Challenges implementing circular economy principles in the management of industrial saline effluents: from technology development to holistic assessment

INTERNATIONAL DOCTORAL SCHOOL OF THE USC Alba Roibás Rozas PhD Thesis Challenges implementing circular economy principles in the management of industrial saline effluents: from technology development to holistic assessment. Santiago de Compostela, 2022 Doctoral Programme in Chemical and Environmental Engineering TESE DE DOUTORAMENTO CHALLENGES IMPLEMENTING CIRCULAR ECONOMY PRINCIPLES IN THE MANAGEMENT OF INDUSTRIAL SALINE EFFLUENTS: FROM TECHNOLOGY DEVELOPMENT TO HOLISTIC ASSESSMENT Alba Roibás Rozas ESCOLA DE DOUTORAMENTO INTERNACIONAL DA UNIVERSIDADE DE SANTIAGO DE COMPOSTELA PROGRAMA DE DOUTORAMENTO EN ENXEÑARÍA QUÍMICA E AMBIENTAL SANTIAGO DE COMPOSTELA 2022 D./Dna. Alba Roibás Rozas Título da tese: Challenges implementing circular economy principles in the management of industrial saline effluents: from technology development to holistic assessment. Presento a miña tese, seguindo o procedemento axeitado ao Regulamento, e declaro que: 1) A tese abarca os resultados da elaboración do meu traballo. 2) De ser o caso, na tese faise referencia ás colaboracións que tivo este traballo. 3) Confirmo que a tese non incorre en ningún tipo de plaxio doutros autores nin de traballos presentados por min para a obtención doutros títulos. 4) A tese é a versión definitiva presentada para a súa defensa e coincide a versión impresa coa presentada en formato electrónico E comprométome a presentar o Compromiso Documental de Supervisión no caso de que o orixinal non estea na Escola. En Santiago de Compostela, 26 de Abril de 2022. Asinado: Alba Roibás Rozas AUTORIZACIÓN DAS DIRECTORAS DE TESE Challenges implementing circular economy principles in the management of industrial saline effluents: from technology development to holistic assessment D./Dna. Anuska Mosquera-Corral D./Dna. Almudena Hospido Quintana INFORMAN: Que a presente tese, correspóndese co traballo realizado por D/Dna. Alba Roibás Rozas, baixo a miña dirección/titorización, e a utorizo a súa presentación , considerando que reúne os r equisitos esixidos no R egulamento de Estudos de Doutoramento da USC, e que como director desta non incorre nas causas de abstención establecidas na Lei 40/2015. De acordo co indicado no Regulamento de Estudos de Doutoramento, declara tamén que a presente tese de doutoramento é idónea para ser defendida en base á modalidade de Monográfica con reproducción de publicaciones , nos que a participación da doutoranda foi decisiva para a súa elaboración e as publicacións se axustan ao Plan de Investigación. En Santiago de Compostela, 26 de Abril de 2022 Anuska Mosquera Corral Almudena Hospido Quintana I’ll tell you what freedom is to me. No fear. Priestess Nina Simone A mis padres, a quienes se lo debo todo. A mi mejor amiga, que es mi madre. A mi padre, la suerte de mi vida. SPI Sustainable Process Index SRT Solid Retention Time SSD Species Sensitivity Distribution SUP Single-use Plastics TAC Total Annual Costs TAETP Terrestrial Ecotoxicity TAG Triacylglycerides TC Total Carbon TEA Techno-Economic Analysis TN Total Nitrogen TOC Total Organic Carbon TS Total Solids TSS Total Suspended Solids TVS Total Volatile Solids VFA Volatile Fatty Acids VSS Volatile Suspended Solids WRA Waste Reduction Algorithm WWTP Wastewater Treatment Plant XF Exposure Factor SUMMARY i SUMMARY The fish canning industry is one of the most important sectors in Galicia, generating 85% of the total sector profits in Spain and about 10% of the world's production. It is estimated that, yearly, about 300,000 t of fish and shellfish are processed in this region, reaching an economic value of 1,25 billion €. But this activity also results in a high generation of effluents that need to be treated before being discharged. These effluents are characterized by a high content of organic matter, nutrients (like nitrogen), and salt. For this reason, fish canneries need to perform high investments in wastewater treatment facilities and effluent management. This is even more important in a context in which the sustainability of the sector is highly dependent on the water quality, as the raw material of this industry comes from the ocean. In recent years, political and governmental institutions, including the EU, are increasingly committed to an economic model that aims to maintain goods and materials in the value chains as long as possible. Here, waste is not only treated with the aim of being managed for dumping, but also for its recovery and use as a raw material. This model is known as circular economy, and it is clearly promoted from the European institutions by executing strategies like the Green Deal or the Circular Economy Action plan. The implementation of circular economy in a sector like the Galician fish canning industry would have special relevance, as this traditionally conservative industry applied moderate innovation patterns that hindered its development in the past. In this context, institutions like the National Association of Fish Canning Producers (by its acronym in Spanish, ANFACO) are advocating for the transformation of the approach applied in the sector, dealing with the specific problems of the industry (like