the test layout used to perform these tests and normally used to perform all withdrawal tests at Institut für Holzbau und Holztechnologie - Graz Technical University (Austria). Figure (I) 3:8. 2D drawing of test setup. F Front View Lateral View Top View F 50 470 Specimen Load Cell device to hold screw head Steel rods Screw Steel beam Specimen Load Cell Steel rods Steel beam 50 170 Steel plate device to hold screw head Screw Steel plate Screw head Hole in Steel plate Steel plate Steel plate Steel rods 102 180 470 170 129 137 129 170 129137129 270 170 270 102 180 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 142 Figure (I) 3:15. Mean test curves for all configurations with 12% of moisture content, expressing the expected yielding failure mode. 3.2.2 Test results for MC=8% Concerning results obtained for specimens tested with MC=8%, the real MC levels vary between 7,7% and 8,9% and present a good distribution (Figure (I) 3:16). However, due to low levels of MC reached, some damages in CLT specimens were observed after conditioning period in an environment of 20ºC and 29% RH. Damages in specimens were detected by direct visual inspection and are listen bellow: - Two specimens (8_GAP0_ML_8% and 9_GAP0_8%) exhibited large cracks through the STS location, in first layer, however it seems to not influence obtained  VR,‚[t‚ (Figure (I) 3:17 (a)); - Some specimens exhibited cracks through STS location in bottom layer (Figure (I) 3:17 (b)); - Specimens of groups with GAP0/GAP4 in one or both outer layers (GAP_FL, GAP_OL and GAP_3L) exhibited an enlargement of gap width of approximately 1mm (Figure (I) 3:17 (c)); - Some specimens present some cracks parallel to the grain in middle layer, for configurations that do not consider gaps in the same layer (Figure (I) 3:17 (d)); - A reduced delamination between layers was verified only in few cases (Figure (I) 3:17 (e)). Furthermore, as a result of some test problems, sampling of group GAP4_ML_8% was reduced from ten to eight specimens. Table (I) 3:4 presents descriptive statistics of results obtained by the group of specimens that intended to reach MC=8%, concerning:  VR,‚[t‚ ,  NO and MC. Similarly to group with MC=12%, obtained results suggest a good distribution and absence of outliers. The highest obtained mean value for  VR,‚[t‚ was 7,34 N/mm 2 for configuration GAP0_3L_8% and lowest was 4,10N/mm 2 for configuration GAP4_3L_8%. 0 1 2 3 4 5 6 7 0 2 4 6 8 10 f ax [N/mm 2 ] Displacement [mm] REF_12% GAP0_FL_12% GAP0_ML_12% GAP0_OL_12% GAP0_3L_12% GAP4_FL_12% GAP4_ML_12% GAP4_OL_12% GAP4_3L_12% Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 143 Figure (I) 3:18 (a), the tendency to  VR,‚[t‚ increase with number of gaps remains. Also due to enlargement of gaps, GAP4_8% configurations present higher decreases for  VR,‚[t‚ than those observed in group with MC=12%. Again the configurations GAP4_OL_8% and GAP4_3L_8% exhibit the higher decreases (Figure (I) 3:18 (b)): 28,4% and 37,3%, respectively. Figure (I) 3:19 compares the results obtained by similar configurations with GAP0 and GAP4. Considering only gap width, the differences still increase with number of gaps: 19,3%, 14,6%, 30,4% and 44,1%, for configurations GAP_FL_8%, GAP_ML_8%, GAP_OL_8% and GAP_3L_8%, respectively. Due to lower  VR,‚[t‚ obtained for configurations with GAP4, these differences are larger than results obtained by group with MC=12%. Figure (I) 3:16. Moisture distribution for group with 8% of moisture content. (a) (b) (c) (d) (e) Figure (I) 3:17. Cracking resulted from conditioning period. (a) crack through the STS location; (b) opening of GAP0 on an exterior layer; (c) lateral view of a crack in an outer layer; (d) and (e) lateral view of delamination and crack in middle layer. 7,0 7,5 8,0 8,5 9,0 9,5 7,0 7,5 8,0 8,5 9,0 9,5 Expected Normal Value MC [%] Expected Value Reference Line P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 144 Table (I) 3:4. Mean values and descriptive statistics of  VR,‚[t‚ ,  NO and ./ of different test configurations with 8% of moisture content. MEAN VALUES DESCRIPTIVE STATISTICS GROUPS  (%)  ,B2@B ( N / MM 2 )   ( KG / M 3 ) NUMBER OF TESTS   , B2@B    SD COV MIN MAX MEDIAN P 5 REF_8% 8,4 6,54 472 10 0,68 0,10 5,47 7,96 6,52 5,47 55,70 0,12 376 553 472 376 0,35 0,04 7,9 8,9 8,6 7,9 GAP0_FL_8% 8,0 6,69 458 10 0,73 0,11 5,75 8,12 6,73 5,75 39,00 0,09 395 514 463 395 0,29 0,03 7,9 8,7 8,3 7,9 GAP0_ML_8% 8,1 6,29 457 10 0,31 0,05 5,91 6,82 6,23 5,91 35,30 0,08 398 504 460 398 0,28 0,03 7,9 8,6 8,4 7,9 GAP0_OL_8% 8,0 6,72 461 10 0,43 0,06 5,92 7,26 6,76 5,92 40,40 0,09 393 517 464 393 0,33 0,04 7,7 8,8 8,2 7,7 GAP0_3L_8% 8,1 7,34 458 10 0,46 0,06 6,42 7,77 7,54 6,42 39,50 0,09 391 505 466 391 0,24 0,03 8,0 8,7 8,4 8,0 GAP4_FL_8% 8,1 5,4 457 10 0,74 0,14 3,93 6,62 5,28 3,93 34,10 0,07 401 505 461 401 0,19 0,02 8,0 8,7 8,4 8,0 GAP4_ML_8% 7,9 5,37 455 8 0,38 0,07 4,72 5,84 5,41 4,72 36,20 0,08 405 516 455 405 0,33 0,04 7,7 8,7 8,4 7,7 GAP4_OL_8% 8,0 4,68 467 10 0,34 0,07 4,25 5,32 4,64 4,25 36,50 0,08 412 522 469 412 0,19 0,02 7,9 8,5 8,1 7,9 GAP4_3L_8% 7,8 4,10 462 10 0,45 0,11 3,25 4,66 4,2 3,25 36,30 0,08 407 513 463 407 0,19 0,02 8,0 8,7 8,2 8,0  – moisture content,   - maximum withdrawal resistance,   – density, SD – standard of deviation, CoV – coefficient of variation, P5 – 5 th percentil, REF_8% - configuration without gaps tested with 8% of moisture content, GAP0_FL_8% - configuration with a gap of 0mm in first layer and tested with 8% of moisture content; GAP0_ML_8% - configuration with a gap of 0mm in middle layer and tested with 8% of moisture content, GAP0_OL_8% - configuration with a gap of 0mm in outer layers and tested with 8% of moisture content, GAP0_3L_8% - configuration with a gap of 0mm in three layers and tested with 8% of moisture content, GAP4_FL_8% - configuration with a gap of 4mm in first layer and tested with 8% of moisture content; GAP4_ML_8% - configuration with a gap of 4mm in middle layer and tested with 8% of moisture content, GAP4_OL_8% - configuration with a gap of 4mm in outer layers and tested with 8% of moisture content, GAP4_3L_8% - configuration with a gap of 4mm in three layers and tested with 8% of moisture content. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 145 Figure (I) 3:20 depicts the visual failures of tested specimens with MC=8%, which are more pronounced than observed failures for specimens with MC=12%. In bottom part, tip of STS crushes timber, while on top, timber breaks and comes out inside free area around the STS provided by test setup. However, the failure mode remains the same: Figure (I) 3:21 shows the yielding failure mode, through the mean curves obtained by relation between  VR,‚[t‚ and STS displacement. More detailed information about test results of specimens with MC=8% can be seen in Annex 1:11. Analyzing  VR,‚[t‚ values obtained in 8% of moisture content group, it was verified that due to reduced moisture content and consequent small enlargement of gaps, the discrepancy between REF_8% and GAP0_8% configurations became reduced. The higher discrepancy is of 12,23% for configuration GAP0_3L_8%. However, as expected and shown in Figure (I) 3:18, the tendency to  VR,‚[t‚ increase with number of gaps remains. Relatively to GAP4_8% configurations, the withdrawal capacity decrease more than in group with 12% of moisture content also due to gap enlargement. Again the configurations of GAP4_OL_8% and GAP4_3L_8% exhibit the higher discrepancies (Figure (I) 3:18): 28,44% and 37,31%, respectively. Figure (I) 3:19 compares the results obtained by similar configurations of GAP0 and GAP4. Considering only gap width, the differences still increase with number of gaps: 19,28%, 14,63%, 30,36% and 44,14%, for configurations GAP_FL_8%, GAP_ML_8%, GAP_OL_8% and GAP_3L_8%, respectively. These differences are larger than results obtained by group with 12% of moisture content. (a) (b) Figure (I) 3:18. Box chart with notched boxes of  VR , ‚[t‚ and scatter plot of density of reference and moisture content for the following groups: (a) REF_8% and GAP0_8%; (b) REF_8% and GAP4_8%. 7,0 7,5 8,0 8,5 9,0 9,5 10,0 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 MC [ % ] MC  12 REF GAP0 FL ML OL 3L 7,0 7,5 8,0 8,5 9,0 9,5 10,0 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 MC [ % ] MC  12 REF GAP4 FL ML OL 3L P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 146 (a) (b) (c) (d) Figure (I) 3:19. Box chart with notched boxes of  VR , ‚[t‚ and scatter plot of density of reference and moisture content for the following groups: (a) REF_8% and GAP_FL_8%; (b) REF_8% and GAP_ML_8%; (c) REF_8% and GAP_OL_8%; and (d) REF_8% and GAP_3L_8%. (a) (b) Figure (I) 3:20. Visual failures of tests performed with 8% of moisture content. (a) Visual failure for configurations REF, GAP0 and GAP4_ML; (b) Visual failure for remaining configurations GAP4. REF GAP0_FL GAP4_FL 7,0 7,5 8,0 8,5 9,0 9,5 10,0 3 4 5 6 7 8 9 fax,test [N/mm2 ] 300 350 400 450 500 550 600 [ Kg/m3]  12 MC [ % ] MC  12 REF GAP0_ML GAP4_ML 7,0 7,5 8,0 8,5 9,0 9,5 10,0 3 4 5 6 7 8 9 fax,test [N/mm 2] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 MC [ % ] MC  12 REF GAP0_OL GAP4_OL 7,0 7,5 8,0 8,5 9,0 9,5 10,0 3 4 5 6 7 8 9 fax,test [N/mm 2] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 MC [ % ] MC  12 REF GAP0_3L GAP4_3L 7,0 7,5 8,0 8,5 9,0 9,5 10,0 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 MC [ % ] MC  12 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 147 Figure (I) 3:21. Mean test curves for all configurations with 8% of moisture content, expressing the expected yielding failure mode. 3.2.3 Test results for MC=18% Considering the group of specimens in which a MC=18% was intended, the range of moisture content values vary between 16,5% and 18,7%, and similarly to both previous groups, obtained values exhibit a good distribution (Figure (I) 3:22). Furthermore, similarly to what was observed to low levels of MC (MC=8%), high levels of MC caused some damages on CLT specimens, namely: ▪ Wood swelling caused the deformation of CLT specimens, which exhibit concave or convex side faces depending on dominant timber direction (Figure (I) 3:23 (a)); ▪ GAP4 close significantly, crushing against the thread of STSs (Figure (I) 3:23 (b)). Table (I) 3:5 presents descriptive statistics of  VR,‚[t‚ ,  NO and ./, for all test configurations with MC=18%. Withdrawal resistance varies in a range between 3,77 N/mm 2 and 7,76 N/mm 2 and, similarly to both moisture groups described before, configurations GAP0 present the highest values (Figure (I) 3:24 (a)) and configurations GAP4 present the lowest values (Figure (I) 3:24 (b)). Both follow the same tendency to increase/decrease  VR,‚[t‚ with the increase of number of gaps. However, differently to what was observed for remaining MC groups, the decrease of  VR,‚[t‚ registered for GAP4 configurations is not so significant. This change is the result of timber swelling and consequent closing of GAP4. The difference between mean  VR,‚[t‚ of configurations REF_18% and GAP4_3L_18% is 18,3%. It is important to remind that for the group with MC=8% the difference between these configurations was of 37,3%. Following the same tendency, similar configurations with GAP0 and GAP4 also present lower differences between them. Figure (I) 3:25 depicts exactly this fact which can be expressed by following percentages: 15,1%, 11,1%, 15,0% and 23,5%. Relatively to visual damages after testing, they were quite similar to those observed for group tested with MC=8%. Due to higher moisture content, crushing close to the tip 0 1 2 3 4 5 6 7 0 2 4 6 8 10 f ax [N/mm 2 ] Displacement [mm] REF_8% GAP0_FL_8% GAP0_ML_8% GAP0_OL_8% GAP0_3L_8% GAP4_FL_8% GAP4_ML_8% GAP4_OL_8% GAP4_3L_8% P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 148 of STS, caused by pulling out action, was a bit more evident, but on the top of specimen timber fractured less (Figure (I) 3:26). Figure (I) 3:27 depicts the mean curves representative of relation between  VR,‚[t‚ and displacement (mm). As expected, and similarly to the remaining MC groups, curves depict a yielding failure mode. More detailed information about test results for configurations with MC=18% can be seen in Annex 1:12. Figure (I) 3:22. Moisture distribution for group with 18% of moisture content. (a) (b) Figure (I) 3:23. Damages caused by high MC levels. (a) deformations on CLT shape caused by timber swelling depending on grain direction; (b) closing of GAP4 after conditioning period. (a) (b) Figure (I) 3:24. Box chart with notched boxes of  VR , ‚[t‚ and scatter plot of density of reference and moisture content for the following groups: (a) REF_18% and GAP0_18%; (b) REF_18% and GAP4_18%. 