Decision-making for renovating the Mediterranean social housing: a practical approach through an interactive open access tool

Decision-making for renovating the Mediterranean social housing: A practical approach through an interactive open access tool C.M. Calama-Gonz´ alez a,* , R. Escand´ on b , R. Su´ arez b , F. Ascione c a Departamento de Construcciones Arquitect´ onicas y su Control, Escuela T´ ecnica Superior de Edificaci´ on, Universidad Polit´ ecnica de Madrid, Avda. Juan de Herrera 4, 28040 Madrid, Spain b Instituto Universitario de Arquitectura y Ciencias de la Construcci´ on, Escuela T´ ecnica Superior de Arquitectura, Universidad de Sevilla, Av. de Reina Mercedes 2, 41012 Seville, Spain c Department of Industrial Engineering, Universit` a degli Studi di Napoli Federico II, Piazzale Tecchio 80, 80125 Napoli, Italy ARTICLE INFO Keywords: Parametric building stock modelling Multi-objective optimization Investment costs Adaptive thermal comfort Clustering retrofit strategies Deep vs partial renovation ABSTRACT To achieve 2050 Climate Neutrality, building stock requires a multidimensional renovation process. This is particularly urgent in most vulnerable households, with higher exposure to climate change, where this procedure should focus on cost-controlled passive measures. Given the complexity of identifying optimal strategies, it is imperative to improve the retrofitting process of the social housing stock to enhance its energy performance guaranteeing health and comfort. For this, an interactive tool was developed focused on the case of southern Spain. Able to provide optimized combinations of energy retrofit strategies, using NSGA-II genetic algorithms and setting two optimization objectives: minimizing thermal discomfort and economic costs. The freely accessible tool was designed with practical and didactic approach to facilitate decision-making. The results obtained suggest the feasibility of implementing phase actions instead of a single large-scale intervention and show the tool’s ability to quantify the percentage of thermal comfort improvement achieves at each phase. 1. Introduction The phenomenon of climate change, and the new reality it represents, is transforming our way of life and precipitating a metamorphosis at energy, political, and social levels [1]. In this context, a variety of processes and energy policies are put forward with the objective of facilitating the ecological transition and ultimately achieving climate neutrality by 2050. The European Green Deal [2] underscores the imperative of long-term residential renovation as a means of enhancing energy efficiency in the building sector. In the wake of the global pandemic, the European Commission advanced a comprehensive funding plan to drive renovation initiatives. Consequently, the strategy, entitled “A Renovation Wave for Europe – Greening our buildings, creating jobs, improving lives” [3] was devised with the objective of doubling the renovation rate by 2030 while also addressing the issue of energy poverty. The novel fourth version of the Energy Performance of Buildings Directive (EPBD) (EU) 2024/1275 [4] provides an opportunity to address the level and pace of renovation actions. This directive encompasses the implementation of measures such as the Building Renovation Passport (BRP), which supplies a personalized roadmap for individual buildings, enabling emission reduction targets to be met. Furthermore, the incorporation of Building Digital Logbooks (DBLs) is expected to address energy poverty and facilitate decision-making processes for retrofit. Despite the absence of a unified methodological approach for the definition of long-term renovation strategies, individual member states are required to adapt this approach in line with their specific conditions. Nevertheless, the approach of introducing the multidimensional concept of deep renovation (DR) [5] has been widely adopted to achieve climate neutrality. The definitions and implications of deep renovation (DR) are broad, given the comprehensive renovation required for this building stock. The process calls for the implementation of a range of simultaneous passive and active measures, along with notable enhancements in performance and reductions in energy consumption when compared to the pre-renovation state [6]. Different member states have been incorporating the concept of DR into long-term renovation strategies (LTRS) using various criteria, and with a primary focus on buildings. This is primarily made possible by a reduction in primary energy consumption, expressed as a percentage of energy savings. However, energy efficiency * Corresponding author. E-mail address: [email protected] (C.M. Calama-Gonz´ alez). Contents lists available at ScienceDirect Energy & Buildings journal homepage: www.elsevier.com/locate/enb https://doi.org/10.1016/j.enbuild.2025.115629 Received 18 January 2025; Received in revised form 4 March 2025; Accepted 16 March 2025 Energy & Buildings 336 (2025) 115629 Available online 21 March 2025 0378-7788/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). requirements implemented must go hand-in-hand with improvements in the quality of life of users and their capacity to adapt to climate conditions, which is particularly crucial against the current background of increasingly frequent heatwaves [7]. Given the potential environmental, social, and economic benefits resulting from the deep renovation process, several different approaches can be identified, namely (with some exploratory studies). - Evaluation of a techno-economic method for deep renovation with multiple intervention scenarios. Agliardi et al. [8] applied this approach to a real case study of social housing in Reggio Emilia (Italy). - Analysis of the deep renovation market and exploration of the policy, economic, social and technical (PEST) factors influencing the market. Mainali et al. [9] proposed marketing strategies for detached houses in Sweden and Denmark. - Application of a package of measures across all buildings, spatially differentiating cost-effectiveness. ¨ Osterbring et al. [10] concluded that the cost-effectiveness of deep renovation should be assessed on a building-by-building basis rather than by area or neighbourhood. Given the inherent complexity of this issue, the issue of retrofitting a building must be viewed as multidimensional, with a solution that is neither singular nor approximate. Furthermore, the process engages numerous stakeholders, including users, technicians, managers, and builders, each with varying approaches and priorities. The renovation of the residential building stock in response to climate change can be considered a ’wicked problem’, one which Rittel and Webber [11] applied to social policy planning, describing a complex problem that varies over time and cannot be immediately verified or easily reversed. It also differs in individual buildings, requires significant investment, and does not completely solve the problem, although it does improve performance. The preliminary measures considered should focus on the building envelope in order to reduce the thermal load. Such measures can