the management of saline effluents). Problems like the climate and the plastic crises, and the COVID-19 pandemic, among others, are causing fluctuations in the petrol cost, and are raising the urgency of implementing new economic and environmental strategies that guarantee the sustainability of the global economic model and of the life system in the Planet. Therefore, the challenges currently faced are not only urgent, but also globally significant, and circular economy seems to be one of the most promising solutions to solve these issues. In this sense, the TREASURE-TECHNOSALTi project aims to provide solutions to increase the circularity applied in the fish canning sector. Here, the challenges of implementing circular economy in the fish canning industry (where saline waste streams are generated) were addressed, as TREASURE-TECHNOSALT targets to unravel the effects of salinity in wastewater treatment and valorisation. These challenges were not limited to technological development, but also to the methodological gaps linked to the environmental evaluation of effluent disposal (as saline i TREASURE – TECHNOSALT: Facing the treatment/recovery of saline wastewater to assure future water availability. CTQ2017-83225-C2-1-R. National Spanish Agency of Research. https://biogroup.usc.es/treasure ii streams might need different theoretical frameworks), and to the economic assessments, where the body of knowledge on resource recovery processes based on saline streams is still scarce. The unravelling of these challenges would aim to increase the readiness level of the technologies linked to application of the circular economy in the fish canning industry and could be extrapolated to other sectors. In this context, this thesis addressed these challenges, and aimed to transform wastes coming from the fish canning industry in value-added products using mixed microbial cultures (MMC). These value-added compounds are, mainly, Polyhydroxyalkanoates (PHA) and Triacylglycerides (TAG). They are completely biodegradable, and their production involves microbiological processes using waste streams (renewable sources), so their introduction in the value chains would fit the principles of circular economy in the context of resource recovery. PHA, a biopolymer that has the potential to replace conventional petrochemical plastics, is currently produced on an industrial scale, although such production is scarce and restricted to processes in which pure cultures of microorganisms are used (under sterile conditions). Moreover, this full-scale production is highly limited and circumscribed to some applications, like biomedical. TAG, a biomaterial produced by some microorganisms when they are in the presence of fats and oils, can be used to produce paints, varnishes, and biodiesel, among other applications. However, pilot-scale or full-scale applications are not even under development, and the current research works for TAG production only cover lab-scale. The industrial-scale production of these and other biomaterials has been hindered as the use of axenic cultures and pure substrates (like corn or glucose) significantly raises process costs. Moreover, it could have other side effects, like subliminal distress of food supply, as land would be used for materials generation instead of food production. The production of biopolymers from waste (which are unexpensive raw materials) and MMC (which do not need sterile conditions) have been a major advance to increase the feasibility of these processes on an industrial scale. In the last years, a lot of work has been performed regarding this production pathway for biomaterials (using MMC and waste streams). Recently, the use of saline streams as a feedstock for these processes has gained more interest, as they present some potential advantages. For example, Downstream Process (DSP) might have operational costs lower than those without saline conditions, due to the use of osmotic shocks to break the cell wall, and the decrease of probability for potential contamination by unwanted microbial strains. However, technical viability itself is not enough to guarantee the feasibility of introducing a new process or product in the value chains, and environmental and economic validations are necessary before a new good is stablished in the market. Life Cycle Assessment (LCA) has proven to be the tool for assessing the environmental impacts of different industrial processes. Particularly, it has been pointed out as the most accurate tool to validate resource recovery processes and emerging technologies, like those in which residual streams are valorised into biopolymers. LCA evaluates the environmental impacts of a product or service along its life cycle, i.e., the stages defining the existence of a process or good, from raw materials extraction to product waste management. LCA is being extended to cover other vectors also defining sustainability, like societal or economic dimensions. Here, the life cycle approach can be SUMMARY iii applied to financial aspects by Life Cycle Cost Assessment (LCCA), providing another powerful tool to perform integral evaluations of products and processes by combining different vectors of sustainability (not only environmental, but also economic). For systems based on emerging