16,0 16,5 17,0 17,5 18,0 18,5 19,0 16,0 16,5 17,0 17,5 18,0 18,5 19,0 Expected Normal Value MC [%] Expected Value Reference Line 16,0 16,5 17,0 17,5 18,0 18,5 19,0 MC  12 MC [ % ] 3 4 5 6 7 8 9 fax,test [N/mm2 ] 300 350 400 450 500 550 600 [ Kg/m3]  12 REF GAP0 FL ML OL 3L 16,0 16,5 17,0 17,5 18,0 18,5 19,0 MC  12 MC [ % ] 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 REF GAP4 FL ML OL 3L Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 149 (a) (b) (c) (d) Figure (I) 3:25. Box chart with notched boxes of  VR , ‚[t‚ and scatter plot of density of reference and moisture content for the following groups: (a) REF_18% and GAP_FL_18%; (b) REF_18% and GAP_ML_18%; (c) REF_18% and GAP_OL_18%; (d) REF_18% and GAP_3L_18%; and (e) REF_18% and GL_18%. (a) (b) Figure (I) 3:26. Visual STS failures of tests performed with 18% of moisture content. (a) Visual failure on top of specimen; (b) Visual failure on bottom of specimen. REF GAP0_FL GAP4_FL 16,0 16,5 17,0 17,5 18,0 18,5 19,0 MC  12 MC [ % ] 3 4 5 6 7 8 9 fax,test [N/mm2 ] 300 350 400 450 500 550 600 [ Kg/m3]  12 REF GAP0_ML GAP4_ML 16,0 16,5 17,0 17,5 18,0 18,5 19,0 MC  12 MC [ % ] 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 REF GAP0_OL GAP4_OL 16,0 16,5 17,0 17,5 18,0 18,5 19,0 MC 12 MC [ % ] 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 REF GAP0_3L GAP4_3L 16,0 16,5 17,0 17,5 18,0 18,5 19,0 MC  12 MC [ % ] 3 4 5 6 7 8 9 fax,test [N/mm 2 ] 300 350 400 450 500 550 600 [ Kg/m 3 ]  12 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 150 Table (I) 3:5. Mean values and descriptive statistics of  VR,‚[t‚ ,  NO and ./ for different test configurations with 18% of moisture content. MEAN VALUES DESCRIPTIVE STATISTICS GROUPS  (%)  ,B2@B ( N / MM 2 )   ( KG / M 3 ) NUMBER OF TESTS   , B2@B    SD COV MIN MAX MEDIAN P 5 REF_18% 17,5 5,86 479 10 0,53 0,09 5,35 7,12 5,68 5,35 52,90 0,11 394 560 475 394 0,43 0,02 17,4 18,7 17,8 17,4 GAP0_FL_18% 17,0 6,24 465 10 0,70 0,11 5,13 7,76 6,2 5,13 39,80 0,09 400 525 467 400 0,45 0,03 16,7 17,9 17,4 16,7 GAP0_ML_18% 17,0 6,03 463 10 0,45 0,07 5,4 6,61 6,15 5,4 33,80 0,07 410 507 467 410 0,44 0,03 16,5 17,9 17,2 16,5 GAP0_OL_18% 17,4 6,25 469 10 0,52 0,08 5,26 6,93 6,32 5,26 40,70 0,09 406 537 469 406 0,36 0,02 17,2 18,2 17,6 17,2 GAP0_3L_18% 17,3 6,26 463 10 0,33 0,05 5,65 6,66 6,37 5,65 39,00 0,08 399 512 468 399 0,40 0,02 17,1 18,4 17,5 17,1 GAP4_FL_18% 17,4 5,3 465 10 0,55 0,10 4,5 6,05 5,28 4,5 34,20 0,07 411 516 469 411 0,51 0,03 17,1 18,6 17,4 17,1 GAP4_ML_18% 17,3 5,36 469 10 0,52 0,10 4,67 6,21 5,28 4,67 36,10 0,08 414 529 471 414 0,45 0,03 16,8 18,3 17,6 16,8 GAP4_OL_18% 17,4 5,31 473 10 0,68 0,13 4,34 6,24 5,4 4,34 38,20 0,08 418 536 472 418 0,32 0,02 17,1 18,2 17,6 17,1 GAP4_3L_18% 17,4 4,79 466 10 0,55 0,12 3,77 5,37 5 3,77 37,30 0,08 408 515 468 408 0,25 0,01 17,4 18,2 18,0 17,4  – moisture content,   , B2@B - maximum withdrawal resistance obtained with test,   – density, SD – standard of deviation, CoV – coefficient of variation, P5 – 5 th percentil, REF_18% - configuration without gaps and tested with 18% of moisture content, GAP0_FL_18% - configuration with a gap of 0mm in first layer and tested with 18% of moisture content; GAP0_ML_18% - configuration with a gap of 0mm in middle layer and tested with 18% of moisture content, GAP0_OL_18% - configuration with a gap of 0mm in outer layers and tested with 18% of moisture content, GAP0_3L_18% - configuration with a gap of 0mm in three layers and tested with 18% of moisture content, GAP4_FL_18% - configuration with a gap of 4mm in first layer and tested with 18% of moisture content; GAP4_ML_18% - configuration with a gap of 4mm in middle layer and tested with 18% of moisture content, GAP4_OL_18% - configuration with a gap of 4mm in outer layers and tested with 18% of moisture content, GAP4_3L_18% - configuration with a gap of 4mm in three layers and tested with 18% of moisture content. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 151 Figure (I) 3:27. Mean test curves for all configurations with 8% of moisture content, expressing a yielding failure mode. 3.3 Modeling the influence of number of gaps (Ƞ  ) and MC levels (Ƞ  ) on withdrawal resistance of STSs inserted in main face of CLT panels 3.3.1 Correction of density of reference ( , ) and withdrawal resistance ( , ) As observed in previous points, against the expectations, GAP0 configurations exhibited higher  VR,‚[t‚ than REF configurations for all moisture groups, always depending on number of gaps. Those results incited the suspicion that the density at GAP location could be higher depending on the way the boards were placed during CLT production. Annual growing rings could be closer to each other in the location where STS was inserted (Figure (I) 3:28). Considering this suspicion, density was corrected by measure and weight a smaller volume of timber around the STS (Figure (I) 3:29). However, corrected density ( NO,‰Ývv ) present a bit lower values (Figure (I) 3:30) than  NO , a difference that was not considered significant. Table (I) 3:6 presents density values, before and after correction, respective CoV and the ratio between both for all configurations tested with different MC levels. Considering values obtained for  NO,‰Ývv ,  VR was also corrected ( VR,‰Ývv ) according to equation (3:6), suggested by CUAP 06.03/08 (2010). Despite this formula has been conceived to be used only in case of high differences between densities of specimens, it was decided to apply it in order to perform some comparisons with results obtained for the experimental campaigns presented in next two chapters. Table (I) 3:7 shows the descriptive statistics of  VR,‰Ývv , which are the values used to perform the analysis presented hereafter. 0 1 2 3 4 5 6 7 0 2 4 6 8 10 f ax [N/mm2] Displacement [mm] REF_18% GAP0_FL_18% GAP0_ML_18% GAP0_OL_18% GAP0_3L_18% GAP4_FL_18% GAP4_ML_18% GAP4_OL_18% GAP4_3L_18% P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 254 (a) (b) Figure (II) 1:8. Stadhaus, 24 Murray Grove, London – UK. (a) external view of building; (b) excessive compartmentalization of structural solution. (a) (b) Figure (II) 1:9. Bridport building, Hackney London, UK. (a) External view; (b) CLT load bearing structure. (a) (b) Figure (II) 1:10. Forté, in Melbourne’s Docklands, Australia. (a) 3D external rendering; (b) Picture of works on site. (a) (b) Figure (II) 1:11. Via Cenni, Milan, Italy. (a) external 3D rendering (b) vertical cross section. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 255 Constructed in 2014, Wood Innovation and Design Center is a six-storey office building designed by MG architecture (Figure (II) 1:15). With the exception of a concrete raft slab that shapes the building foundation, it is entirely built with mass timber products: glulam columns and beams; CLT walls and stairwells/lift cores and CLT innovative staggered CLT panel design for floors. Brock Commons is a seventeen-storey building, developed by Acton Ostry architects for the University of British Columbia (Figure (II) 1:16). The building rises from a concrete podium and its structural system comprises two concrete cores, a grid of glulam columns connected between floors by means of special metal connectors and CLT floors acting as two-ways diaphragm. Vertical loads are carried by the timber structure while concrete cores provide lateral stability. Some other hybrid systems were developed and published but not applied until today. Barents house, project by Reiulf Ramstad Architects (Reid, 2010), and FFTT system (find the forests through the trees), developed by Michael Green and Eric Karsh (Green & Eric Karsh, 2012), are two of these proposals. Both systems look for solutions able to offer greater spatial amplitude, especially suitable to non-residential uses and reach higher heights. Barents house is a project for a timber building with 20 stories which bet in a structural system that combines CLT floors, glulam beams, columns and diagonals, and a concrete core (Figure (II) 1:17). FFTT system is able to reach 30 stories and bets on the combination of CLT walls and floors with steel beams (Figure (II) 1:18). The reduction of timber amount, and consequent reduction of space partition, is possible due to the increase in the building stiffness warranted by concrete and steel elements (C. Silva et al., 2012). (a) (b) Figure (II) 1:12. Limnologen project, Växjö, Sweden. (a) External view of two of four buildings; (b) floor elements (Serrano, 2009). (a) (b) Figure (II) 1:13. LifeCycle Tower ONE. (a) 3D digital external rendering of the LCT ONE in daylight; (b) Hybrid structural system. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 256 (a) (b) Figure (II) 1:14. Wagramerstrasse timber building. (a) 3D external rendering; (b) Picture of works on site. (a) (b) Figure (II) 1:15. Wood Innovation Design Center. (a) External view of building; (b) picture of inner space. (a) (b) Figure (II) 1:16. Brock Commons. (a) External view of building; (b) hybrid structural system. (a) (b) Figure (II) 1:17. Barents House Project. (a) 3D external rendering; (b) 3D structural rendering. (a) (b) Figure (II) 1:18. FFTT system (a) 30 storey timber building model proposed by mgb Architecture + Design; (b) Hybrid timber-steel solution. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 257 CHAPTER 2 (II) 2 URBAN TIMBER SYSTEM: A TIMBER HYBRID SOLUTION P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 258 2.1 Introduction Based on the belief that timber is the construction material of the future, and considering all the points discussed in previous chapter, the Urban Timber system (UT system) has been developed. This system got inspiration in some constructed buildings and was designed attending to some essential principals: - Explore the possibilities of timber as structural material promoting the rebirth of timber architecture as natural part of dense urban centers; - Respect economic, social and environmental sustainability; - Evaluate the advantages and disadvantages of timber based structures; - Assuring the flexibility of the construction system, regarding the building shape, opening sizes and inner design; - Ensure all safety building requirements; - Explore the combination of different wood-based materials, preferring those that use low quality timber to shape advanced building products; - Provide a fast and simple construction system; - Offer possibilities for the building end of life; - Consider the effects of moisture induced effects on timber based structural elements. In order to provide a system able to shape flexible buildings, UT system was designed based on a timber hybrid construction concept that combines CLT and glulam as main structural materials. This combination of panel-shape and linear-shape elements is the key point regarding the architectural flexibility, widening the range of possibilities for architects. UT system denies the selfish “all with CLT” concept, resulting in a lighter solution able to conceive more economical and challenging buildings. However, it is important to mention that in similarity to structures built just with CLT, UT system also chooses to work with materials that use low quality timber to shape advanced building products, enabling long-lasting carbon storage in a new assortment of wood. Advantages associated with prefabrication were also exploited either regarding practical issues like reduction of construction time and costs, or regarding the possibilities offered by CNC technology. As it will be described ahead in the present chapter, UT system was detailed considering safety requirements, such as structural behavior or fire safety, comfort defined by acoustic and thermal behavior, and moisture protection. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 259 Regarding building safety, structural behavior of UT system was based on the bundled tube concept, often used on the design of multi-storey buildings. Dias (2017) developed a structural evaluation of UT system, designing a 3D model in RFEM-5 software , a finite element software by Dlubal. The author conclude that UT system is able to shape large spans (8,8m) and consequently allows the proclaimed spatial flexibility. Fire safety, acoustic and thermal comfort and moisture protection were evaluated based on published researches in which floor and wall sections are suggested. The protection of timber structural elements from moisture is a crucial issue. Despite being a subject with reduced published research, the moisture effects on timber elements is of huge importance. This way, the experimental campaign presented in Part I of the present thesis was of great significance during the definition of the UT system. In the present chapter, UT system is described and evaluated considering its structural concept and behavior, architectural considerations, production and construction processes, end of life possibilities and its detailing. Specificities related with moisture effects will be discussed in detail in next chapter. 2.2 Description of Urban Timber system (UT system) and its structural elements Inspired in the bundled tube concept, the UT system works like a cluster of individual tubes connected together in order to make them behave as a single unit. As result, the strength and stiffness of the cross frames are improved (Ali & Moon 2007). This concept is often applied in order to achieve a greater number of floors and get a greater freedom of spatial design. One of the greatest advantages of bundled tube concept is the free location of lift and stairs cores. The UT system is based in the combination of CLT and glulam, in which CLT shapes floors, walls and deep beams, while glulam it is used as double beams. CLT floors are directly supported by longitudinal and transversal double glulam beams. The double beams work together with CLT floors distributing the loads to the CLT walls, increasing the stiffness of the building and avoid the effect of progressive collapse. CLT walls have the function to resist all gravity loads, driving them to the foundations by means of vertical or oblique paths. At last, CLT deep beams sew up all the individual tubes in the building perimeter (Figure (II) 1:19). P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 260 Figure (II) 1:19. Simplified Scheme of the UT system identifying main timber structural elements. 2.2.1 Structural elements The UT system was developed based on some basic principles of structural systems. As an architect, the author of the present thesis does not have the necessary background to perform a structural design of UT system. However, collaborate in the development of a master thesis, Dias (2017), aimed to make a structural evaluation of the UT system. To develop such work, the author considered CLT made of timber of C24 strength class and glulam of GL24h strength class. Regarding CLT, as it is not yet standardized, material characterized from Stora Enso was considered. Cross sections of CLT floors were initially pre-designed by Calculatis , a simple design software developed by Stora Enso. This pre-design was done in a conservative way, once CLT floor slabs were designed as unidirectional elements. Remaining elements were designed through the elaboration of a 3D model in RFEM-5 software , a finite element software by Dlubal. In the following sections, CLT walls, CLT slabs, double glulam beams and CLT deep beams will be analysed regarding its structural behavior in a 10 storey building, such as considered by Dias (2017). 2.2.1.1 CLT walls One of the first critical reactions to the UT system, uses to be about CLT walls. Usually, the question is: Would not be better to use glulam columns instead of CLT walls? What size for CLT walls? What are the advantages of using CLT in independent small elements? Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 261 In fact, it would be possible to design the same scheme using glulam columns with smaller cross sections when comparing with CLT walls. However, in this case it was preferred to bet on CLT walls for two main reasons: one based on structural issues and the other based on aesthetic principals. Regarding the structural issues, tall buildings have greater vulnerability to lateral forces, especially wind loads, which means that the building perimeter has more structural significance (Ali & Moon 2007). CLT walls work like shear-plates positioned in building perimeter, oriented perpendicularly to the plane facade, providing more stiffness than the corresponding elements in a frame construction (Falk 2005). In respect to aesthetic reasons, the motivation to choose CLT walls instead of glulam columns is based in the possibility to expose CLT walls in the building. This way it is possible to expose the boards of external layers that looks like a beautiful wood cladding, instead of lamination lines that characterize glulam elements. The face of CLT walls that expose lamination lines can be hidden as depicted in Figure (II) 1:1, and final result would be perfect. Based on half miter technic, the connection between two elements, CLT wall and capping part, can be made by mechanical connections or by gluing technique. Figure (II) 1:20. Detail of how to hide lamination lines on CLT walls. Dias (2017) performed the structural evaluation of the UT system considering the plan shown in Figure (II) 1:21. In order to answer to some structural needs, some CLT walls with different dimensions are suggested for central core and lift cores. Table (I) 3:1 exhibits the dimensions of different wall elements considered for his evaluation. The large length of CLT elements is related with construction system, theme that will be discussed in section 0, ahead in present chapter. So, in order to answer to the question P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 262 related to the dimension needs, the work developed by Dias (2017) concluded that, considering a building with 10 storeys, general CLT walls needs to have a width of 1,5m and a thickness of 600mm (Table (I) 3:1). In a first impression, a thickness of 600mm may seem too much, however, considering a building that is 10 storey high it is not that inappropriate and its impact in interior space can be hidden by separation walls or minored by means of a well-planned open space concept. To achieve the proposed dimensions, the most important parameter to define was the buckling of the CLT walls. This parameter was successfully verified, however it imposed the need to perform experimental tests to determine the stiffness of connections between CLT walls and concrete base and between CLT walls and double glulam beams. Figure (II) 1 : 21 . Simplified Scheme of UT system identifying main timber structural elements. Dimensions in meters. 1 2 3 4 5 6 7 BCD E 8,8 8,0 8,0 8,8 A F G 5,62,1 3.9 2,15,6 3,9 CLT WALLS GLULAM BEAMS CLT DEEP BEAMS LEGENDA: Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 263 Table (II) 1:1. Dimensions of CLT walls considered by Dias (2017), depending on its location. LOCATION OF CLT WALLS DIMENSIONS LENGTH (m) WIDTH (m) THICKNESS (mm) GENERAL CLT WALLS 9,24 and 9,00 and 12,24 1,50 600 CENTRAL CORE PERIMETER 9,24 and 9,00 and 12,24 1,50 ≤ W ≤ 2,60 160 LIFT CORE 9,24 and 9,00 and 12,24 1,60 ≤ W ≤ 2,35 320 Despite the horizontal loads has been considered during the development of UT system, the connections between CLT walls and concrete foundations has to be evaluated and adapted. Generally, when it is considered a monolithic CLT construction, based on a platform system concept, the connection between CLT walls and concrete foundation are made by means of steel angles designed to resist tensile and shear actions. However, the UT system is based on the ballom frame concept and the number of CLT walls is much less. So, steel angles would not be able to resist internal stresses caused by lateral loads. Furthermore, due to reduced number of walls, this kind of connections would be susceptible to brittle failures and consequent collapse. To overcome this issue, Dias (2017) suggested to combine UT system with Pres-lam (Prestressed Laminated timber) technology, in order to provide ductility to the structural system. This combination will be described ahead in the present chapter in section dedicated to connections. 2.2.1.2 CLT slabs It is common to hear that one of the biggest advantages of CLT is its ability to shape slabs of solid timber which can distribute the loads in both orthogonal directions. However, in most cases, CLT slabs are calculated and applied as unidirectional structural elements. And why is that? Despite being considered a great advantage of CLT slabs, its bidirectional behavior is not easy to predict, involving a vast number of parameters, such as, support conditions, relative stiffness of panels in support zones, yielding moment of transversal and longitudinal layers, shear stress in both directions, etc. Depending on support conditionings, designing a bidirectional CLT slab can be considered: (1) when the slab is supported by punctual elements such as columns and (2) when the slab is supported over three or four edges. Gagnon and Pirvu (2011) proposed a method of calculation for CLT slabs that is based in the following assumption: P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 270 Figure (II) 1:24. Connection between double glulam beams and CLT walls by means of an external steel connector. Figure (II) 1:25 shows a connection in which double glulam beams are connected to CLT walls by means of an internal steel connector. The need to develop a hidden solution arises essentially due to fire and aesthetic concerns. Considering a scenario in which the connection is exposed and not protected by nonflammable materials, steel connectors are considered weaknesses of structural system. Once the steel connector is hidden inside the timber element it is protected from high temperatures that can be reached in case of fire. The concept of this connection is the same of the solution presented before, parallel glulam beams are tied together with CLT walls and steel connectors by means of long steel dowels. The main difference is on the shape of steel connector that instead of a U shape it has a T shape. Perpendicular glulam beams are fixed to the steel connector by means of steel dowels/rods. In this way, perpendicular glulam beams have to be prepared on-site in order to fit perfectly in the steel connector. Here, also both parallel and perpendicular glulam beams have a support on its underside. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 271 Figure (II) 1:25. Connection between double glulam beams and CLT walls by means of an internal steel connector. At last, Figure (II) 1:26 presents a solution inspired in the timber to timber connection designed by Shigeru Ban architects to Tamedia Building. A large hardwood connector is responsible to tie together CTL walls and both, parallel and perpendicular beams. A smaller hardwood element and some self tapping screws, are used to lock CLT walls and parallel glulam beams, while perpendicular beams are fixed to large hardwood connector by means of steel dowels/rods. Here, all the elements must reach the site with all the cuts necessary to assemble the connection. The idea to apply timber to timber connections in tall timber buildings is of great interest, especially regarding environmental issues. However, to much work and research is required before state any conclusions about its efficiency. According to Dias (2017), different possible solutions must be tested in order to quantify the stiffness of connections associated to Pres-Lam system, especially when connection are exposed to dynamic loads. According to the same author, one of the problems associated to this connection is the consequent deformation of CLT floors. This problem occurs when CLT panels are exposed to oscillation effects resulted from rotational constraint created by connection between CLT walls and glulam beams. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 272 Figure (II) 1:26. Connection between double glulam beams and CLT walls by means of an internal hardwood connector. 2.2.2.2 Rigid connections 2.2.2.2.1 Connection between the reinforced concrete foundation/wall and CLT walls As already mentioned, the need to apply Pres-Lam to the UT system, arise due to the weakness detected on the connection between the CLT wall and the reinforced concrete foundation. The contact area between both elements is too small to answer to the loads generated in a multi-storey building. Multistorey timber buildings are characterized by a high flexibility due to the high number of connections needed and also due to the mechanical properties of timber. In structural terms, shear design for the base of the building will impose larger sections when lateral loads are considered. Pres-Lam technology has been developed in New Zealand and it meant to be applied in large timber structures ((Palermo et al., 2006) (Smith et al., 2008) and (Iqbal et al., 2007)). During last decade, some Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 273 few commercial building has been built using this technique, namely: Nelson Marlborough Institute of Technology (Devereux et al., 2011) and Carterton Events Center (Palermo et al., 2012). Pres-Lam is constituted by high resistance pre-stress cables (Macalloy), that do not need to be continuous through all storeys, combined with located sinks (solid steel bars). Dias (2017) estimated the located stresses on foundation when the building is subjected to extreme wind loads. He found that the maximum tension is of 539kN/m. So, due to the reduced weight of the building it can collapse when submitted to high horizontal loads. This ways, considering a building with 10 storey building, it would be necessary to apply pre-stressed bars with a diameter of 25mm of Macalloy type, located in the middle of each CLT wall. Figure (II) 1:27 and Figure (II) 1:28 present two proposals for connection between CLT walls and the reinforced concrete foundation. Connection presented in Figure (II) 1:27 was proposed before the structural evaluation performed by Dias (2017). In that time, the connection was designed using a steel connector that rests directly in reinforced concrete foundation and is fixed by means of steel dowels/rods. The same steel connector goes inside the CLT walls by means of several steel plates. Necessary grooves on CLT wall, where the steel plates go inside, must be done previously. Steel dowels/rods would be responsible to fix steel connector to CLT walls. The solution present in Figure (II) 1:28 resulted from the evaluation performed by Dias (2017) and is obviously based on an already tested solution proposed by Palermo et al. , (2012). The solution is extremely simple and the main element is the Macalloy steel bars in which pre-stress is applied. The connection is composed by lateral dowel bars casted in concrete and grouted in CLT walls and by Macalloy steel bars and Macalloy bar couplers. Lateral dowel bars work like energy dissipation devices, while prestressed Macalloy steel bars are responsible to answer high loads accumulated on the base of the building. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 274 Figure (II) 1 : 27 . C onnection be tween reinforced concrete wall/foundation and CLT walls proposed before evaluation of the UT system. Figure (II) 1 : 28 . C onnection between reinforced concrete wall/foundation and CLT walls after Dias (2017) evaluation of the UT system. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 275 2.2.2.2.2 Connections between CLT floors/CLT walls and CLT deep beams CLT deep beams are strategic elements which sew up all the elements that compounds the structural system making it to work as a single unit. When a structure works like that, the stiffness of the building increases. According to Dias (2017), the connection between CLT floors/CLT walls and CLT deep beams must be rigid and can be performed by means of self-tapping screws inserted diagonally to CLT planes. Screws are arranged under an angle of 45°, (or less if CLT deep beams are sloping) between screw axis and member axis in order to provide a higher load-carrying capacity compared to common shear connections due to the high withdrawal capacity of the self-tapping screws. It is important to mention here that screws must always be inserted in the internal side of the building, in order to minimize the possibilities of contact with moisture sources. This subject will be discussed in more detail in the next chapter. 2.2.2.2.3 Connections between adjacent CLT floor panels Diaphragms play a significant role on the performance of a building regarding lateral loads that can resulted from wind or earthquake actions. Together with glulam beams, diaphragms connect the elements which are responsible to resist lateral loads. Further, diaphragms are those elements which absorb lateral loads along all their perimeter, ensuring load transfer between façade elements and vertical loads resisting elements. So, the structural analysis performed by Dias (2017) shown that this kind of connections must be rigid. Regarding the dynamic behavior of a building, the research performed by Moroder (2016) shown some disadvantages associated to flexible diaphragms: increase of internal stresses in the diaphragms of lower floors and the displacements between floors are higher than displacements observed in structural walls. Generally, connections between adjacent floors are performed by self tapping screws. It is an effective and easy solution to apply. 