be implemented in a phased manner, have a long operational lifespan, and facilitate energy savings over extended periods. This suggests that, in the context of deep renovation, a positive cumulative effect can be expected. However, as decision-making with combined intervention strategies is a complex task, it becomes difficult to identify the most technically efficient and cost-effective intervention solutions among a diverse set of alternatives. A preliminary comprehensive energy assessment is thus required to ascertain potential energy enhancements, along with an appropriate economic evaluation to contrast investment costs and the improvement predicted. The incorporation of digital technologies is one of the essential mechanisms which can be implemented to adequately address this process. The use of sophisticated methodologies to forecast, examine and mitigate the impacts of climate change on constructions, coupled with the deployment of artificial intelligence (AI) technologies, can be efficient in addressing this challenge, bringing about enhanced, optimized and superior outcomes. The evaluation of improvement potential is typically conducted using Building Energy Modelling (BEM), at the base of Building Energy Simulation (BES). This technique, which allows the construction of simulation models of representative buildings or archetypes based on traditional mathematical models, offers accurate predictions of energy behaviour at building level. However, this methodology has a clear limitation in obtaining results at building stock level, as this larger scale would require significant computational resources to simulate all potential cases [12]. To address these limitations, new emerging digital tools [13] such as artificial intelligence techniques have been developed for application to improving energy efficiency and comfort conditions [14]. These techniques enable the implementation of learning algorithms and computer codes capable of interrelating and coupling with the energy simulation models themselves. This way it is possible to create parameterized Building Stock Models (BSMs) based on existing building archetypes, capable of representing a larger building collective, using fewer computational resources and shorter calculation times, and achieving a larger building stock scale [15]. Based on a data-driven approach, Machine Learning algorithms provide significant advantages in analytical prediction processes, drastically reducing calculation times and resulting in substantial progress for building management, accurate estimations and decision-making for the built environment [16]. Among these, the use of evolutionary techniques for multi-objective numerical optimization [17] such as NSGA-II genetic algorithms [18], automates simulation and energy evaluation processes. This approach equally benefits selective processes, and the assessment of optimal retrofit solutions based on selected variables. It constitutes a rigorous methodological framework offering a broad panorama of combinations of possibilities and strategies, enabling users to choose the options best suited to their specific constraints [19], and proposing effective energy-efficient and energy-saving actions [20]. Thus, the use of automated mathematical building performance optimization (BPO) paired with building performance simulation (BPS) is a means to evaluating many different design options and obtain the optimal or near optimal while achieving fixed objectives. Despite the existing limitations, including model uncertainty, computation time or difficulty of use, as outlined by Attia et al. [21] in their review for their application in net zero energy buildings, the advantages still outweigh these limitations. Considering this methodology, there is a high number of studies that optimized thermal envelope conditions [22], their combined effect with energy supply systems [23], building system operation schedules [24] and solar protection systems [25], among others, with the main optimization objectives of minimizing primary energy consumption or CO 2 emissions [26], energy-related global costs [27] or visual and thermal comfort [28]. In this context, it is also of the utmost importance to highlight the detected research gap on existing open access and free tools that provide real optimized retrofit solutions applied to the housing building at the level stock. Several works provide information on current building thermal and energy performance, of both specific case studies at the single building level or specific neighbourhoods, normally through graphical information, mapping techniques or GIS platforms, that usually classify buildings according to their energy demand or consumption values, which are commonly obtained from Energy Performance Certificates or static calculations [29,30]. Although several papers present energy retrofit strategies applied to the existing housing stock using dynamic modelling, these works normally focus on a single building level case study. In fact, as can be observed in the review of 153 papers conducted by Hashempour et al. [31], only 14.3 % of the studies on building performance present retrofit strategies and, 85.7 % of them consider the single building level. For instance, the work conducted by Ascione et al. [32] is significantly relevant, since optimized retrofit strategies for the Italian housing building stock considering cost savings, carbon dioxide emissions and primary energy consumption are determined. Nevertheless, these authors do not provide any actual retrofit tool which contain the obtained results, simply presenting their findings and conclusions through static figures and written descriptions. Throughout the renovation process, there is one key agent, the homeowner or occupant, especially in cases of energy and socioeconomic vulnerability [33]. Generally, a major barrier arises due to the limited training of homeowners and property managers, coupled with a lack of reliable information [34]. Moreover, few studies connect energy efficiency policies and objectives with homeowners’ perspectives and their ability to renovate their homes [35]. The necessary renovation process should be understood as part of long-term housing stock management, so that the element of time should also be incorporated [36]. In this context and literature framework, several research questions arise: What are the opportunities for homeowners to undertake deep energy renovations? Can different effective renovation strategies be implemented in homes that align with the economic capabilities of their C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 2 owners? The novel objective of the present research project is to develop an open-access and interactive tool for assessing energy retrofit strategies, helping to obtain valuable information for addressing the urgently needed renovation of existing social housing buildings in southern Spain (Mediterranean climate). The main aim of this tool is to provide a comprehensive range of diverse, optimized combinations of energy retrofit solutions, in combination with a basic analytical decisionmaking process that considers