technologies, LCCA can be combined with other strategies for the economic assessment like the Techno-Economic Analysis (TEA), that specifically studies the performance of industrial processes, to provide a more robust framework to analyse a process’ viability. Therefore, the life cycle approach allows to analyse the performance of processes and products, assessing from a holistic perspective the consumption of resources and energy, the generation of waste and emissions, and expressing environmental burdens in a set of impact categories (such as climate change, eutrophication, human toxicity, etc.). However, there are still methodological gaps to be covered in LCA, as modelling all environmental interactions is a difficult task. For example, most LCA studies ignore the effect of salinity, although it is known that modifications in the salt concentration of ecosystems can cause significant environmental damage. This is important when circular economy implementation is being assessed in a sector like the fish canning industry, as effluents are generally saline, and the effects of wastewater disposal might need tools that consider this effect as well. Therefore, the assessment of salinity effects in LCA represents one of the current methodological gaps found that needs to be addressed in the field and, for this reason, TREASURE-TECHNOSALT addressed this issue. On the other hand, the challenges for process validation are not only linked to methodological gaps. Moreover, because there is no data on the production of PHA/TAG on an industrial scale using MMC, LCA studies assessing the production of some biomaterials are still scarce. Therefore, the environmental benefits generated by the production of these substances compared to the production of traditional materials have not yet been properly validated. Furthermore, as circular economy-based systems are multifunctional (waste is managed, and by-products are generated), validation goes in two directions. Then, the assessments need to include not only a comparison between conventional (petrochemical) and innovative (biobased) materials. Moreover, also circular economy-based effluent management techniques (including resource recovery) need to be compared to traditional linear economy-based waste treatment systems (where valorisation is not included). This scenario is appliable not only to LCA, but also to analogous life-cycle approach methodologies needed to obtain the integral validation of a technology, like LCCA. Here, although some efforts were made to evaluate the economic profile of PHA production, studies show wide ranges of results, and generally agree that economic validation is still not achieved. For TAG and other innovative biomaterials, the number of studies is highly limited, so it is even more difficult to analyse issues like the marketability or the current technology status. In this sense, the lack of full and pilot-scale data is also hindering the economic validation of PHA production and, this, the maturity level increase. Therefore, new case studies and economic assessments are needed to improve the current framework for PHA and biomaterials validation. For this purpose, integral studies where several dimensions of the process validation (like technological, environmental, and economic) are properly assessed and included are needed. Otherwise, the readiness level of the technology is stuck. iv The challenges faced by this thesis focus on solving some of the problems that appear in the management of fish canning industrial effluents, trying to find strategies to improve the degree of circularity applied in this sector. Because these streams have a high concentration of salt, especial attention has been paid to determine how the salinity affects these biological processes and how these features could be addressed in the field of environmental evaluation. Therefore, the issues addressed were framed in a context where circular economy needs to be implemented in different sectors and industries. Thus, a systematic approach to assess the readiness of an emerging technology based on resource recovery process was stablished. To do so, this thesis aims at i) studying and optimising the production of high valueadded storage substances (PHA) from the organic fraction of fish canning industrial effluents, using mixed (non-sterile) cultures of microorganisms, ii) assessing and identifying the current tools available to validate emerging technologies and try to fill some of the potential gaps found in the field, and iii) evaluating the technological solution through a life cycle approach, including both the environmental and the economic vectors. By doing so, the final goal is to provide a robust framework for technology readiness level increase in emerging technologies based on circular economy, with a special focus in the fish canning sector. In Chapter 1, the necessity of producing biomaterials is contextualized in the current plastic and climate crises. The state of the art of PHA production is analysedii, and an overview of the application of circular economy and valorisation strategies applied in the fish canning industry is also provided. Special attention is paid to technical, environmental, and economic challenges linked to the plastic crisis and PHA