2.2.3 Floor and wall sections The performance of floor and wall systems, not only in the safety level but specially in terms of comfort, is a parameter of huge significance to make homes more attractive to potential buyers, in particular in cases of multi-storey timber buildings. In these cases, vertical and horizontal barriers must ensure that disturbance from one unit to the other are avoided (C. Sigrist et al., 1999). P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 276 Relatively to non-structural wall assemblies, the UT system suggests a non-load bearing light system based on gypsum boards with high levels of sound insulation. In contrast, CLT floor assemblies are a bit more complex. Figure (II) 1:29 and Figure (II) 1:30 present two possible solutions to apply in constructions which are based on the UT system. Presented proposals are based on suggestions published by Hu and Adams, Davide L. (2013). It was decided to present two possible solutions in order to keep the versatility concept always present, however both solutions suggested must be properly tested in order to guarantee its efficiency. Further, it is important to state that much more solutions are possible to apply in this structural system. Figure (II) 1:29 shows a CLT floor assembly in which structural system is totally hidden by flooring and celling elements. According to Hu and Adams, Davide L. (2013), if a normal timber flooring system is required, it would be necessary to create an additional acoustic barrier in order to get the minimum acoustic comfort. This way, it is suggested a simple flooring solution composed by five layers, namely: timber floor, flooring underlayment, two low density wood fiberboards and flooring underlayment. Then, in order to fulfill the acoustic requirements, a ceiling structure needs to be created. So, celling must to be coated by two 15mm gypsum boards layers fixed by means of sound insulation clips. Further a 200mm fiberglass layer is located above gypsum boards. The location of rubber mat between elements is a key point reducing sound propagation. Figure (II) 1:30 presents a solution that does not need a ceiling complement. For that, a less conventional floor system is proposed. This solution is composed by six layers, namely: pre-fabricated concrete topping with 20mm; kaft paper underlayment, subfloor of 25mm, 2 honeycomb acoustic infill of 30mm and again kaft paper underlayment. Here the huge advantage is that timber ceiling can be exposed, however it is important to alert that a concrete floor is not the suitable for residential proposes. Both solutions respect sound transmission classes (STC) and impact insulation class (IIC) as well as reduce flanking transmissions that goes through shared structural building components (Hu & Adams, Davide L. (Adams Associates, 2013). 2.2.3.1 Fire protection Combustibility of wood is the main reason why majority of building codes or regulatory requirements restricts the use of timber as structural material as well as the number of floors allowed (from 2 to 5 storeys for non-sprinkled buildings and from 3 to 8 storeys for sprinkled buildings (Gerard & Barber, 2013)). But, contrary to the general belief, wood can be more fire resistant than both steel and concrete: Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 277 (1) moisture content of wood delays its burning; (2) wood burns with a predicted charring rate; and (3) it creates a protective char layer which allows timber elements to be exposed for extended periods of time during a fire without sacrificing structural integrity and allowing time to evacuate the building (Mahlum et al., 2014). Despite all these findings there is still who claims that it is unrealistic to suppose that fire authorities will permit the unlimited use of timber for tall buildings worldwide in the near future (Frangi et al., 2008) . The main reason for this is related with specificities of fire safety for tall buildings. The worst scenario predicted in a case of a fire in a tall building is when a number of occupants located in the upper part of the building cannot leave the building nether the fire brigade can reach the fire compartment. In these cases, it is assumed that the fire cannot be extinguished and it continues until all combustible material in the fire compartment has burned. Figure (II) 1:29. Acoustic solution that hides all structural system. Figure (II) 1:30. Acoustic solution that exhibits part of structural system. REDUCED FLANKING SOUND TIMBER FLOOR FLOORING UNDERLAYMENT 2 LOW-DENSITY WOOD FIBERBOARDS FLOORING UNDERLAYMENT 5 LAYER CLT PANEL OF 175 MM GLULAM BEAM FIBERGLASS OF 200 MM 2 GYPSUM BOARD OF 15 MM SOUND ISOLATION CLIPS RUBBER MAT RESPECT STC & IIC TIMBER FOOTING REDUCED FLANKING SOUND PREFABRICATED CONCRETE TOPPING OF 20 MM KRAFT PAPER UNDERLAYMENT 2 HONEYCOMB ACOUSTIC INFILL OF 30MM KRAFT PAPER UNDERLAYMENT 5 LAYER CLT PANEL OF 175 MM GLULAM BEAM RUBBER MAT RESPECT STC & IIC SUBFLOOR OF 25MM TIMBER FOOTING P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 278 The UT system was obviously designed thinking on the advantages of exposed timber even if necessary to overdesign structural elements and use active fire protection devices installed for detection, alarm and suppression of fire. However, if it is not proven safe, the UT system can also use passive fire protection systems, designed to confine fire and smoke to designated zones, as well as “building encapsulation” which protects timber elements with a sufficient number of non-combustible claddings like gypsum plasterboards boards in order to prevent timber burning and consequent structural collapse. Considering the known strength decrease of steel connections when exposed to high temperatures, the UT system looks to use embedded steel connects as much as possible. When it is not possible, encapsulation is the easiest solution. 2.3 Architectural considerations and adaptive design 2.3.1 External building appearance The main architectural focus of the UT system is to maximize the freedom in the creative act that architecture represents. Figure (II) 1:31 shows a comparison between cellular construction and the UT system, regarding building shape possibilities. Beginning with plan shape, both structural systems allow the building to have different shapes (rectangular, triangular, hexagonal (…) or even organic shapes). The great difference between these different construction systems arises when the building is extruded. While buildings based on cellular construction system will result on a simple extrusion of plan shape, buildings based on the UT system can grow following angled paths. So, regarding general volumetric shape of building, the great innovative proposal of UT system is the possibility of creating angles between ground and façade planes. Keeping the discussion on external view, the UT system places CLT walls perpendicular to façade allowing wider openings. This possibility combined with CLT deep beams allows to endow the building a strong sense of horizontality, what is totally impossible with a typical cellular construction . In a construction system that is totally based on structural walls, the continuity of opening is always broken by the location of vertical load paths on the façade plan. Relatively to balconies and floating elements, both construction systems are compatible with them. These kind of elements are super common and in addition to its main function, it can easily be used as an architectural tool, in order to endow the building of a personalized shape. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 279 CLT deep beams proposed by the UT system, perform an important role regarding either solar protection or aesthetic possibilities of the building. CLT deep beams can vary their height and shape and change their position relatively to floor plane, resulting in a façade system super versatile. That means that the technology behind the CLT production can be useful to architectural expression. The CNC technology enables the application of CLT panels with different shapes without impair the production process of structural elements. Despite prefabrication is often correlated with modular architecture, this does not means that it does not allow personalized solutions (Larsson et al. 2012). In the case of buildings based on cellular construction, different shapes can also be given to the openings, offering the building a different dynamic, however vertical load paths will be always present. It is true that vertical supporting elements of the UT system are also in contact with façade elements, however the area of contact is reduced and its presence is hidden by CLT deep beams. When a rain screen walls is chosen, the presence of vertical supporting elements becomes even more unnoticed (Figure (II) 1:32). The UT system is the result of a process that tries to dissolve the box suggested by cellular construction . As result, the perception of the plates as a bearing skeleton increases and the space enclosure is generated by a composition of separate elements. Considering the space configurations proposed by Bejder (2012), the UT system is located between the called dissolved box and the ‘ floating’ structure . Vertical elements do not define the interior space while CLT deep beams define the relation between interior and exterior. 2.3.2 Inner building flexibility Relatively to internal spaces and its flexibility, the UT system does not offer an extreme open space concept, as proposed by Michael Green with the FFTT system, however, it allows to place interior frame lines without seriously compromising interior space planning (Figure (II) 1:31). When a construction system allows special flexibility, it is always an advantage. And why is that? Because the use of the building will not be defined by the building configuration. As more flexible is the space or as much less barriers exist in the space, more adaptive it will be. In other words, the use for the space will be defined by the owner/user and not by the architecture itself. This fact is valid for the first use, as well as for different uses that the space can shelter during its life time. So, the goal of UT system goal was to make a space as much flexible as possible, being able to answer people shifting needs and serve a larger range of people for long periods of time. This way, future changes can be simpler and more cost effective when planned for early in the design process and it can also P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 286 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 287 CHAPTER 3 (II) 3 INTEGRATED DESIGN OF THE UT SYSTEM CONSIDERING MOISTURE INDUCED EFFECTS P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 288 3.1 Introduction As already observed, when timber is used as structural material it is, most of the times, hidden behind different wall coverings, either in outside façade or in inner spaces. In other words, timber is used merely as a substitute for steel or concrete, without being considered their unique aesthetic qualities and physical properties. In most examples, architecture is too often more or less reduced to an anonymous white gypsum surfaces where it is impossible to identify the structural building materials hidden underneath. This protective measure is often taken due to the lake of knowledge about timber and its based materials. Obviously, timber construction has to proof to be safe in case of fire (Frangi et al., 2008) (Gerard & Barber, 2013), earthquakes (M. Fragiacomo et al., 2011), to be acoustically and thermally comfortable (Christophe Sigrist et al., 1999) (Hu & Chui, 2013) and moisture resistant (C. Silva, Branco, & Lourenço, 2014) (Green & Eric Karsh, 2012) (Build it Green, 2008), however, in most of times, independent of its properties, timber structural elements end up totally hidden. In past few years, some innovative proposals for multi-storey timber buildings are looking for sustainable solutions for our denser cities. Cross laminated timber (CLT) has been a transversal material being the leading figure in majority of constructed buildings and proposals for new construction systems. For this reason, CLT was the main material selected to shape the UT system. Further, due to the lack of knowledge, the UT system was developed attending special attention to moisture effects on timber structural elements. The present chapter will use the UT system to describe and illustrate construction details which aims to prevent unwanted moisture effects. Experimental evaluation performed focused on the effect of moisture induced effects on CLT, considering timber movements and consequent stresses created between layers, as well as the consequences on mechanical connections performed by self-tapping screws. Experimental campaign and obtained results are described in Part I of the present thesis. Obtained results were crucial to the development of the UT system, either regarding the location of structural elements or the definition of construction details. Structural elements were positioned considering swelling and shrinkage elements and the effect of such movements in the building as a unit. The connections between structural elements were also object of special attention. Different possibilities of barriers between interior and exterior spaces were considered: façade elements, eventual balconies, different positions for balconies and roof. Despite Part I and Part II of the present thesis look to be totally independent researches, they complement each other. In fact, the contact with material and the awareness about the behavior of CLT when Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 289 submitted to moisture changes was crucial for the author to propose a construction system. It is not common to the architect to get a deep knowledge about the construction material and most of times he always applies recommendations of manufacturer without questioning or innovating. Besides the results obtained by experimental campaign developed in the scope of this thesis, bibliographic suggestions were also of great significance, once they also performed tests based on important subjects not directly studied in here. 3.2 Moisture induced effects on CLT elements All the main concepts linked to the effects of moisture variation in CLT elements are properly explained in Chapter I of Part I of the present thesis. Anyway, a brief explanation will be made here. In very simple words, timber hygroscopicity means that the material is permanently ready to release and absorb water (absorption and desorption phenomenon) when exposed to fluctuating atmospheric humidity. Fragiacomo et al. (2011) calls this exposure of “humidity load”, which should be subdivided into various categories, such as external conditions (sheltered or not from precipitation and sun radiation), interior conditions (heated or unheated buildings), and even the activity that building shelters. Both indoor and outdoor structures are exposed to an environment with changing relative humidity and temperature, and in consequence, the moisture content in wood structures will normally change during their use. The most severe effects often take place when the building is new. It is quite common that wood with higher moisture content is built in and then dried after the building is heated. Timber hygroscopic behavior is a very important issue regard to timber building physics because changes on timber moisture content affect physical, mechanical and rheological properties of wood, such as: shrinkage/swelling, internal stresses, strength properties, durability, and decrease of connections resistance. 