a range of criteria (including the improvement of users’ thermal comfort, the minimization of economic aspects and the consideration of the level of building intervention), rather than advocating for a single optimal solution. This open-access and interactive tool may prove beneficial for public stakeholders and users involved in the decision-making processes related to building retrofitting. This research stands out from existing studies in the following ways: - Although the implementation of parametric modelling and artificial intelligence techniques on a multi-objective analysis framework may be standard in numeric optimization approaches, the novelty lays in their application to predict precise thermal performance results of an specific building stock which is in urgent need of energy renovation given its high social, economic and climate vulnerability: the social housing building in southern Spain. Besides, these methodologies are applied at the building stock level rather than at single building level, with reduced computational costs and high predictive capacity. To do so, validated parameterized stock building models are created using characterization building information contained in a large building database (around 39,500 dwellings) which has been previously statistically analysed. - This approach allows to obtain energy simulation results of the stock level based on dynamic calculations that offer the possibility to simultaneously and concisely assess different parameters (such as, thermal comfort and stress, intervention costs, etc.), and integrate time-based scenario evaluations In fact, to more accurately represent the real thermal performance of this vulnerable social stock, adaptive thermal comfort has been considered as an optimization objective, in contrast to the most commonly used approach based on energyrelated aspects. Specifically, a proposal of optimized and thermally efficient retrofit strategy packages for existing buildings through a process of divergent and thinking analysis involving multiple options and alternatives is presented through the development of an openaccess tool. This tool helps regional decision-making, allowing users, technicians or public entities to choose the most suitable options for their specific needs., since it also includes temporal aspects that consider staging renovation solutions. 2. Methodology The methodology followed in this research is explained in the work stages shown in Fig. 1 and includes the combination of different methods: statistical techniques for the assessment of extensive databases and validation of building energy models, collection of on-site measurements of real case studies, construction of dynamic simulation models through energy simulation tools and intelligent computation through multi-objective analysis and automatic numerical optimization with genetic algorithms. 2.1. Description of the methodology stages 2.1.1. Stage 1. Building characterization of the existing social housing stock Building characterization data have been obtained from a public database provided by the Andalusian Housing and Rehabilitation Agency (AVRA in Spanish), whose content has been enhanced by incorporating new study variables. Following the statistical analyses carried out on this database, containing information on approximately 39,500 public social housing dwellings in southern Spain built between 1950–2010, the most representative variability ranges of the typological, morphological, and construction characteristic variables of the existing building stock have been defined. In the present research, the studies have focused on the predominant building typologies (H-block and linear block, which together represent more than 82 % of the buildings included in the aforementioned database) [37,38] located in the climatic zone with the largest extension and representation within the Andalusian Mediterranean area (including cities such as Seville, C´ ordoba, or Huelva). According to the K¨ oppen-Geiger classification this area is classified as Mediterranean climate (Csa) [39]. 2.1.2. Stage 2. Construction of validated building energy models of representative building archetypes At this stage, representative building archetypes of the two predominant building typologies included in the public database (H-block and linear block) are selected and monitored over an extended length of time, considering different seasonal periods (summer, winter, and midseason). Subsequently, dynamic simulation models at single-building level are constructed based on the geometric, physical and constructive data of the case studies, using the EnergyPlus version 9.2 energy simulation software and linking climatic data through “.epw” climate files. These models undergo a prior validation process and are calibrated using Bayesian statistical techniques and the comparison of monitored in-situ data with dynamic simulation software predictions. The results are then validated based on the uncertainty coefficients established in the ASHRAE Guidelines [40]. Previously published works have detailed this calibration and validation process for both the building archetype of the H-block [41] and linear block models [42]. 2.1.3. Stage 3. Construction of parameterized and validated building stock energy models A parameterization process of the validated single-building level models is later carried out, aiming to construct parameterized and validated models at building stock level, applying a statistical bottom-up approach to analyse buildings at a neighbourhood or regional scale. This step allows the construction of representative archetypes of the existing residential stock based on predominant building typologies (linear and Fig. 1. Work stages included in the methodology followed. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 3 H-block buildings, as represented in Fig. 2), and through automatic input setting variation for the parameterized variables, simultaneous simulations can be carried out for thousands of representative buildings, rather than just at single-building level. For this purpose, jEPlus version 2.1 energy simulation software and both EP and Python programming languages have been used, carrying out the parameterization of different variables: 1) Geometric variables. Firstly, the three dimensions of space (X, Y, Z) of the building archetype model are parameterized. Two key variables are defined: the average built area of the dwelling (m 2 ), and the floor-to-ceiling height (m), both parameters which depend on the geometric X, Y and Z variables. Additionally, the building orientation (◦with respect to North) and the window-to-wall ratio of the facades (%) are parameterized, as are the number of floors of the building and the type of urban typology. This last variable allows the analysis of different cases of building block positioning within the urban context: a single building with no shading effects from the surroundings (Isolated), attached case with two party walls (Terraced), or attached case with one party wall (Corner). In order to consider the number of floors and urban typologies, it is necessary to construct as many “.idf” files in EnergyPlus as defined options, which are later imported into the simulation software as a parameterized file where mathematical calculations can be conducted (“.imf”). All this allows the representation of different building geometries. 