production. Finally, the objectives and structure of this thesis are explained. In Chapter 2, there is an overview of the materials and methods used. Considering the scope of this thesis, it is divided in two sections covering: i) the experimental protocols and analytical methods used to follow up the bioreactors operation, and ii) the evaluation methodologies used for the environmental and techno-economic assessment of the technology (i.e., LCA, which special focus on the Life Cycle Impact Assessment (LCIA) stage, LCCA, and TEA). Chapter 3 assesses the production of PHA using wastewater coming from the fish canning industry. To do so, saline Mussels Cooking Wastewater (MCW) was valorised to produce PHA with MMC in a three-stage system consisting on: 1) an acidification unit, where Chemical Oxygen Demand (COD) is fermented into Volatile Fatty Acids (VFA), 2) an enrichment unit, where the VFA-rich stream is used to feed a Sequencing Batch Reactor (SBR), and 3) an accumulation unit to maximise PHA storage in a Fed-Batch Reactor (FBR). One of the major findings was that the relatively high protein content in the wastewater hindered biopolymer production and MMC enrichment, as they are hardly fermented into VFA, and their presence in the SBR difficult the enrichment performance. Due to this relatively high protein content (1.8 – 5.7 g CODPROT/L), PHA accumulating capacity was initially below 10 % in the system, so several strategies were tested during the operational time to improve its transformation. ii Please note that even when TAG production was addressed by TREASURE-TECHNOSALT, its technology development is not covered by this thesis. SUMMARY v In the acidification unit, sodium bicarbonate (NaHCO3) was added to supply alkalinity, which was also low in the wastewater, to avoid the low pH and VFA production inhibition. This action increased protein conversion into VFA from 10.3% to 69.2%, and the subsequent PHA accumulation raise from 6.9 to 14.7%. In the enrichment unit, the strategy was to modify the Aerobic Dynamic Feeding (ADF) used to enrich the MMC by incorporating a settling stage after the feast phase. This provoked a shift in the proteins’ oxidation profile, as their consumption switched from the feast to the famine phase of the cycle, where the nitrogen released was used by the MMC for growth during the starvation period. This increased the biomass concentration, so the tolerated COD by the system was also higher (from 1.6 to 4.2 g VSS/L and from 2.2 to 4.4 g COD/L, respectively). Finally, increasing the proteins/VFA ratio for MMC acclimation to proteins was done by modifying the wastewater proportions in the SBR feeding. This modification allowed to increase PHA accumulation from 8.8 to 41.5 % in the FBR, overcoming the proposed threshold for industrial viability of 40%. Chapter 4 is a review on the methodological choices applied to perform LCA of PHA production. As already said, the fact that a material is biodegradable or renewable, does not automatically guarantee environmental benefits linked to its production, so a validation through LCA is needed. It was found that the few LCA studies for PHA production using MMC obtained contradictory results, mainly due to the lack of actual data, but also to the absence of a consensus methodological framework. The review covered the research works published in the last 20 years, hoping to find the methodological gaps hampering the environmental validation of the process. Consequently, it focuses on key methodological characteristics of any LCA, such as the functional unit and the allocation choices, but also on the data sources of the existing studies, to have a clear picture of the present knowledge related to the environmental profile of biomaterials like PHA. Besides some methodological gaps found; the main output of this study is that most of the LCA performed till date on PHA production do not use primary data. Contrarily, they usually employ data from references based on studies performed 20 years ago. Therefore, LCA practitioners need to stop relying on data generated decades ago to evaluate nowadays processes, and environmental assessments based on primary information need to be urgently performed. Chapter 5 aims at covering one of the main outputs of Chapter 4, the need of studies using primary information. Therefore, the results from a two-year laboratory-scale operation for PHA production using MCW and MMC (Chapter 3) were scaled-up to define and compare a circular economy scenario with the current linear approach (i.e., effluent generation, treatment, and discharge). Here, the multifunctionality of the PHA production system is addressed, as the replacement is not only referred to the material (petrochemical plastic vs biopolymer), but also to the waste management system. Results show an average improvement of ca. 25% for nine out of ten impact categories evaluated if the circular economy approach is implemented, where the only category performing worse (ozone depletion) is linked to the DSP. The sludge management strategy was a key factor for the environmental validation of the process, and if composting is applied instead of anaerobic digestion, the improvement is lost in the fossil depletion impact category but maintained in the remaining 8 categories. vi