3.2.1 Shrinkage and swelling effects on CLT The shrinkage/swelling capacity of CLT is not affected regarding its thickness direction, however the crosswise arrangement, shrinkage/swelling movements in-plane movements are reduced. Augustin (2008) says that moisture movements of CLT panels are too small to be measure over the panel surface: less than 1% in across the grain direction and 2% in cross-grain direction. These numbers were verified by an experimental campaign developed in the context of the presented thesis, in which high shrinkage/swelling movements were: 0,3%/-0,27% in across the grain direction (X) and 1,55%/1,28% in cross-grain direction (Y), respectively – see point 2.4.2 from chapter 2 of Part I of the present thesis. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 290 The moisture content of mass timber elements must be controlled from the production to the end of construction, in order to ensure lowest swelling/shrinkage during the life of a building. Beyond this important caution, if the behavior of material is well known, the way it is applied can be changed/adapted in order to reduce the consequences caused by unavoidable changes caused on material properties by moisture changes. Obviously, is an advantage when solutions like cross lamination proves to be efficient on the reduction of timber movements caused by moisture changes. Anyway, if the material is properly known it is possible to work leading with its characteristics and adapt the design to them. One good example of this adaptive design is the Ballom Frame construction which reduces the shrinkage/swelling through the height of a building (Green & Eric Karsh, 2012). According to Glass et al. (2013), when Platform frame construction system is chosen to shape a timber building, quantified cumulative shrinkage over the height of the building is around 3mm per story. Three millimeters may seem insignificant, however it is reason to require special attention during the design of external envelop of the building, in order to avoid possible future damages. Besides the simple variation of dimensions, changes on moisture content of CLT elements can cause important changes on mechanical properties of material. Gülzow et al. (2010) studied the effect of changes of moisture content on MOE and shear modulus of CLT and concluded that, similarly to solid timber, both parameters decrease at they mean level towards an increase of moisture content. However, authors emphasized that swelling of the timber grain leads to an apparent increase in the modulus of elasticity for small service loads due to internal component friction. In the other hand, cracking resulted from the reduction of moisture content leads directly to a distinct decrease in the bending stiffness perpendicular to the grain direction on the face layers. 3.2.2 Moisture induced stresses When CLT panels are exposed to changes in the relative humidity of the ambient air, the restriction of wood movement caused by crosswise bonding of layers can result in moisture-induced stresses and deformations (warping and checking). Moisture gradients are also an important moisture effect which may affect the stress of wood. This effect results from the slow moisture diffusion in wood when humidity load is variable or different from initial equilibrium. As previously mentioned, when wood is exposed to variable humidity conditions it absorbs and desorbs moisture from the air. However, wood needs long time periods to reach equilibrium for different levels of relative humidity, depending on timber size it can take several weeks or even months Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 291 (Time, 1998). This means that timber structures are affected by climatic variations (fast climatic changes), which do not let wood reach equilibrium. As result, moisture gradients are induced in wood sections and hence internal stresses arise. Moisture gradients combined with the restriction of wood movement caused by CLT crosswise bonding induce important differences in shrinkage and swelling of wood, which will develop so-called moisture induced-stresses (MIS), due to constrained swelling or shrinkage strains. Often MIS lead to cracks either in timber surface or in central part of timber sections. In fact, serious structural damages can occur and shape distortions may reduce the serviceability of structural elements (Kim et al., 2010) (Sjödin & Johansson, 2007). One solution to prevent these phenomenon is based on the drastic application of coatings on the timber surfaces exposed to weather conditions in order to reduce the moisture exchange with the environment and thus moisture induced stresses low (Angst & Malo, 2012b). Further, to minimize the impacts of MIS in CLT panels, it is also important to use timber boards with the same moisture content. If the moisture content varies between individual boards residual stresses will develop differently as each layer shrinks or swells, what can significantly weaken the bond strength, causing joint failure and lead to excessive cracking (Harch, 2010). 3.2.3 Timber decay Wood is an organic material which needs to be protected from moisture in order to avoid a material decay. As a timber based material, CLT demands exactly the same precautions. Fungi are one of the biggest wood enemies, however they need the right environment to live and feed on the cellulose, hemicellulose and lignin from wood cells. To perform an attack, fungi needs sufficient moisture, nutrients, oxygen, and a reasonably warm environment. Further, this kind of attack results on reduction of timber mechanical properties only if the same condition remains for a significant period of time (Wang et al., 2010). According to McClung (2013), wood is considered safe from decay if it is exposed to temperatures below 10ºC and a maximum moisture content of 20%, which provides a safety margin for wood species with poor decay resistance. 3.3 The UT system answering to the conditioning of a moisture sensitive material When a structural system based on a moisture sensitive material is developed, some special cares needs to be taken care of. As already mention before, when it is possible to adapt the design of the system to the material characteristics it is always preferable. It is important to keep always present that the material properties must be explored and not camouflaged or altered. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 292 The UT system is an attempt to follow this principle. From definition of structural elements and the way they are linked to each other to the design of a mechanical connection, the influence of moisture induced effects were considered. Present section will describe all structural system and explain how it was thought regarding moisture induced effects on CLT. 3.3.1 Structural elements Structural elements of the UT system are reduced to a small number of elements. There are CLT walls that work almost like columns, glulam beams that join all CLT walls, CLT floors/diaphragms that rigidify the structure and CLT deep beams that sew up all the elements. CLT walls are the main load-carrying elements which get the loads of each floor from glulam beams and drive loads to the foundations, either through a vertical path or an oblique one. These vertical/oblique walls work also like shear walls, resisting lateral loads by being oriented perpendicular to the façades (Ali & Moon, 2007). CLT floors work together with double glulam beams, improving the building stiffness and avoiding the effect of progressive collapse. Finally, CLT deep beams sew up all individual tubes in the building perimeter. Focusing on the adaptive design, the system is based on the Ballom Frame construction principals. Based on this decision, vertical structural elements are designed to be continuous avoiding all possibilities of accumulative shrinkage/swelling inconvenient resulted from floor structures. Similarly to FFTT system, shrinkage/swelling movements of CLT floor, especially on its thickness, will not accumulate over the building height. Further, drainage wall does not need special care to protect lateral side of CLT floor once it will not face the building facade. Being continuous, CLT walls will just accumulate the shrinkage/swelling movements developed through longitudinal direction of CLT panels (X direction indicated in experimental campaign described in chapter II of Part I). According to test results obtained within this thesis, movements on X direction will be always less than 1% even when submitted to extreme humidity changes. It must be recalled that performed tests submitted CLT specimens to RH cyclic changes that varies between 30% and 90%. Here it is also important to mention that despite structural and architectural reasons, CLT walls are placed perpendicular to the façade reducing the area of contact with external environment. However, there is the disadvantage to put in contact with the exterior environment the face of CLT that absorbs water more quickly. In fact, the area in contact with external environment is reduced, however the protective measures against moisture must be effective. Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 293 Double glulam beams, those are placed sideward to CLT walls and fixed to them by means of steel connectors or hardwood elements. Different possibilities for this connection were presented in chapter II and will be analysed regarding moisture content issues forward in the present chapter. Here it is important to argue why double glulam beams are placed sideward of CLT walls. In terms of timber grain direction, the thickness of CLT wall will correspond to radial direction. So, considering the performed tests by the author of present thesis, in worst scenario it will swells in maximum 2% of its thickness and shrinks less than 1% of its thickness. For the building evaluated by Dias (2017), it was considered a thickness of 600mm resulting in movements of 12mm when it swells and 6mm when it shrinks. Adding to CLT wall movements, there is the movements of the glulam beams which needs also to be considered. Still considering the movements on CLT walls thickness, glulam beams will swell and shrink in tangential direction. Further glulam beams will also swell and shrink in radial direction, though its height. That is why it is suggested that CLT Floor are not directly connected to CLT walls. Similarly to glulam beams, CLT floor will swell and shrink in its radial direction without interfere with CLT Walls. CLT deep beams will be placed in the building perimeter hugging all structural elements. In this way, it will be free to move either in radial or tangential directions without compromise the remaining structural elements that are directly in contact with them. If the timber movements are considered high, it could be considered the possibility of reducing them by means of conditioning moisture flow through timber fibres. As observed in Chapter 2 of part I, the possibility to reduce the moisture induced strains is to condition moisture flow on longitudinal direction, however the effect of such measure must be tested on large scale panels in order to verify if such measure do not result in significant deformations. Foundations are made of reinforced concrete in order to avoid prevention works related with subterranean termite and the use of preservative treated wood. Further, despite its low eco-friendly profile, concrete elements can be significantly smaller, when comparing to a fully concrete building due to low weight of timber. Assuming a conservative stance, the UT system resorts to reinforced concrete to shape foundations as well as underground floors and first floor. Connection between concrete and CLT walls are made 3 meters up from ground level and is performed by a built-in steel connector anchored in concrete wall and steel dowels are used to fix CLT wall or by means of Pres-lam system. If the exposed concrete is not an attractive option for designer, concrete wall can be easily covered with timber boards obtained a final look similar to CLT. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 294 3.3.2 Façades Despite the significance of the building envelop on its energy efficiency, indoor air quality and occupant comfort, a special attention has to be given to durability when a timber building is being designed. It is essential to control water intrusion, air flow between interior and exterior as well as water vapour diffusion. As already mentioned in the state of the art (chapter I of part II), timber buildings can last as any other. However, it is mandatory to detail the building properly in order to keep structural elements with safe levels of moisture content. So, detailing a façade of a CLT building must consider two main issues: ensure effective drainage of the façade plane and ensure a diffusion-open, but airtight solution. Unfortunately, a CLT panel cannot be used simultaneously as cladding and structural material. Similarly to solid wood, CLT is a moisture sensitive material and have to be adequately protected from any moisture sources. Obviously the main reason for that is related with durability issues and protection of timber element from rotting. However, there are other compelling reasons for such protective measures: the simple contact of panels with two distinct environments (exterior and interior) can cause significant strains in timber elements and consequently distortions on CLT panels. That is why an effective drainage of façade plane is crucial for a good performance of a timber building. However, the accumulation of water on timber elements may not be result of a poor drainage. One of the most observed errors when a façade is detailed is the application of non-organic materials, free of hygroscopic properties, like, mineral wool, together with organic materials like timber. In a situation like this, the risk to trap water vapour between insulation material and timber element is too high. Wood is a hygroscopic material that absorb and realises water vapour naturally, so betting on a simple airtight solution will not work properly. A diffusion-open envelop, that self-regulates moisture flows, interacting with surrounding environment ensures the ideal conditions for timber elements to last. All wood-based products in the cellulose-based system offer hygroscopic properties which allow them to absorb and release moisture. Diffusion-open vapour barriers can be used to ensure airtightness, but the principle is to let the building envelope breathe by taking up, storing and releasing moisture. This gives a pleasant indoor climate and better air quality. Aware of this real problem, countries where timber construction is much common are adapting modern building techniques to the inherent properties of timber. This means that they are looking for a solution that create an airtight but diffusion-open building envelope that self-regulates natural absorption and releasing moisture. In other words, they are looking for an organic wall that interacts with surrounding environment in order to balance humidity (Rönnelid et al., 2013). It is important to mention here that, Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 295 carefully CLT produced panels are able to airtight a building, however cracking will inevitably happen over time. So, an airtight solution may always be suggested. Looking for an effective solution of a facade compatible with the UT system, four different possibilities for façade drainage are proposed. Figure (II) 1:34 and Figure (II) 1:35 present solutions based on the concept of rain screen system, which is known as the most efficient solution for façade drainage of multi-storey buildings (Build it Green, 2008). Figure (II) 1:37 and Figure (II) 1:38 present solutions based on the glazed curtain façade which is a much appealing solution for architects. As already mentioned, the exposure of timber elements to two environments with different moisture levels can be a problem and cause important distortions and damages. Considering this fact, the four solutions proposed by the UT system also considering the position of CLT walls regarding the barrier between interior and exterior, trying to reduce the exposure of CLT walls to different environments: (I) when the UT system consider a rain screen system CLT wall are entirely on inside environment while CLT deep beams are the elements that make the border between interior and exterior (Figure (II) 1:34); (II) when it consider balconies in the entire perimeter of building, CLT wall are almost entirely on outside environment (Figure (II) 1:35); (III) when it consider an external continuous glazed curtain wall, entire building structure are on inside environment (Figure (II) 1:37); and (IV) when it consider a double glazed curtain wall CLT walls and deep beams are placed in an intermediate environment (Figure (II) 1:38). 3.3.2.1 Rain screen system Rain screen system must ensure an efficient vapour permeable wall set. According to Glass et al. (2013), uncontrolled air leakage can cause moisture accumulation and interticial condensation and consequently CLT decay, besides other problems related with building energy performance, indoor air quality, and occupant comfort. Rönnelid, Wik and Janols (2013), suggest the substitution of mineral wool by an impregnated plywood and a weather shielding sheet. The wood fiber insulation on the other hand has very good hygroscopic capabilities, which means that water is absorbed and released in balance with the moisture level of the surroundings. Contrary to some proposals (Skogstad et al., 2011), which defend CLT air leakage, a vapour permeable and water resistant barrier is recommended. It will prevent moisture causing effects on CLT elements as will also allow drying of construction moisture and drying in service. Vapour permeable materials are desirable for all wall set in order to guarantee the desirable drying capability of entire assembly. Therefore, besides the main drainage plane, created by exterior cladding, two extra drainage planes are created: an air gap between exterior cladding and insulation layer and a P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 302 unexpected infiltration does not cause major damages on inner floor assembly, which is sealed with bitumen sheet when it faces balcony floor. This way, in a scenario in which CLT floor of balcony is damaged for water intrusion, the plate can be easily replaced. The problematic point of the proposed assembly are the double glulam beams which are common for both, inner and outer floor assemblies. As double glulam beams are fixed to CLT walls in order to distribute the loads of diaphragms, there is no possibility to propose different supports for inner and outer CLT floor assemblies. So, glulam beams will be exposed to internal and external environments. In terms of water absorption, it can be controlled or reduced by means of water repellents, however distortions and damages, either on timber elements or in mechanical connections, caused by moisture variations would be a problem. The solution proposed in here suggests that the part of glulam beams that are placed in exterior must be protected by exterior cladding but assuring the ventilation of glulam elements, however further research is required in order to improve this solution and guarantee its efficiency. The proposed solution with balconies also adopts the rain screen system for facades. Despite CLT deep beams do not affect directly thermal behaviour of building, insulation layer is continuous all around the building perimeter and balconies. This way, thermal bridges through balcony structures are not a problem. 3.3.2.4 Windows The main concern related with a window assembly is the maximum control of water path. As mentioned before, the UT system suggests four solutions for openings which vary in windows size, location and support. However, all of them have one thing in common: the concern about control the water path keeping it out of the building. First solution which places the windows aligned with the rain screen façade (Details P1.A and P2.A depicted in Figure (II) 1:36), proposes wider windows, not so high, which are supported on CLT deep beams. As recommended for CLT construction, the exterior side of the window assembly is placed in the same plane of vapour permeable water resistive membrane (Glass et al., 2013). The water is prevented from entering the building by means of sloped metal flashings with end dams located up and bellow the window which keep the water draining by exterior cladding. To keep unexpected water away from window sill, CLT deep beams are finished with a sloped timber shim which is covered with two self-adhered membranes which must be water resistive and vapour impermeable. First membrane overlaps the vapour permeable water resistive membrane while second overlaps the insulation layer. Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 303 Figure (II) 1:40. Detailed solution in which UT system proposes a continuous rain screen system with external balconies. P2.B P2.D P2.C P2.A 27 18 28 27 18 29 30 11 4 31 19 23 24 25 4 18 17 32 15 5 20 9 13 0.1 0.5m LEGEND: 1. Sloped metal roof cap-flashing with end dams 2. Bug screen 3. Five layers CLT deep beam 4. Bitumen sheet 5. Rigid insulation 6. Vapor permeable water resistive membrane/ Air/wind barrier 7. Drained and ventilated air space; wood strapping for cladding fixation 8. Non-combustible cladding 9. Shim 10. Pre-fabricated concrete plates 11. Two layers of water proof rigid insulation with staggered joints 12. Tapered rigid insulation layer 13. Five layers CLT slab 14. Glulam beam 15. Fiberglass of 200mm 16. Sound insulation clip 17. Sloped metal window flashing with end dams 18. Non combustible outer window rim 19. Window assembly 20. Two gypsum boards of 15mm 21. Timber inner window rim 22. Timber footing 23. Timber floor 24. Two low density wood fiberboards 25. Flooring underlayment 26. Rubber mat 27. Sloped metal facade flashing with end dams 28. Cover balcony 29. Non combustible balcony floor 30. Strapping for floor fixation 31. Three layers CLT floor for balconies 32. Timber section for window fixation 33. Sloped metal roof cap-flashing aligned with glass wall 34. Timber frame for glass wall fixation (UNIGLAS ®) 35. Five layers deep beam with water resistive protection 36. Fixed glass panels (UNIGLAS ®) 37. Metal connection between timber frame and CLT deep beams 38. Second inner window assembly P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 304 The proposal that places balconies in the building perimeter (see detail P2.C of Figure (II) 1:40) protect window assemblies from rain by drawing back the window from façade plan, which means less possibilities of water infiltration. The difficulty of this solution is related with the drainage of balcony floor which will prevent water accumulation and consequent water infiltration inside the building. Despite the window assembly is inside, the opening in the facade remains the same. So, sloped metal flashings are also placed on top and bottom of opening, in order to keep water running through the external cladding and the tops of CLT deep beams are also protected with water proofing membranes. The balcony floor (similar to a low-slop roof assembly) meets the inner window assembly by means of two water proofing membranes, which protect inside from accumulative water, and a metal flashing, which makes the water from windows runs to balcony drainage. Regarding Glazed curtain wall system described in section 3.3.2.2, it is important to take in consideration the significance of locating integrated windows in the system. The ventilation needs of such a system imposes the location of enough windows that ensures the adequate ventilation of the building. Further, considering low thermal behaviour and/or condensation effects, a system like this will need air conditioning and mechanical ventilation to perform well. So, it is important to consider that this kind of systems can resort to photovoltaic glass to build the glazed facade and consequently reduce the costs associated to air conditioning and mechanical ventilation. The last proposal, based on the concept of double glass wall (see detail P3.D in Figure (II) 1:39), applies the glazed curtain wall system to building limit and add a second glass layer inside the building, creating inner balconies. The advantages of this system comparing with a single glazed curtain wall are essentially related with a better thermal behavior and with the possibility of natural ventilation. These two glass layers improve the insulation and condensation control. These inner balconies help to conserve energy by balancing interior and exterior temperatures either during cold or warm seasons (Sanders, 2006). 3.3.3 Roof The UT system fits with aesthetics of low-slop roof assembly, which despite its higher propensity for water accumulation it does not need much more care than concrete structures. A conventional solution for a concrete low-slop roof is: application of a sloped layer; placement of water proofing membranes over the entire roof surface and bypassing parapet flashing, position of rigid insulation and sealing its joints to reduce air flow and avoid water penetration, and floor assembly for accessible roofs and for radiation protection. Extra cares related with CLT roof assemblies are: adding another layer of water proofing and Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 305 vapour resistant membrane on the top of insulation layer and ensure the removal of water accumulation near the roof-wall intersections. This last is made by the placement of a sloped shim in the roof-wall intersection (see detail P1.A in Figure (II) 1:36). Some other solutions for low-slop CLT roofs has been suggested, such as: application of closed-cell spray polyurethane foam insulation directly over a membrane on the CLT and covered with a polymeric top coat for UV protection (Glass et al., 2013); the creation of a ventilation gap, similar to rain screen wall system, that allows materials to dry easily (Andreas Ringhofer & Schickhofer, 2014) (Andreas Ringhofer & Schickhofer, 2013). 3.3.4 Connections The main connections developed for the UT system are presented in section 2.2.2 and those that needs special care regarding moisture induced effects are the connections between CLT floors/CLT walls and CLT deep beams and the connections between double glulam beams and CLT walls. 3.3.4.1 Connections between CLT floors/CLT walls and CLT deep beams Regarding the effects of moisture induced effects on the mechanical connections, the main concerns are obviously related with connections that are exposed to most extreme changes on relative humidity and that are more susceptible to accumulate water. Considering that, connections that are placed in façade plane are the most worrying. In façade plane the connections that can be exposed to most moisture variations are those that connects CLT floors with CLT deep beams. As explained in section 2.2.2.2.2 for these kind of connections are performed by self-tapping screws inserted diagonally from the inside side of the building. If the screws are inserted from the outside they would be ended to be an open path to water intrusion. As a result, water could accumulate in timber element and reduce the mechanical capacity of screwed connection. As evaluated in chapter 3 of part I of the present thesis, when 12%≤MC≤18% withdrawal capacity of connection reduces in a range of 1,8% per each percentage unit of moisture content added. When the timber moisture content decreases 8%≤MC≤12%, it would be expected that withdrawal capacity remains the same. However, if the screw goes through an inner gap, a reduction of withdrawal capacity can also occur. It is also important to mention that screws must be inserted diagonally in order to avoid the insertion of screws through parallel direction of timber fibers. In such case, when 14%≤MC≤25% withdrawal capacity of self-tapping screws could decrease up to 4% per each moisture content added. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 306 3.3.4.2 Connections between double Glulam beams and CLT walls The connection between double glulam beams and CLT walls is the most complex in the entire system, and the effects of moisture changes were also considered during its design. In section 2.2.2.1.2 were proposed three different solutions for this connection. All three solutions try to ensure that mechanical connection would allow the movements of timber elements in all three directions in case of moisture induced movements. Steel dowels were chosen in order to ensure a free movement of timber elements without damage the connection and consequently reduce its efficiency. Evaluating the possibilities of timber movements, the most worried situation is if the CLT walls swells in its thickness direction. In this situation located timber smash can occur. In the case of timber shrinkage, the consequences would not be worry because metal dowels would keep the structure safe and reduce timber cracking that could be intensified in case of screwed connections. Another worrying situation is the possibility of delamination of glulam beams resulting in the failure the connections. That is why solutions presented in Figure (II) 1:24 and Figure (II) 1:25 suggest a metal support in the bottom of timber beam. This way, in the case of some delamination, structure collapse can be avoided and metal dowels keeps joining structural elements together. With the exception of the solution that considers external balconies, connections between glulam beams and CLT walls are placed in internal space. Therefore, the variations on environment conditions are not expected to be extreme and predicted timber movements are low. Anyway, all suggested solutions must be properly tested also considering extreme changes on moisture content in order to predict the worst scenario. This way, the ultimate limits of suggested connections would be known. 3.4 Conclusions In the present chapter it was described how moisture induced effects on timber elements were considered during the design of the UT system. Those effects were considered taking into account the location of structural elements, the design of different possibilities for building façade, building roof and mechanical connections between different structural elements. It is important to state here that all proposed solutions were based on the knowledge acquired during the development of Part I of present thesis. The acquired knowledge was essential to develop the solutions here presented and discussed. However, it also must be said that all solutions must be properly tested and studied in order to guarantee their viability. Regarding the presented study, some important conclusions must be pointed: Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 307 - The location/orientation of structural elements regarding building façade can be of significant importance when moisture induced effects are considered; - It is important to reduce the area of contact of timber elements that create the barrier between interior and exterior spaces; - Rain screen system is the solution most applied in multi-storey timber buildings, however glazed curtain wall is the most appealing solution for architects and it also promises to be also an effective solution regarding timber structures; - Solution that suggest the placement of balconies in the building perimeter is the riskiest solution, once it assumes that double glulam beams share internal and external spaces of the building in such a way that the ends of glulam elements are placed in balconies. Despite protected by means of a coating and ventilated the risk of water intrusion is higher; - Connections between CLT floors/CLT walls and CLT deep beams and connections between double glulam beams and CLT walls needs to be tested to be properly known. In present thesis it was decided to use the knowledge obtained by experimental tests on the development of a specific structural system, however the it also could be used to create generic recommendations for designers. So, one of our recommendations for future works is the development of a guide of recommendation about how to design timber buildings considering moisture induced effects. For that, experimental work presented in present thesis would ground some recommendations, but much more experimental evaluations must be done. P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 308 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 309 CHAPTER 4 (II) 4 MAIN CONCLUSIONS OF PART II P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 310 Part II of present thesis was dedicated to the development of a structural system for multi-storey buildings, based on CLT as structural material. Based on some existing buildings and on the results obtained by experiments presented in Part I, UT system was proposed. Chapter 1 summarizes the bibliographic research, explaining the role of timber in the cities of the future. Actually construction of this new kind of buildings has been mainly supported by qualitative reasons, such as sustainable advantages in all its three strands: environmental, social and economic. So, it is mandatory to promote a search for more feasible solutions able to support taller buildings and more versatile architectural solutions. For that, CLT is presented as a precious ally. CLT is a transversal material, being the leading figure in the majority of the constructed buildings as well as in the proposals for new construction systems. One of the advantages of this material is that it can be combined with other structural materials, such as concrete and steel. Some of the existing building and systems in development show that marriage between CLT and other structural materials in order to explore to the maximum the properties of CLT. It is in chapter 2 that, UT system tries to render CLT as a more versatile material and more attractive to construction market. It aims to answer adequately quantitative and qualitative requests of buildings that will fill the cities of the future. Besides environmental advantages, this construction system provides spatial versatility, trying to answer more properly the demands of today’s society, and its structural behavior was evaluated in order to understand its viability. Inspired in the bundled tube concept, UT system works like a cluster of individual tubes connected together in order to make them behave as a single unit. UT system combines CLT and glulam structural elements, in which CLT shapes floors, walls and deep beams, while glulam shapes only beams. As observed in some demonstration buildings, bet in a hybrid solution can be of great advantage in order to explore the properties of the materials involved. This way, either the building structure and architectural demands have less limitations. UT system, ended to be a timber based hybrid solution, once CLT panels are combined with glulam beams. The versatility of construction system was the central issue regarding architectural considerations. In terms of external appearance, UT system offer the possibility to shape sloped facades and it allows to endow the building a sense of horizontality by means of the shape of windows. The interior space can be drawn taking into account few limitations once, despite being placed perpendicular to the façade plane, the location of CLT walls are punctual. Furthermore, it offers the possibility to create external balconies, Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 311 CLT walls can be used as solar protection, different cuts can be made in CLT walls allowing the structure to have an active role on the definition of interior spaces. Regarding structural behavior, UT system was evaluated by Dias (2017), who used as case study a building with 10 storeys. Structural elements were designed and its cross sections defined: CLT walls request a cross section of 600x1500mm (thickness x width); CLT slabs a thickness = 280mm; and glulam beams a cross section of 300x300mm. It was concluded that UT system is a viable solution, however much more evaluation needs to be done, especially concerning lateral loads and vibration of CLT floors. The mechanical connections are those points that raise the most doubts. For the connection between CLT walls and reinforced concrete foundations, Dias (2017) suggested the use of Pres-Lam system, while for the remaining connections some possible solutions were pointed out. The most complex connection is the one that links CLT walls with glulam beams. For that connection three different possibilities were presented, but all of them must be properly tested before state their efficiency. The UT system was developed considering the moisture induced effects, studied in Part I, and it is in chapter 3 that those considerations are explained. The positioning of structural elements, the design of building façade and roof and the definition of mechanical connections are the main processes in which the moisture induced effects were carefully considered. The reduction of the contact area between timber and external environment is of great significance, reason why UT system places structural walls oriented perpendicular to the façade plane. However, if a façade based on a glazed curtain wall can easily neutralize the effect of external environment in structural elements. Anyway, the problematic associated to the drastic changes in temperature and its effects on hygroscopic behavior of timber is still a problem. Connections between structural elements also needs to predict the effect of moisture induced effects, in order to guarantee a long health to the connection. Connections based on self-tapping screws must consider that: screws cannot be inserted from exterior to the interior, preventing that moisture goes inside the timber element using screw as path; screws should preferably be inserted diagonally drilling the main face of CLT panels, avoiding the insertion through side face of panels, especially through longitudinal direction of wood fibers; during design of connections it must be considered the possibility of existence of gaps through screw path as well as the effect of moisture changes. Connections between CLT walls and glulam beams presented in chapter 2, were developed taking into account two main issues related with wood hygroscopic behavior: the movements of timber and their P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 318 Kelly, J. F., Weidmann, B., & Walsh, M. (2011). The Housing We’d Chosse (Grattan Institute Report ; No. 2011-5., Issue June). Kifetew, Girma, Lindberg, H., & Wiklund, M. (1997). Tangential and radial deformation field measurements on wood during drying. Wood Science and Technology , 31 (1), 35–44. https://doi.org/10.1007/s002260050012 Kifetew, Grima. (1996). Application of the deformation field measurement method to wood during drying. Wood Science and Technology , 30 (6). https://doi.org/10.1007/BF00244440 Kim, C., Oh, J., & Lee, J. (2010). Effect of moisture content on performance of dowel-type connection. In A. Ceccoti & J.-W. van de Kuilen (Eds.), World Conference on Timber Engineering . https://www.researchgate.net/publication/271198762_Effect_of_moisture_content_on_perform ance_of_dowel-type_connection KLH-Massivholzplatten. (2011). ETA-06 / 0138 (Vol. 3, pp. 1–17). OIB/EOTA. Krenn, H., & Schickhofer, G. (2009). Joints with inclined Screws and Steel Plates as outer Members. CIBW18/42-7-2 . Kucharek, J.-C. (2009). Wood for the Hood. Riba Journal , March , 51–53. Langenbach, R. (2008). Building Tall with Timber: A Paean to Wood Construction. Structural Engineering International , 18 (2), 130–132. https://doi.org/10.2749/101686608784218725 Larsson, M., Kaiser, A., & Girhammar, U. A. (2012). Case Study Houses 2 . 0 : Mass-Cusomised MultiStorey Timber Buildings – Competitive high-rise timber structures from an architectural and engineering point of view. In P. Quenneville (Ed.), World Conference on Timber Engineering (pp. 442–447). Curran Associates, Inc. Lecompte, D., Smits, A., Bossuyt, S., Sol, H., Vantomme, J., Van Hemelrijck, D., & Habraken, A. M. (2006). Quality assessment of speckle patterns for digital image correlation. Optics and Lasers in Engineering , 44 (11), 1132–1145. https://doi.org/10.1016/j.optlaseng.2005.10.004 Lehmann, S. (2012). Sustainable Construction for Urban Infill Development Using Engineered Massive Wood Panel Systems. Sustainability , 4 (10), 2707–2742. https://doi.org/10.3390/su4102707 Lehmann, S., Reinschmidt, A., & Mustillo, L. (2012). Transition strategies : Accelerating social acceptance and removing the barriers to prefabricated multi-storey timber urban infill developments in Australia using CLT construction systems . (Vol. 61, Issue December). Forest & Wood Products Australia. Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 319 Lepage, R. T. M. (2012). Moisture Response of Wall Assemblies of CrossLaminated Timber Construction in Cold Canadian Climates [Master Thesis, Department of Civil Enginering, University of Waterloo]. http://hdl.handle.net/10012/6569 Lewandowski, C. M., Co-investigator, N., & Lewandowski, C. M. (2015). Summary for Policymakers. In Intergovernmental Panel on Climate Change (Ed.), Climate Change 2013 - The Physical Science Basis (Vol. 1, pp. 1–30). Cambridge University Press. https://doi.org/10.1017/CBO9781107415324.004 Mahapatra, K., & Gustavsson, L. (2008). Multi-storey timber buildings: breaking industry path dependency. Building Research & Information , 36 (6), 638–648. https://doi.org/10.1080/09613210802386123 Mahlum. (2014). CLT feasibility study: A study of alternative construction methods in the Pacific Northwest . https://pdf4pro.com/view/clt-feasibility-study-mahlum-198059.html Mahlum, Co., W. C. C., & Porter Lundeen. (2014). CLT feasibility study. A study of alternative construction methods in the pacific northwest . McClung, R., Ge, H., Straube, J., & Wang, J. (2014). Hygrothermal performance of cross-laminated timber wall assemblies with built-in moisture: field measurements and simulations. Building and Environment , 71 (January), 95–110. https://doi.org/10.1016/j.buildenv.2013.09.008 McClung, V. R. (2013). Field Study of Hygrothermal Performance of Cross-Laminated Timber Wall Assemblies with Built-In Moisture [Master Thesis, Department of Building Science, Ryerson University]. http://digitalcommons.ryerson.ca/dissertations/1059/ Mohammad, M., Gagnon, S., Karacabeyli, E., & Popovsky, M. (2011). Innovative Mid-Rise Timber Structures Offer New Opportunities for Designers. Annual Convention of the Structural Engineers Association of California . Moroder, D. (2016). Floor diaphragms in multistorey timber buildings [Doctoral Thesis, Department of civil engineering, University of Canterbury]. https://doi.org/10.26021/2006 Muñoz, W., Mohammad, M., & Gagnon, S. (2010). Lateral and Withdrawal Resistance of typical CLt connections. In A. Ceccoti & J.