2) Physical and constructive variables related to the envelope. This includes parameterized variables to define the thermal conductivity properties (W/m⋅K), specific heat (J/kg⋅K), density (kg/m 3 ), and thickness (m) of the roof, facade, and intermediate floors, along with the solar absorptance properties of the roof and facade. It is thus possible to simulate numerous construction envelope solutions with different U-values (W/m 2⋅ K). Additionally, partition wall thickness (m), window glazing and frame types (for the evaluation of windows with different U-values), and infiltration rate (ACH) have been parameterized in the models to consider different scenarios. 3) Operational variables. In this case, people density (people/m 2 ) has been parameterized to simulate different occupancy loads according to the relevant Spanish regulations, as well as the opening of the most commonly used exterior solar protection system in southern Spain (roller blinds). Similarly, the models have been parameterized with the variable of night-time natural ventilation rate (ACH) applied during summer periods in residential buildings in southern Spain (Mediterranean climate) to simulate different scenarios. In other words, all defining variables of the building archetypes have been parameterized. Thus, the variability ranges which result from the statistical evaluation of the extensive building database containing information of the social housing stock in southern Spain (presented in subsection 2.1.1 and later 2.2), have been assigned to the parameterized variables of the archetypes simulation models, as possible value ranges. As a result, the current performance of the existing public social housing of southern Spain (named as “base case”, with no retrofit strategies implemented) at the stock level (in contrast to the single-building level) can be assessed, providing general results at regional scale, through dynamic simulations of representative case study buildings, and also reporting valuable information for the later decision-making retrofit process. In regard to simulation assumptions, shadows casted by neighbouring buildings are not considered in the modelling process given that these aspects cannot be generally parameterized. However, shading casted by the urban grouping of the building typologies themselves, have been taken into account in the case of terrace and corner building typologies. In relation to thermal bridges, adjustments to the conduction calculations for linear and point thermal bridges have been considered during the calibration and validation process of the parameterized building stock model. 2.1.4. Stage 4. Definition of passive and low-cost energy retrofit solutions applicable to the social housing stock The proposal of possible retrofit solutions to improve the energy performance of the existing social housing stock has been formulated considering different criteria, which takes into account expert knowledge and construction practices in the region. Firstly, users’ social vulnerability based on socio-economic aspects was taken into account, as well as the inherent need to propose cost-appropriate retrofit solutions for the users of these dwellings, in keeping with the results reported in [33]. In addition, retrofit measures proposed target the most influential variables on thermal building performance, previously analysed though sensitivity analysis conducted in other works [43]. In this case, where the use of HVAC systems in social dwellings is limited due to the lower spending power of the users for their installation and operation, thermal comfort has been considered a relevant factor in the performance balance of social residential buildings. For this reason, the retrofit measures proposed focus primarily on passive improvement of the thermal envelope of the buildings, the enhancement of operational measures relating to air-exchange variables, and window-configuration aspects. These measures also address the possible application of low-cost active solutions, mostly linked to mechanical ventilation systems. All these considerations are in line with the current European and national energy strategies aimed at achieving social well-being through the decarbonization and retrofit of existing building stock. In view of all the above, a set of energy retrofit solutions is proposed in section 3.2. To effectively incorporate the retrofit strategies proposed in the energy simulation model at the building stock level, it is necessary to perform a second parameterization process, which may allow to simulate thermal performance after energy retrofitting the building stock. Fig. 2. Sample of linear and H-block buildings archetypes. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 4 This entails: A) incorporating new parametrized variables (for example, defining the possible implementation of a mechanical ventilation system or the implementation of constructive retrofit solutions for the building envelope over the existing building (base case); or B) modifying the already parameterized variables (for instance, adding new schedules for the use of natural ventilation or different scenarios for solar protection systems). Thus, this improvement makes it possible to assess the performance of the existing building social housing stock at regional scale through predominant building simulation case studies, considering possible energy retrofit strategies. 2.1.5. Stage 5. Multi-objective analysis for numerical optimization of retrofit solutions The objective variables defined in this research aim to minimize the thermal discomfort of the social users in their home, taking into account their economic and thermal adaptation capabilities. Therefore, the objective variables correspond to the annual percentage of overheating hours (%), the annual percentage of undercooling hours (%), and the initial investment costs in relation to the retrofit solutions ( € /m 2 , where m 2 refers to the built area per housing unit). All three parameters are equally weighted in the optimization problem. Percentages of overheating and undercooling hours are calculated based on the adaptive thermal comfort model defined in EN 167981:2019 [44] and are determined in relation to the percentage of hours exceeding the upper limit of the adaptive comfort band and that falling below the lower limit of the band, respectively. The adaptive thermal comfort band is obtained by defining a predicted percentage of dissatisfied (PPD) lower than 10 %, corresponding to an interval of +3◦C and −4◦C (upper and lower limits, respectively) on the adaptive comfort temperature (T c ). This temperature is derived from the running mean dry bulb outdoor temperature (T e ) which in turn depends on the daily mean dry bulb outdoor temperature for the previous 1 to 7 days (T e1 to T e7 ) (Eqs. (1) and (2)). Further information on this methodology