Moreover, when a conservative replacement yield of fossil-based plastic was tested, the circular economy approach was the preferable option in 8 out of 10 categories (in this case, also fossil depletion performed worse, as less petrochemical plastic is avoided). The significance of the DSP was also confirmed by this study, although it was not a barrier to show the feasibility of producing value-added bioproducts under a circular economy approach. Accordingly, Chapter 6 addresses the integration of techno-economic and environmental evaluations to perform a holistic assessment that aims to contribute to the readiness level of the technology. To do so and considering the framework of the TREASURE-TECHNOSALT project, industrial effluents (wastewater and oil) from the fish canning industry were valorised to produce PHA and TAG using MMC at benchscale. Once validated at lab scale, those process systems were up-scaled to perform a techno-economic and environmental analysis and to compare them with the treatments currently applied in the industry. Here, especial attention was paid to the effects of process scale and (des)centralization. In this sense, one of the major findings was that centralized extraction was an attractive alternative when DSP on-site was not technically feasible due to the process scale. Under this approach, wastewater treatment plants (WWTP) would act as PHA-rich biomass suppliers where a centralized biorefinery approach would oversee the DSP of the solids generated in the factories. For system that used wastewater as a feedstock, effluent management cost would decrease from 0.53 €/kg COD removed when wastewater is only treated, to 0.48 €/kg COD when wastewater is valorised if the extraction process is not included. This is mainly due to the replacing of activated sludge reactors, with residence times of several days for industrial wastewater, by sequential ones, which have a lower volume due to their shorter residence times of 24 hours (so lower direct and indirect costs). If several of these industrial WWTPs provided biomass for centralized biomaterials extraction, PHA production cost could be 0.95 – 1.18 €/kg PHA. When oil is separated from the WWTP influent and used to produce biomaterials, wastewater treatment cost decreases from 0.52 to 0.30 – 0.33 €/kg COD if DSP is not included. However, the cost of biomaterials production (thus, the extraction process performance) increases with respect to the systems that used wastewater instead of oil to produce biomaterials (1.56 – 3.35 €/kg PHA, and 6.75 – 9.21 €/kg TAG) due to the low biomass production in the biological reactor fed with oil. Finally, the environmental validation of the processes is linked to the operational approach selected and the wise use of chemicals, as well as to the use of low-cost extraction processes to produce low grade PHA. As mentioned above, Chapter 5 aimed to solve some of the general gaps linked to PHA environmental validation. However, other research necessities were raised in Chapter 4 when particularly assessing the use of saline streams in industrial processes, as it was found that there was no methodological framework to assess the effects of salinity variations in aquatic environments. Chapter 7 aims at covering this gap by addressing the issue of salinity variations in aquatic environments in LCA. Variations in the salt concentration (both derived directly from anthropogenic activities, like irrigation, treated wastewater discharge or dam management, but also indirectly from climate change) do impact on ecosystems, but there was no methodology to quantify or describe those impacts. SUMMARY vii Salinity is changing in aquatic systems and, although there are studies on the effects of salinity variations on individual species, little is known about the effects on overall ecosystems. Moreover, these impacts could be more uncertain in transitional waters such as estuaries or fiords, and, particularly, in the Galician rías or estuaries. Complementary to the current models used for ecotoxicity assessment, where an increase in the concentration of a pollutant generates an increase in the impacts (thus, disregarding the effects of water freshening), this thesis introduces a general framework to address the impacts of salinity variations, including emission-related positive effects. To do so, both negative and positive impacts linked to a decrease in the concentration of a chemical (in this case, sodium chloride) were modelled for the first time. The framework developed was validated by applying it to the Galician ría of Arousa, where sharp drops in the salt concentration have caused mass mortalities of shellfish in recent decades. The characterization factors developed ranged 0 – 0.89, in units of Potential Disappeared Fraction of species (PDF)‧m3‧month/kg. Chapter 8 summaries the main knowledge contributions of this thesis, .i.e. the provision of a systematic approach that can be used to increase technology readiness level for resource recovery emerging processes based on circular economy, with special emphasis to the ones employing saline effluents and/or streams coming from the fish canning industry. Firstly, a process for PHA production using saline waste streams and MMC was developed and validated, both