-W. van de Kuilen (Eds.), World Conference on Timber Engineering . https://www.researchgate.net/publication/265521436_Lateral_and_withdrawal_resistance_of_t ypical_CLT_connections P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 320 NoDujic, B., Klobcar, S., & Zarnic, R. (2007). Influence of Openings on Shear Capacity of Wooden Walls. New Zealand Timber Design Journal , 16 (1), 5–17. https://www.researchgate.net/publication/277819809_Influence_of_openings_on_shear_capac ity_of_wooden_walls Omland, I., & Tonning, L. (2009). Can wooden architecture bond the urban past to the urban future? International IAPS-CSBE & HOUSING Network 2009 . https://www.researchgate.net/publication/237795896_Can_wooden_architecture_bond_the_ur ban_past_to_the_urban_future_-_what_does_materiality_mean_for_the_genius_a_city ÖNORM EN384:2010. (n.d.). Structural timber ― Determination of characteristic values of mechanical properties and density . CEN. Östman, B., & Källsner, B. (2011). National Building Regulations In Relation To Multistorey Wooden Buildings In Europe . Palermo, A., Pampanin, S., Fragiacomo, M., Buchanan, A., & Deam, L. (2006). InnovativeSeismicSolutions for Multi-Storey LVL TimberBuildings. 9th World Conference OnTimberEngineering . Palermo, A., Sarti, F., Baird, A., & Dekker, D. (2012). From theory to practice: design, analysis and construction of dissipative timber rocking post-Tensioning wall system for Carterton Events Centre, New Zealand. World Conference on Earthquake Engineering . Pan, B., & Li, K. (2011). A fast digital image correlation method for deformation measurement. Optics and Lasers in Engineering , 49 (7), 841–847. https://doi.org/10.1016/j.optlaseng.2011.02.023 Pan, B., Qian, K., Xie, H., & Asundi, A. (2009). Two-dimensional digital image correlation for in-plane displacement and strain measurement: a review. Measurement Science and Technology , 20 (6), 062001. https://doi.org/10.1088/0957-0233/20/6/062001 Pan, B., Xie, H., Wang, Z., Qian, K., & Wang, Z. (2008a). Study on subset size selection in digital image correlation for speckle patterns. Optics Express , 16 (10), 7037–7048. https://doi.org/10.1364/oe.16.007037 Pan, B., Xie, H., Wang, Z., Qian, K., & Wang, Z. (2008b). Study on subset size selection in digital image correlation for speckle patterns. Optics Express , 16 (10), 7037. https://doi.org/10.1364/OE.16.007037 Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 321 Patterson, D. (2013). Forté. Creating the World’s Tallest CLT Apartment Building. US CLT Symposium . Pirnbacher, G., Brandner, R., & Schickhofer, G. (2009). Base Parameters of self-tapping Screws. International Council for Research and Innovation in Building and Construction . Popovski, M., Schneider, J., & Schweinsteiger, M. (2010). Lateral Load Resistance of Cross-laminated Wood Panels. World Conference on Timber Engineering . https://www.researchgate.net/publication/265613292_Lateral_load_resistance_of_crosslaminated_wood_panels Qin, B., & Han, S. S. (2013). Planning parameters and household carbon emission: Evidence from highand low-carbon neighborhoods in Beijing. Habitat International , 37 , 52–60. https://doi.org/10.1016/j.habitatint.2011.12.017 Ranta-Maunus, A. (2001). Moisture Gradient as Loading of Curved Timber Beams. IABSE Symposium Report , 85 (9), 12–17. https://doi.org/10.2749/222137801796348467 Reid, R. L. (2010). Norwegian Team Designs World’s Tallest Timber Building. Civil Engineering . Ringhofer, A, Brandner, R., & Schickhofer, G. (2015). Withdrawal resistance of self-tapping screws in unidirectional and orthogonal layered timber products. Materials and Structures , 48 (5), 1435– 1447. https://doi.org/10.1617/s11527-013-0244-9 Ringhofer, Andreas, Grabner, M., Silva, C. V., & Branco, J. (2014). The influence of moisture content variation on the withdrawal capacity of self-tapping screws. Holztechnologie , 55 (3), 33–40. http://hdl.handle.net/1822/31243 Ringhofer, Andreas, & Schickhofer, G. (2014). Multi-storey residential buildings in CLT - Interdisciplinary principles of design and construction. WCTE 2014 - World Conference on Timber Engineering, Proceedings . http://www.scopus.com/inward/record.url?eid=2-s2.084924944216&partnerID=tZOtx3y1 Ringhofer, Andreas, & Schickhofer, G. (2013). Timber-in-Town – current examples for residential buildings in CLT and tasks for the future. Focus Solid Timber Solutions - European Conference on Cross Laminated Timber (CLT) , 196–218. Rönnelid, M., Wik, T., & Janols, H. (2013). Passive cross laminated timber buildings: Final report Cerbofproject no. 76 (Solar Energy Research Center, Dalarna University, ISSN 1401-7555 ; 104). urn:nbn:se:du-13667 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 322 Roos, A., Woxblom, L., & McCluskey, D. (2010). The influence of architects and structural engineers on timber in construction – perceptions and roles. Silva Fennica , 44 (5), 871–884. https://doi.org/10.14214/sf.126 Sanders, R. M. (2006). Curtain Walls: Not Just Another Pretty Façade. Journal of Architectural Technology , 23 (1), 1–8. Schwab, E., Steffen, A., & Korte, C. (1997). In-plane swelling and shrinkage of wood-based panels. Holz Als RohUnd Werkstoff , 55 (4), 227–233. Serrano, E. (2009). Documentation of the Limnologen Project. Overview and Summaries of Sub Projects Results . Sigrist, C., Gerber, C., Weber, H., & Brunner, R. (1999). Evaluation of floor systems for multi-storey buildings. Pacific Timber Engineering Conference . Sigrist, Christophe, Gerber, C., Weber, H., & Brunner, R. (1999). Evaluation of floor systems for multistorey buildings. Pacific Timber Engineering Conference . Silva, C., Branco, J. M., Camões, A., & Lourenço, P. B. (2014). Dimensional variation of three softwood due to hygroscopic behavior. Construction and Building Materials , 59 , 25–31. https://doi.org/10.1016/j.conbuildmat.2014.02.037 Silva, C., Branco, J. M., & Lourenço, P. B. (2012). MLCC na construção em altura. Congresso Construção 2012 - 4 o Congresso Nacional , 1–12. http://hdl.handle.net/1822/22216 Silva, C., Branco, J. M., & Lourenço, P. B. (2014). UT System : a Structural System To Build Taller Urban Timber Houses With the Aspired Spatial Flexibility. 41st IAHS WORLD CONGRESS , 1–12. https://doi.org/10.13140/2.1.3833.2486 Silva, T. (2014). Explorar a Potencialidade de um Edifício Construído com Madeira Lamelada Colada Cruzada ( MLCC ). Da organização espacial interior à relação interior - exterior . Master Thesis, Department of Architecture, University of Minho. Silva, C. V, Branco, J. M., & Lourenço, P. B. (2013). A project contribution to the development of sustainable multistorey timber buildings. In L. Bragança, M. Pinheiro, & R. Mateus (Eds.), Portugal SB13 - Contribution of Sustainable Building to Meet EU 20-20-20 Targets (pp. 379–386). IISBE Portugal, University of Minho, Instituto Superior Técnico. http://hdl.handle.net/1822/26947 Ta l l b u i l d i n g s u s i n g C LT. A n i n te g r a t e d d e s i g n c o n s i d e r i n g m o i s t u re i n d u c e d e f f e c t s 323 Sjödin, J., & Johansson, C.-J. (2007). Influence of initial moisture induced stresses in multiple steel-totimber dowel joints. Holz Als RohUnd Werkstoff , 65 (1), 71–77. https://doi.org/10.1007/s00107006-0136-6 Skogstad, H. B., Gullbrekken, L., & Nore, K. (2011). Air leakages through cross laminated timber ( CLT ) constructions. NSB 2011 9th Nordic Symposium on Building Physics , 89–96. Smith, T., Pampanin, S., Buchanan, A. H., & Fragiacomo, M. (2008). Feasibility and Detailing of Posttensioned Timber Buildings for Seismic Areas. New Zealand Society of Earthquake Engineering . Sousa, H. S., Branco, J. M., & Lourenço, P. B. (2014). Characterization of Cross-Sections from Old Chestnut Beams Weakened by Decay. International Journal of Architectural Heritage , 8 (3), 436– 451. https://doi.org/10.1080/15583058.2013.826303 Sousa, R. (2010). Avaliação Experimental da Estabilidade Dimensional de Elementos em Pinho Bravo e Madeira Tratada Termicamente [Master Thesis, Department of civil engineering, University of Minho]. http://hdl.handle.net/1822/24768 Stehn, L., & Bergström, M. (2002). Integrated design and production of multi-storey timber frame houses – production effects caused by customer-oriented design. International Journal of Production Economics , 77 (3), 259–269. https://doi.org/10.1016/S0925-5273(00)00153-5 Svensson, S., Turk, G., & Hozjan, T. (2011). Predicting moisture state of timber members in a continuously varying climate. Engineering Structures , 33 (11), 3064–3070. https://doi.org/10.1016/j.engstruct.2011.04.029 Teibinger, M., & Matzinger, I. (2013). Technical Brochure “Construction with Cross-Laminated Timber in Multi-Storey Buildings: Focus on Building Physics” available. Bauphysik , 35 (5), 345–345. https://doi.org/10.1002/bapi.201390039 Time, B. (1998). Hygroscopic Moisture Transport in Wood (Issue February). Doctoral Thesis, Department of Building and Construction Engineering, Norwegian University of Science and Technology. Uibel, T., & Blaß, H. J. (2007). Edge Joints with Dowel Type Fasteners in Cross Laminated Timber. International Council for Research and Innovation in Building and Construction. Working Commission W18–Timber Structures. 40th Meeting. https://www.researchgate.net/publication/36452723_Edge_Joints_with_Dowel_Type_Fasteners _in_Cross_Laminated_Timber P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 324 UN General Assembly. (2015). Transforming our world: the 2030 agenda for sustainable development. In A/RES/70/1 . http://www.un.org/ga/search/view_doc.asp?symbol=A/RES/70/1&Lang=E Wallner, B. (2012). Versuchstechnische Evaluierung feuchteinduzierter Kräfte in Brettschicholz verursacht durch das Einbringen von Schraubstangen . Master Thesis, Institut fur Holzbau und Holztechnologie, Technische Universitat Graz. Wang, J., Clark, J., Symons, P., & Morris, P. (2010). Time to Initiation of Decay in Plywood, OSB, and Solid Wood Under Critical Moisture Conditions. International Conference on Building Envelope Systems and Technology. Wells, M. (2011). Tall Timber Buildings : Applications of Solid Timber Construction in Multistory Buildings. Council on Tall Buildings and Urban Habitat , 1 . https://global.ctbuh.org/resources/papers/download/319-tall-timber-buildings-applications-ofsolid-timber-construction-in-multistory-buildings.pdf Winter, W., Weber, G., Hernández, S., & Brigola, B. (2012). Strategies to increase use of timber in multistorey buildings - case studies. In P. Quenneville (Ed.), World Conference on Timber Engineering (pp. 262–269). Curran Associates, Inc. https://www.researchgate.net/publication/289047278_Strategies_to_increase_use_of_timber_i n_multistorey_buildings_-_Case_studies Xia, B., O’Neill, T., Zuo, J., Skitmore, M., & Chen, Q. (2014). Perceived obstacles to multi-storey timberframe construction: an Australian study. Architectural Science Review , 57 (3), 169–176. https://doi.org/10.1080/00038628.2014.912198 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 325 ANNEX P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 326 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 327 Annex 1:1 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 334 C1_BEFORE CUT_X DIRECTION_STRAIN A1 A2 DAY 35 C1 A1_C1_35_F1 A2_C1_35_F1 DAY 56 A1_C1_56_F1 A2_C1_56_F1 DAY 161 A1_C1_161_F1 A2_C1_161_F1 DAY 182 A1_C1_182_F1 A2_C1_182_F1 DAY 324 A1_C1_324_F1 A2_C1_324_F1 C1_A -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 335 C2_BEFORE CUT_X DIRECTION_STRAIN A1 A2 DAY 35 C2 A1_C2_35_F1 A2_C2_35_F1 DAY 56 A1_C2_56_F1 A2_C2_56_F1 DAY 161 A1_C2_161_F1 A2_C2_161_F1 DAY 182 A1_C2_182_F1 A2_C2_182_F1 DAY 324 A1_C2_324_F1 A2_C2_324_F1 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 336 C3_BEFORE CUT_X DIRECTION_STRAIN A1 A2 DAY 35 C3 A1_C3_35_F1 A2_C3_35_F1 DAY 56 A1_C3_56_F1 A2_C3_56_F1 DAY 161 A1_C3_161_F1 A2_C3_161_F1 DAY 182 A1_C3_182_F1 A2_C3_182_F1 DAY 324 A1_C3_324_F1 A2_C3_324_F1 C3_A -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 337 C4_BEFORE CUT_X DIRECTION_STRAIN A1 A2 DAY 35 C4 A1_C4_35_F1 A2_C4_35_F1 DAY 56 A1_C4_56_F1 A2_C4_56_F1 DAY 161 A1_C4_161_F1 A2_C4_161_F1 DAY 182 A1_C4_182_F1 A2_C4_182_F1 DAY 324 A1_C4_324_F1 A2_C4_324_F1 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 338 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 339 Annex 1:3 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 340 C1_AFTER CUT_Y DIRECTION_STRAIN A1 A2 DAY 35 C1 A1_C1_35_F1 A2_C1_35_F1 DAY 56 A1_C1_56_F1 A2_C1_56_F1 DAY 161 A1_C1_161_F1 A2_C1_161_F1 DAY 182 A1_C1_182_F1 A2_C1_182_F1 DAY 324 A1_C1_324_F1 A2_C1_324_F1 C1_A -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 341 C2_AFTER CUT_Y DIRECTION_STRAIN A1 A2 DAY 35 C2 A1_C2_35_F1 A2_C2_35_F1 DAY 56 A1_C2_56_F1 A2_C2_56_F1 DAY 161 A1_C2_161_F1 A2_C2_161_F1 DAY 182 A1_C2_182_F1 A2_C2_182_F1 DAY 324 A1_C2_324_F1 A2_C2_324_F1 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 342 C3_AFTER CUT_Y DIRECTION_STRAIN A1 A2 DAY 35 C3 A1_C3_35_F1 A2_C3_35_F1 DAY 56 A1_C3_56_F1 A2_C3_56_F1 DAY 161 A1_C3_161_F1 A2_C3_161_F1 DAY 182 A1_C3_182_F1 A2_C3_182_F1 DAY 324 A1_C3_324_F1 A2_C3_324_F1 C3_A -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 Ta l l b u i l d i n g s u s i n g C LT. A n i n t e g r a t e d d e s i g n c o n s i d e r i n g m o i s t u r e i n d u c e d e f f e c t s 343 C4_AFTER CUT_Y DIRECTION_STRAIN A1 A2 DAY 35 C4 A1_C4_35_F1 A2_C4_35_F1 DAY 56 A1_C4_56_F1 A2_C4_56_F1 DAY 161 A1_C4_161_F1 A2_C4_161_F1 DAY 182 A1_C4_182_F1 A2_C4_182_F1 DAY 324 A1_C4_324_F1 A2_C4_324_F1 -2,4E-02 -2,1E-02 -1,8E-02 -1,5E-02 -1,2E-02 -9,0E-03 -6,0E-03 -3,0E-03 0,0E+00 3,0E-03 6,0E-03 1,2E-02 1,5E-02 1,5E-02 1,8E-02 2,1E-02 2,4E-02 P h D P r o g r a m i n C i v i l E n g i n e e r i n g – U n i v e r s i t y o f M i n h o 350 [Document text truncated for crawler view.]