can be found in the referenced standard. Tc=0.33 ×Te+18.8 (1) Te= (Te1 +0.8⋅Te2 +0.6⋅Te3 +0.5⋅Te4 +0.4⋅Te5 +0.3⋅Te6 +0.2⋅Te7)/3.8 (2) Regarding investment costs, each retrofit solution has been associated with a specific Initial Investment Cost ( € /m 2 or € /unit) for its effective implementation in the building. These costs include aspects such as prices linked to the dismantling of layers or elements of the original state, whenever required, or the cost of scaffolding or auxiliary elements, aiming to provide the most realistic cost assessments possible. It should also be noted that these costs were obtained from the Spanish CYPE Price Construction Generator [45] and/or real execution projects. However, they should only be used for the comparison of solutions, as they are subject to characteristic changes in the construction sector and may vary with time. In the optimization process, the NSGA-II genetic machine learning algorithm was used in the definition of the multi-objective problem approach, using the Latin Hypercube Sampling method for randomly and statistically defining the case studies to be analysed. The maximum number of iterations was set at 100, with a crossover rate of 100 % (parameter which represents the frequency with which new solutions are defined by merging features of previous ones). The mutation rate, which refers to the frequency of random changes in new solutions, was considered to be 20 %. Finally, the tournament selection size was set at 2, that is to say, from two random solutions in the population, the algorithm only kept the best solution if it was considered fit. These values chosen were based on the computational resources available and previous experiences conducting multi-objective calculations [46]. In the optimization problem, a set of dominated solutions are obtained, which represent the “Pareto Front” or set of minimum values, since all the three optimization objectives are minimized. This means that a set of optimal strategies may be obtained, offering users a basic process of primary analysis and decision-making adapted to their specific needs. For this purpose, the jEPlus +EA version 2.1 simulation software, as well as the EP and Python programming languages are used for the formulation of the optimization problem. 2.1.6. Stage 6. Creation of an interactive open-access tool for visualizing optimized retrofit strategies To promote the dissemination of results while ensuring a significant social impact, a free tool that contains the scientific findings reported in this research in terms of optimized retrofit strategies applied to the existing social housing stock in southern Spain has been generated using the open-access package HiPlot (High-dimensional interactive plotting), an interactive visualization tool for high-dimensional data through parallel plots [47] created with Python programming language. The interactive parallel axes plot has been exported to “.html” format, enabling any user to visualize and manipulate the results via a web browser. Moreover, users have the capability to export the data to a “. csv” file for further data processing. Thus, the tool becomes a fully open and operational resource for the user. 2.2. Case study description 2.2.1. Statistical characterization of the social housing building in southern Spain The climatic severity of the locations evaluated can be defined by Cooling Degree-Days (CDD) and Heating Degree-Days (HDD). A base temperature of 20 ◦C, the value used by the current Spanish CTE standard [48] to define climatic areas, has been defined for the calculation of CDD and HDD. CDD values for the analysed climate zone range from 1190 to 1490, while HDD values range from 700 to 970. Table 1 summarizes the variability ranges detected for the characterization of the most representative building typologies (H-block and linear block). Taking into consideration these ranges, it should be noted that 64 % of the H-shaped buildings found in these territories are 3 to 5 floors high. Furthermore, 89 % of them have a window-to-wall ratio of between 10–30 %, and in 91 % of cases, the dwellings have an average built area of around 70 to 115 m 2 . Regarding the linear blocks, 88 % of the buildings of this typology are between 3 and 4 floors high, with 78 % having a window-to-wall ratio of between 10 and 20 %. Similarly, a simulation of linear block cases with average dwelling built areas of between 70 and 115 m 2 would include 81 % of the buildings in this climatic region. These data further support the representativeness of the variability ranges obtained. 2.2.2. Energy retrofit solutions proposed for the social housing stock Based on the criteria described in section 2.4, the following energy retrofit solutions have been considered (Table 2): 1) For the roof retrofit, a total of 15 solutions were proposed, in addition to the non-retrofitted case. Both the improvement associated with the application of a low emissivity external paint, with a solar absorptance below 0.3 (P), and solutions involving the addition of thermal insulation to the original roof were considered. In this last case, thermal insulation was added internally (In) or externally (Out) to the base case roof, in different thicknesses (0.06, 0.08, 0.09, 0.10, 0.12, 0.14 m) depending on the type of insulation used (MW or mineral wool with a thermal conductivity of 0.045 W/m⋅K and XPS C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 5 or extruded polystyrene with a thermal conductivity of 0.034 W/ m⋅K). Additionally, the possibility of incorporating a green roof solution (Green) was also considered. The 15 solutions are defined in relation to the combination of all these parameters, presenting the most commonly used retrofit solutions in the market. 2) For the facade, 15 possible solutions were analysed, in addition to the non-retrofitted case. As seen on the roofs, the possible application of a low emissivity external paint with a solar absorptance lower than 0.3 (P) was also considered for the facade, as was the incorporation of thermal insulation on the base case solution. For the facade, insulation thicknesses considered were 0.04, 0.05, 0.06, 0.075, 0.08, 0.10, and 0.12 m, depending on the type of insulation used and its position on the façade. As seen in the table mentioned, the proposed insulations correspond to MW (mineral wool with thermal conductivity of 0.037 W/m⋅K), PUR (polyurethane with 0.028 W/m⋅K), EPS (expanded polystyrene with 0.037 W/m⋅K), and RW (mineral wool with 0.035 W/m⋅K). Regarding the position of the insulation, interior insulation (In), exterior insulation (Out), and a scenario of injecting thermal insulation into the pre-existing air cavity (Ca) were contemplated. A ventilated facade solution (FV) adapted to the Mediterranean climate was also proposed, considering an opaque ceramic cladding. 