technically and environmentally. Then, a state-of-the-art overview of the methodological framework available for this evaluation was provided, and part of the research work developed after that aimed to contribute to some of the nowadays general necessities found after this review, to provide general validation for PHA production systems. Therefore, an economic assessment is provided after the environmental validation. Here, it was proven that WWTPs could work as raw materials providers for centralised biorefinery schemes. The scope of this evaluation was extended by including other research works in the frame of TREASURE-TECHNOSALT for TAG production, validating the centralized biorefinery approach for more resource recovery systems. This thesis proved the feasibility of implementing circular economy-based approaches and processes in the Galician fish canning industry and provided a framework to assist the increase of the readiness level of emerging technologies, as explained in Chapter 1. Moreover, it proved the usefulness of life thinking strategies to holistically validate the feasibility of developing processes based on circular economy, like the resource recovery ones. Finally, this thesis paid special attention to the particular challenges that need to be addressed when working with saline streams. Therefore, the results obtained led to the generation of a new methodology to assess the impacts of salinity variations in aquatic environments, including both negative and positive effects in emission-related impact assessment for the first time, and providing a new field of discussion in the LCA community. viii RESUMO xv proporcionasen biomasa para a extracción centralizada de biomateriais, o custo de produción de PHA podería ser de 0,95 a 1,18 €/kg de PHA. Cando as graxas e os aceites son separados do influente que alimenta outra EDAR industrial para producir TAG e PHA, o custo de tratamento de augas residuais diminúe de 0,52 a 0,30 – 0.33 €/kg DQO, pero o custo de produción destes biomateriais aumenta con respecto aos sistemas que empregaban augas residuais sen aceite coma materia prima (1,56 – 3,35 €/kg de PHA, e 6,75 – 9,21 €/kg TAG) debido á baixa produción de biomasa no reactor biolóxico alimentado con aceite. Finalmente, a validación ambiental dos procesos está ligada ao enfoque operativo seleccionado e ao uso moderado de produtos químicos, así como á implantación de procesos de extracción de baixo custo para producir PHA de baixo grao (usado, por exemplo, para embalaxes). Por último, nesta liña, o capítulo 5 tivera como obxectivo resolver algunhas das lagoas xerais relacionadas coa validación ambiental PHA. Non obstante, detectáronse outras necesidades de investigación á hora de valorar particularmente o uso de correntes salinas nos procesos industriais, xa que se comprobou que non existía un marco metodolóxico para avaliar os efectos das variacións de salinidade nos medios acuáticos. O Capítulo 7 pretende por tanto cubrir esta lagoa abordando o problema da salinidade no ACV. As variacións na concentración de sal (derivadas directamente de actividades antropoxénicas, como o rego ou a xestión de encoros, pero tamén indirectamente polo cambio climático) repercuten nos ecosistemas, pero ate o momento non había unha metodoloxía para cuantificar ou describir eses impactos. A salinidade está cambiando nos sistemas acuáticos e, aínda que existen estudos sobre os efectos das variacións da salinidade en especies individuais, pouco se sabe sobre os impactos sobre os ecosistemas globais. Ademais, estes efectos poderían ser máis incertos en augas de transición como rías ou fiordos e, en particular, nas rías galegas. Complementariamente aos modelos actuais utilizados para a avaliación da ecotoxicidade, onde un aumento da concentración dun contaminante xera un aumento dos impactos (polo tanto, sen ter en conta os efectos da dulcificación da auga), esta tese introduce un marco xeral para abordar os impactos das variacións da salinidade, incluíndo efectos positivos relacionados cos impactos por emisións (tradicionalmente só negativos). Para iso, modeláronse por primeira vez tanto impactos negativos como positivos vinculados á diminución da concentración dun produto químico (neste caso, cloruro de sodio), introducindo así este novo marco. O devandito marco desenvolvido validouse aplicándoo á ría galega de Arousa, onde fortes descensos da concentración de sal provocaron mortalidades masivas de marisco nas últimas décadas. Os factores de caracterización estimados neste capítulo oscilaron entre 0 e 0,89, en unidades de fracción de especies potencialmente desaparecidas‧m3‧ mes/kg. O Capítulo 8 ofrece as principais achegas ao coñecemento desta tese. Polo tanto, co traballo aquí realizado, ofrécese un enfoque sistemático que pode ser utilizado para aumentar o nivel de madurez tecnolóxica dos procesos emerxentes de recuperación de recursos baseados na economía circular, con especial énfase nos que empregan efluentes salinos e/ou correntes procedentes da industria conserveira de peixe. xvi En primeiro lugar, desenvolveuse e validouse un proceso para a produción de PHA utilizando fluxos de residuos salinos e CMM, tanto a nivel técnico coma a nivel ambiental. A continuación, presentouse unha visión xeral do estado da arte do marco