3) 12 configurations were analysed for windows, in addition to the nonretrofitted case (single glazing with aluminium frame being the most common solution based on the information recorded in the public database). These solutions arose from the combination of different double glazing with different frame types (PVC or aluminium with thermal bridge break). Regarding double glazing, the possibility of including low emissivity glasses is contemplated, both on the exterior and interior faces, with thicknesses of 4 and 6 mm, and with 8, 10, or 12 mm air gaps. Triple glazing windows and cavity gases with better performance (argon, xenon, krypton) were not considered due to their high economic cost for social housing. Likewise, retrofit measures linked to operational and usage parameters of the dwellings are proposed (Table 3), specifically associated with: 1) Natural ventilation. Optimizing the use schedule of the natural ventilation systems in the dwellings (windows) according to the seasonal period (summer and winter) was considered as a retrofit strategy. As a result, four different schedules for natural ventilation have been included. 2) Solar protection systems. Four possible scenarios were taken into account: absence of solar protection systems (non-retrofitted case); implementation of external mobile solar protection systems based on PVC roller blinds, either with an average annual aperture of 50 % of the time open; or with an optimized opening schedule depending on the season (summer and winter); and, finally, the incorporation of a fixed external solar protection system with slats. 3) Mechanical ventilation system. The consideration of mechanical ventilation as a low-cost HVAC system was included, whose operation is established based on the requirements included in the applicable Spanish regulations [50], which establish a continuous use schedule. Therefore, to assess its implementation, two possible scenarios are considered (ON, with mechanical ventilation system) and OFF (without the mechanical system). 3. Analysis and results 3.1. Optimization of retrofit solutions The results analysed in this section have been obtained from numerous dynamic simulations completed. In the case of the H-block building, a total of 115,538 simulations were performed, yielding 15,740 sub-optimal solutions (13.6 %). For the linear block, 7,440 suboptimal solutions (20 %) were obtained out of 37,031 simulations conducted in total. It is important to note that the execution of such a high number of simulations is due to several factors: 1) The existing differences in the variability ranges of the building variables which define the archetypes of the H-blocks and linear blocks. 2) The varying levels of geometric complexity and definition of the parametric stock simulation models associated with each building typology. 3) The need to provide a tool which allows results to be obtained based on different case studies with specific initial conditions. That is, certain defined inputs which enable the user to select the case study that most closely resembles their starting conditions (orientation, number of building floors, urban typology, average built area of the dwelling, and percentage of glazing surface), later analysing the output variables (retrofit solution packages, thermal comfort Table 1 Variability ranges of the existing social housing stock for the predominant typologies. Variables Hblock Linear block Geometry Urban typology Isolated, Terraced, Corner* Orientation (◦) N-S, E-W Floor area (m 2 ) 70–115 Floor height (m) 2.50–3.00 Window-to-wall ratio (%) 10–30 10–20 Number of storeys 3–5 3–4 Building envelope Roof solar absorptance 0.3–0.9 Roof U-value (W/m 2 ⋅K) 1.2–2.4 Roof thickness (m) 0.25–0.45 Roof thermal conductivity (W/ m⋅K) 0.3–0.6 Roof density (kg/m 3 ) 1000–1800 Roof specific heat (J/kg⋅K) 500–1500 Floor U-value (W/m 2 ⋅K) 3.0–7.00 Floor thickness (m) 0.15–0.30 Floor thermal conductivity (W/ m⋅K) 0.7–1.7 Floor density (kg/m 3 ) 1200–1800 Floor specific heat (J/kg⋅K) 500–1500 Facade solar absorptance 0.3–0.9 Facade U-value (W/m 2 ⋅K) 1.2–2.5 Facade thickness (m) 0.10–0.30 Facade conductivity (W/m⋅K) 0.2–0.4 Facade density (kg/m 3 ) 1000–3000 Facade specific heat (J/kg⋅K) 500–1500 Partition thickness (m) 0.07–0.12 Type of window glass Single Type of window frame Aluminium Window U-value (W/m 2 ⋅K) 5.50–5.70 Infiltration rate (ACH) 0.30–1.00 Operation People density (people/m 2 ) 0.01–0.15 Natural ventilation rate (ACH) 0–4 Summer night-time natural ventilation 22:00–8:00 Blinds aperture No blind, 50 %, totally open * Explained in detail in subsections 2.3 and 2.6. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 6 Table 2 Passive solutions considered for retrofitting the social housing stock. Element Label TI position TI type TI thickness (m) Pavement Green coverage Rendering Out. Glass (mm) Gap (mm) In. glass (mm) Frame U-value (W/ m 2 ⋅K) Approximate investment cost ( € /m 2 ) Roof Unretrofitted (non-retrofitted) − − − X− − − − − − 1.2–2.4 − P_Roof − − − X−P− − − − 1.2–2.4 29.12 InMW_0.08 In MW 0.08 X − − − − − − 0.26–0.37 16.63 InMW_0.09 0.09 X − − − − − − 0.24–0.33 18.71 InMW_0.10 0.10 X − − − − − − 0.23–0.30 19.97 InMW_0.12 0.12 X − − − − − − 0.20–0.26 23.28 P_InMW_0.08 0.08 X −P− − − − 0.26–0.37 45.75 OutXPS_0.06 Ex XPS 0.06 X − − − − − − 0.29–0.43 34.59 OutXPS_0.08 0.08 X − − − − − − 0.25–0.34 37.16 OutXPS_0.10 0.10 X − − − − − − 0.22–0.28 46.30 OutXPS_0.14 0.14 X − − − − − − 0.17–0.21 53.62 P_OutXPS_0.08 0.08 X −P− − − − 0.25–0.34 66.28 Green_OutXPS_0.06 0.06 −X− − − − − 0.27–0.39 129.17 Green_OutXPS_0.08 0.08 −X− − − − − 0.23–0.32 133.36 Green_OutXPS_0.10 0.10 −X− − − − − 0.21–0.27 138.8 Green_OutXPS_0.14 0.14 −X− − − − − 0.17–0.20 146.12 Wall Unretrofitted (non-retrofitted) − − − − − X− − − − 1.2–2.5 − P_Wall − − − − − P− − − − 1.2–2.5 23.96 InMW_0.05 In MW 0.05 − − X− − − − 0.37–0.56 60.31 InMW_0.06 0.06 − − X− − − − 0.33–0.48 61.02 InMW_0.075 0.075 − − X− − − − 0.29–0.41 64.22 P_InMW_0.06 0.06 − − P− − − − 0.33–0.48 84.98 CaPUR_0.04 Ca PUR 0.04 − − X− − − − 0.36–0.54 36.61 CaPUR_0.05 0.05 − − X− − − − 0.32–0.45 38.54 P_CaPUR_0.05 0.05 − − P− − − − 0.32–0.45 62.51 OutEPS_0.05 Ex EPS 0.05 − − X− − − − 0.37–0.56 79.47 OutEPS_0.06 0.06 − − X− − − − 0.33–0.49 80.46 OutEPS_0.08 0.08 − − X− − − − 0.28–0.39 82.45 OutEPS_0.10 0.10 − − X− − − − 0.25–0.32 84.44 OutEPS_0.12 0.12 − − X− − − − 0.22–0.27 86.40 P_OutEPS_0.06 0.06 − − P− − − − 0.33–0.49 104.43 VF_OutRW_0.06 RW 0.06 − − − − − − − 0.38–0.56 133.00 Window Unretrofitted (non-retrofitted) − − − − − − 4− − Al 5.5–5.7 From 616.91 to 702.07 € /unit 4LE-Air8-6 − − − − − − 4LE Air8 6 PVC or Al TBB 2.0–2.3 4LE-Air10-6 − − − − − − 4LE Air10 6 1.8–2.0 4LE-Air12-6 − − − − − − 4LE Air12 6 1.7–1.9 4-Air8-6LE − − − − − − 4 Air8 6LE 2.0–2.3 4-Air10-6LE − − − − − − 4 Air10 6LE 1.8–2.0 4-Air12-6LE − − − − − − 4 Air12 6LE 1.7–1.9 TI: thermal insulation. Out: external. In: internal. P: low emissivity external paint. MW: mineral wool (0.045 W/m⋅K in roof, 0.037 W/m⋅K in wall). XPS: extruded polystyrene (0.034 W/m⋅K). PUR: polyurethane (0.028 W/ m⋅K). EPS: expanded polystyrene (0.037 W/m⋅K). RW: mineral wool (0.035 W/m⋅K). Ca: air cavity. VF: ceramic ventilated facade. LE: low emissivity. Al: aluminium. TBB: thermal bridge break. All roof and wall solutions maintain the existing base solution. Roof solutions offer the possibility of modifying the external coverage (pavement or green). Investment costs include dismantling old elements in the solution when needed. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 7 conditions and initial investment costs of the interventions) in order to select the most suitable option. The scatter plots in Fig. 3 show the set of non-dominated or optimal solutions (Pareto Front) obtained in the multi-objective analysis for the H-block and linear block, respectively. The annual percentage of overheating hours on the X-axis and undercooling hours on the Y-axis are indicated. The solutions have been classified by intervention cost level: low cost (up to 50 € /m 2 , in green), medium cost (between 50 and 100 € /m 2 , in blue), and high cost (over 100 € /m 2 , in red). Each point on the graph corresponds to a package or set of retrofit solutions that affect the thermal envelope and both operational and solar protection systems. Generally, it can be observed that there is a wide range of optimized retrofit solutions, especially low-cost ones, which lead to a significant improvement in the thermal performance of existing homes. It can also be deduced that solutions offering a more comprehensive and complete energy retrofit (high-cost strategies) do not always result in a substantial improvement or the best thermal performance from an optimization perspective. Instead, they present more bounded results, with percentages of overheating hours ranging from approximately 30 to 67 % and undercooling hours from 20 to 50 %. This contrasts with the greater dispersion of the rest of the lower-cost solutions, for example, with ranges of 33 to 75 % of overheating hours and 20 to 75 % of undercooling hours for the low-cost strategies. Moreover, in solutions up to 50 € /m 2 , a clearer linear trend is observed, especially in the case of the linear block. Another interesting aspect is that, the higher the economic cost of the retrofit strategies, the greater the reduction in the percentage of undercooling hours, which is more significant than that of overheating hours. This is a crucial aspect: since the analysis has been developed on an annual basis, incorporating both summer and winter, achieving a balance between solutions to ensure thermal improvement throughout the whole year is a complex task. 3.2. Interactive open-access tool The interactive tool developed to visualize the optimized combinations of energy retrofit strategies for the existing social housing stock in southern Spain, and the corresponding user manual, are freely available as Mendeley Research Data: https://doi.org/10.17632/c7fcc3yryj.1 (Hblock buildings) and https://doi.org/10.17632/9kz9dchhj3.1 (linearblock buildings). This tool is also registered in the intellectual property registries 04/2024/3551 and 04/2024/3552. Fig. 4, which serves to illustrate the tool’s capabilities, will subsequently be employed to explain its usage and functionality in this paper. In defining the parallel axes plot and in order to ensure the most userfriendly manipulation possible, a set of input parameters or variables was defined to allow the user to specify the initial conditions of their particular case study. Consequently, the tool provides results on how the selection of different energy retrofit packages leads to specific thermal comfort outcomes in social dwellings in southern Spain, based on predetermined intervention costs. Thus, the user is also able to select the desired level of intervention in the building, any possible economic constraints, or the desired improvement in comfort conditions to be achieved. Thus, the following variables are defined (Fig. 4a): 1) Input data. “Orientation” is the first variable considered, analysing the predominant North-South (“N-S”) and East-West (“E-W”) orientations. Following this, the number of “Building floors” (3, 4, or 5 floors) is established. Subsequently, the selection of the urban “Typology” category allows the user to choose between an “Isolated” building (without shading effects from the surroundings), a “Terraced” building (an attached case with two party walls), or “Corner,” which is an attached case with one party wall. In the last case, and depending on the selected orientation, for buildings with northor east-facing party walls, the user should select “Corner (upper, right)”, while if the party wall faces south or west, the “Corner (lower, left)” option should be selected. Additionally, the user can approximate the average built area of their home according to the ranges included in the graph on the “Area” axis (from 70 to 115 m 2 , each 5 m 2 ). The last input corresponds to the percentage of glazing surface (“WWR”) of the home (from 10 to 20 % or 30 % every 5 %, depending on the predominant building typology analysed). All these parameters and their possible options have been defined based on the predominant variability ranges obtained in the building characterization in stage 1. 2) Output data. This corresponds to the variables that the user can analyse. Firstly, the outputs related to the retrofit solution packages are provided. These include the type of glazing and frame (“Glass” and “Frame” in the interactive figure), as well as the constructive solution of the opaque envelope (“Wall” and “Roof”). Additionally, variables referring to the type of solar protection system (“Solar protection”), the schedule of the natural ventilation system (“Nat Vent”), and the use of a mechanical ventilation system (“Mech Vent”) are included. All the aforementioned variables include a nonretrofitted case. The description of the symbols used in the interactive figure to define each possible option can be found in the “Label” column in Tables 2 and 3 of this paper. Finally, also as output data, the graph displays the results of thermal comfort conditions, based on the percentage of annual “Overheating hours” and “Undercooling Table 3 Operational solutions considered for retrofitting the social housing stock. Solution Label Summer Winter Approximate investment cost ( € / unit) Natural ventilation Off −−− Sum&Win_8-9 h 8:00–9:00 8:00–9:00 − Sum_8-9 h, Win_14-15 h 8:00–9:00 14:00–15:00 − Sum_22-8 h, Win_8-9 h 22:00–8:00 8:00–9:00 − Sum_22-8 h, Win_14-15 h 22:00–8:00 14:00–15:00 − Roller blinds No protection No solar protection system − Blinds 50 % 50 % opened 137.53 Blinds optimized 0*% from 8:00–16:00 50 % from 16:00–21:00 100 % 21:00–7:00 100 % from 9:00–19:00 0*% from 19:00–9:00 137.53 External protection External solar protection (slats) External solar protection (slats) 394.06 Mechanical ventilation On continuous ON 650.00 OFF OFF − * 0 % blind aperture level means totally closed and 100% refers to totally open. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 8 Fig. 3. Dispersion plots of the annual % of overheating and undercooling hours of the optimized results, classified according to economic investment level: a) up to 50 € /m 2 for the H-block and b) up to 50 € /m 2 for the linear block; c) between 50 and 100 € /m 2 for the H-block and d) between 50 and 100 € /m 2 for the linear block; e) over 100 € /m 2 for the H-block and f) over 100 € /m 2 for the linear block. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 9 [25] A. Allouhi, M.B. Amine, C. Reisch, Multi-objective optimization of solar energy systems for electricity and hot water generation in collective residential buildings considering the power-to-heat concept, Appl. Therm. Eng. 230 (2023) 120658, https://doi.org/10.1016/j.applthermaleng.2023.120658. [26] G. Manc` o, U. Tesio, E. Guelpa, V. Verda, A review on multi energy systems modelling and optimization, Appl. Therm. Eng. 121871 (2023), https://doi.org/ 10.1016/j.applthermaleng.2023.121871. [27] F. Ascione, N. Bianco, G.M. Mauro, D.F. Napolitano, Building envelope design: multi-objective optimization to minimize energy consumption, global cost and thermal discomfort. Application to different Italian climatic zones, Energy 174 (2019) 359–374, https://doi.org/10.1016/j.energy.2019.02.182. [28] M. Baghoolizadeh, M. Rostamzadeh-Renani, R. Rostamzadeh-Renani, D. Toghraie, Multi-objective optimization of Venetian blinds in office buildings to reduce electricity consumption and improve visual and thermal comfort by NSGA-II, Energ. Buildings 278 (2023) 112639, https://doi.org/10.1016/j. enbuild.2022.112639. [29] C. Beltr´ an-Velamaz´ an, M. Monz´ on-Chavarrías, B. L´ opez-Mesa, A new approach for national-scale Building Energy Models based on Energy Performance Certificates in European countries: the case of Spain, Heliyon 10 (3) (2024), https://doi.org/ 10.1016/j.heliyon.2024.e25473. [30] C. Villanueva-Díaz, M. ´ Alvarez-Sanz, ´ A. Campos-Celador, J. Ter´ es-Zubiaga, The open data potential for the geospatial characterisation of building stock on an urban scale: methodology and implementation in a case study, Sustainability 16 (2) (2024) 652, https://doi.org/10.3390/su16020652. [31] N. Hashempour, R. Taherkhani, M. Mahdikhani, Energy performance optimization of existing buildings: a literature review, Sustain. Cities Soc. 54 (2020), https://doi. org/10.1016/j.scs.2019.101967. [32] F. Ascione, N. Bianco, C. De Stasio, G.M. Mauro, G.P. Vanoli, Addressing large-scale energy retrofit of a building stock via representative building samples: public and private perspectives, Sustainability 9 (6) (2017) 940, https://doi.org/10.3390/ su9060940. [33] C.M. Calama-Gonz´ alez, R. Escand´ on, R. Su´ arez, A. Alonso, ´ A.L. Le´ on-Rodríguez, Household energy vulnerability evaluation in southern Spain through parametric energy simulation models and socio-economic data, Sustain. Cities Soc. 103 (2024) 105276, https://doi.org/10.1016/j.scs.2024.105276. [34] P. Tuominen, K. Klobut, A. Tolman, A. Adjei, M. de Best-Waldhober, Energy savings potential in buildings and overcoming market barriers in member states of the European Union, Energ. Buildings 51 (2012) 48–55, https://doi.org/10.1016/j. enbuild.2012.04.015. [35] R. Chen, R. Fan, Q. Yao, R. Qian, Evolutionary dynamics of homeowners’ energyefficiency retrofit decision-making in complex network, J. Environ. Manage. 326 (2023) 116849, https://doi.org/10.1016/j.jenvman.2022.116849. [36] T. Fawcett, Exploring the time dimension of low carbon retrofit: owner-occupied housing, Build. Res. Inf. 42 (4) (2014) 477–488, https://doi.org/10.1080/ 09613218.2013.804769. [37] C.M. Calama-Gonz´ alez, R. Su´ arez, A.L. Le´ on-Rodríguez, 2020. Building characterisation and assessment methodology of social housing stock in the warmer Mediterranean climate: the case of southern Spain, In: IOP Conference Series: Earth and Environmental Science, Vol. 410, No. 1, p. 012049, IOP Publishing, doi: 10.1088/1755-1315/410/1/012049. [38] C.M. Calama-Gonz´ alez, R. Escand´ on, A. Alonso, A.L. Le´ on-Rodríguez, R. Su´ arez, Building assessment and statistical characterisation of the mediterranean social housing stock in Southern Spain, in IOP Conference Series: Earth and Environmental Science, Vol. 1050, No. 1, p. 012020, IOP Publishing, 2022, July. doi: 10.1088/1755-1315/1050/1/012020. [39] M. Kottek, J. Grieser, C. Beck, B. Rudolf, F. Rubel, World map of the K¨ oppen-Geiger climate classification updated, Meteorol. Z. 15 (3) (2006) 259–263, https://doi. org/10.1127/0941-2948/2006/0130. [40] ASHRAE, American, Society of Heating, Refrigerating and Air Conditioning Engineers. ASHRAE Guideline 14-2014: Measurement of Energy, Demand and Water Savings, ASHRAE, 2014. [41] C.M. Calama-Gonz´ alez, R. Su´ arez, ´ A.L. Le´ on-Rodríguez, Thermal comfort prediction of the existing housing stock in southern Spain through calibrated and validated parameterized simulation models, Energ. Buildings 254 (2022) 1–14, https://doi.org/10.1016/j.enbuild.2021.111562. [42] R. Escand´ on, F. Ascione, N. Bianco, G.M. Mauro, R. Su´ arez, J.J. Sendra, Thermal comfort prediction in a building category: Artificial Neural Network generation from calibrated models for a social housing stock in southern Europe, Appl. Therm. Eng. 150 (2019) 492–505, https://doi.org/10.1016/j. applthermaleng.2019.01.013. [43] C.M. Calama-Gonz´ alez, R. Su´ arez, R. Escand´ on, Building parameter influence on overheating and undercooling risks in the Mediterranean social housing stock of southern Spain, in: E3S Web of Conferences, Vol. 545, p. 04001, EDP Sciences, 2024, doi: 10.1051/e3sconf/202454504001. [44] European standards committee, EN 16798–1:2019 Energy Performance of Buildings – Ventilation for buildings - Part 1: Indoor Environmental Input Parameters for Design and Assessment of Energy Performance of Buildings Addressing Indoor Air Quality 2019 Thermal Environment Lighting and Acoustics - Module M1 6. [45] C.Y.P.E. Ingenieros, Generador de precios de la construcci´ on. Espa˜ na, http://www. generadordeprecios.info/#gsc.tab=0(Accessed 17 May 2024). [46] C.M. Calama-Gonz´ alez, P. Symonds, ´ A.L. Le´ on-Rodríguez, R. Su´ arez, Optimal retrofit solutions considering thermal comfort and intervention costs for the Mediterranean social housing stock, Energ. Buildings 259 (2022) 1–13, https://doi. org/10.1016/j.enbuild.2022.111915. [47] HiPlot, High Dimensional Interactive Plotting Circle Cl. Available online: htt ps://pypi.org/project/hiplot/ (accessed 17 May 2024). [48] CTE, Spanish Technical Building Code, Basic document: energy saving, 2019. Spanish Government. Available online: https://www.codigotecnico.org (accessed 17 May 2024). [49] L. De Boeck, S. Verbeke, A. Audenaert, L. De Mesmaeker, Improving the energy performance of residential buildings: a literature review, Renew. Sustain. Energy Rev. 52 (2015) 960–975, https://doi.org/10.1016/j.rser.2015.07.037. [50] G. Semprini, R. Gulli, A. Ferrante, Deep regeneration vs shallow renovation to achieve nearly Zero Energy in existing buildings: energy saving and economic impact of design solutions in the housing stock of Bologna, Energ. Buildings 156 (2017) 327–342, https://doi.org/10.1016/j.enbuild.2017.09.044. [51] S. Fritz, M. Pehnt, P. Mellwig, J. Volt, Planned staged deep renovations as the main driver for a decarbonized European building stock, in: ECEEE Summer Study Proceedings, Vol. 7, 2019. [52] P. Femenías, K. Mj¨ ornell, L. Thuvander, Rethinking deep renovation: the perspective of rental housing in Sweden, J. Clean. Prod. 195 (2018) 1457–1467, https://doi.org/10.1016/j.jclepro.2017.12.282. C.M. Calama-Gonz´ alez et al. Energy & Buildings 336 (2025) 115629 16