metodolóxico dispoñible para esta avaliación, e parte do traballo desenvolvido posteriormente tivo como obxectivo contribuír a algunhas das necesidades xerais que se atoparon tras esta revisión, coa fin de proporcionar unha validación xeral para sistemas de produción de PHA. Polo tanto, nesta tese realizouse unha avaliación económica despois da validación ambiental. Aquí, comprobouse que as EDAR poderían funcionar como provedores de materias primas para esquemas de biorefinería centralizada. O alcance desta avaliación ampliouse incluíndo outros traballos de investigación no marco de TREASURETECHNOSALT para a produción de TAG, validando o enfoque de biorefinería centralizada para máis sistemas de recuperación de recursos. Esta tese demostrou a viabilidade de implantar enfoques e procesos baseados na economía circular na industria conserveira de peixe galega, e proporcionou un marco para aumentar o nivel de madurez das tecnoloxías emerxentes. Ademais, demostrou a utilidade das estratexias de pensamento de ciclo de vida para validar holísticamente a viabilidade de desenvolver procesos baseados na economía circular, como os de recuperación de recursos. Finalmente, esta tese prestou especial atención aos retos particulares que se deben abordar cando se traballa con correntes salinas. Polo tanto, os resultados obtidos levaron á xeración dunha nova metodoloxía para avaliar os impactos das variacións da salinidade nos medios acuáticos, incluíndo por primeira vez os efectos tanto negativos como positivos na avaliación dos impactos relacionado efectos de emisións, e proporcionando un novo campo de discusión na comunidade do ACV. CHAPTER 1: INTRODUCTION 1 1. INTRODUCTION SUMMARY This chapter provides an overview of the motivation of the thesis and its scope. Here, the necessity of producing bio-based materials like Polyhydroxyalkanoates (PHA) in the framework of circular economy arises, contextualized in the current climatic and plastic crises. The current state-of-the-art of biomaterials production (mainly PHA, but also triacylglycerydes) is assessed, and the status of PHA environmental and techno-economic evaluations, which are the base for technology maturity increase, is assessed. Finally, an overview of the thesis structure and how that is related to technology development, is provided. Part of the content of this chapter was published as A. Roibás-Rozas1, M. Saavedra del Oso1, G. Zarroli1,2, M. Mauricio-Iglesias1, A. Mosquera-Corral1, S. Fiore2, and A. Hospido1 (2022). How can we validate the environmental profile of bioplastics? Towards the introduction of Polyhydroxyalkanoates (PHA) in the value-chains. Chapter 20 in: C. Teodosiu, S. Fiore, A. Hospido (Eds.), Assessing Progress Towards Sustainability, Elsevier. https://www.elsevier.com/books/assessing-progress-towardssustainability/teodosiu/978-0-323-85851-9 a CRETUS, Department of Chemical Engineering, Universidade de Santiago de Compostela, 15782 Santiago de Compostela, Spain b DIATI (Department of Engineering for Environment, Land and Infrastructures), Politecnico di Torino, 10129 Torino, Italy ALBA ROIBÁS ROZAS 2 1.1 CHALLENGES IN THE CONTEXT OF A GLOBAL CRISIS The plastic and the climate change crises are two of the most relevant problems that need to be addressed by the humankind to avoid a global collapse. Although they have been treated in the past as separated issues, the truth is that they are fundamentally linked. Primarily, plastics contribute to greenhouse gas (GHG) emission in all the stages of its life cycle (from raw materials extraction to manufacturing, transport, and end of life (EoL)). Secondly, both issues are narrowly related with the oceans and the marine environment. Finally, the root cause of both crises is the same: the overconsumption of finite resources and a bad EoL management. In this sense, engagement solving plastic pollution can increase action against climate change, where integrated solutions can be summed up with strategies like the implementation of a circular economy (Ford et al., 2022). Plastic goods have been in our lives for no longer than one century. However, in about seventy years, they have become one of the greatest concerns and environmental threats worldwide. The growth of plastics market during the twentieth century was remarkably fast, only surpassed by some construction goods, as steel and cement. Plastics’ production increased from 2 Mt in 1950 to 380 Mt in 2015, yielding in about 83,000 Mt of total produced virgin plastics in that period (Geyer et al., 2017). Moreover, the plastic sector employs about 1.5 M people in the European Union (EU) and generates around 340 billion € yearly (European Commission (EC), 2018a). However, its production and use are linked to the generation of extremely high amounts of wastes. It is estimated that, of all the plastics ever produced, only about the 30% of them still remain in the value chains, so the remaining 70% of them were wasted. Of this enormous amount of waste, only 9% is recycled, 12% is incinerated, and 79% is accumulated in landfills or directly disposed in the natural environment with no treatment at all (Geyer et al., 2017). The related impacts on ecosystems, health and economy cannot be quantified yet due to the lack of consistent frameworks and actual data (Boulay et al., 2021). However, the number of studies on this topic is growing very fast, and their results show that these effects are serious and global (Halsband and Herzke, 2019; Saling et al., 2020). In fact, plastics are slowly decomposing and leaking in the environment. First, they affect terrestrial ecosystems (de Souza Machado et al., 2018), from where they are transferred into freshwater (Wagner et al., 2014). And finally, plastics reach the oceans in the forms of macro, micro and nano plastics (Galloway et al., 2017; Geyer et al., 2017; Halsband and Herzke, 2019; Saling et al., 2020). Modern plastics are complex mixtures of polymers, residual monomers, and chemical additives, some of them being endocrine disruptors that alter metabolic and reproductive patterns (Galloway et al., 2017). Plastic debris accumulate in the environment and are transferred to many trophic levels through food chains, thereby affecting animal and human health (Akhbarizadeh et al., 2020). Moreover, unmanaged plastic waste also affects the economy, seriously impacting fishing, aquaculture, and tourism sectors, and it implies direct costs for governments and administrations. As an example, the Netherlands spend millions of euros each year on removing litter from beaches and coastal areas (Conejo-Watt and Luisetti, 2019). Micro and nano plastics are vehicles that transport potentially hazardous and toxic substances, like heavy metals or organic pollutants. Moreover, these carried pollutants interact sometimes with the carriers (the micro and nano plastics), boosting the harmful effects of each pollutant by itself (Liu et al., 2022; Xiang et al., 2022). Humans uptake CHAPTER 1: INTRODUCTION 3 plastic by several exposure pathways, mainly: drinking water, the consumption of food products, and air (inhalation of dust), potentially generating pulmonary damage, oxidative stress, cellular damage, DNA damage, and inflammatory and immune reactions, among other effects (Xiang et al., 2022). Moreover, the COVID-19 pandemic crisis resulted in an unprecedent surge of production, consumption, and disposal of single-use plastics (SUP) due to the necessity of using disposable personal protective equipment (PPE), like face masks. This not only increased the amount of plastic pollution, but also increased health risks due to mismanaged potentially infectious materials disposed in urban and natural environments (Ammendolia and Walker, 2022). In fact, contaminated PPE pollution is considered a new type of environmental hazard, linked to the increase of plastic pollution (they are a new source of micro and nano plastics), and to their status of potential carriers for harmful contaminants (Kutralam-Muniasamy et al., 2022). Unfortunately, the introduction of plastics in the trophic chains has already been confirmed extensively. High amounts of microplastics were found in some of the most fished and consumed species (Masiá et al., 2022). About 80% of the cans examined in a study containing tuna and mackerel had at least one microplastic that is going to be consumed by humans (Akhbarizadeh et al., 2020). Consequently, it is estimated that a person intake between several tens of thousands to several millions of microplastic particles annually, where the main source of exposure might be the ingestion of water bottled in plastic. Here, factors like size, chemical composition, or hydrophobicity, would shape the type and magnitude of the effects, ranging from cytotoxicity to hormone disruption, among other effects (Kannan and Vimalkumar, 2021). Regrettably, the presence of plastic particles in humans has already been confirmed, as they were found in all tested samples of liver and adipose tissue (ACS Newsroom, 2020) and human stool (Schwabl et al., 2019). Results were linked with strong evidences of relationship between the presence of these particles in human tissues and harmful effects on the lungs and the immune system (Zarus et al., 2021). Moreover, maternal transfer was already proved by finding pieces of plastic in human placentas (Ragusa et al., 2021). Therefore, (micro)plastic is everywhere (Lim, 2021). In this sense, the amount of plastic accumulated in the ocean is uncertain, and recent studies are pointing out that the current measurements are in fact underestimating the magnitude of the problem (Way et al., 2022). Nevertheless, some studies estimate discharges of about 6.4 million tons of plastic debris yearly, where about 80% of this debris originates from land (remaining 20% correspond to ship or military operations). Here, the biggest contributor to this plastic release is China (Jambeck et al., 2015). However, this is not only due to waste mismanagement in this country, but to the global plastic waste trade networks. In fact, the main sources of plastic waste are found in Europe and the United States (Wang et al., 2020), as rich states sell their waste to poorer and/or developing countries. As the importer countries use these wastes as raw materials to recycle and convert them into new products, these world trades were linked with reductions in GHG emissions (Liu et al., 2021). However, other effects, like the ones linked with marine debris and human toxicity, can only increase under this trade model that does not include reduction. ALBA ROIBÁS ROZAS 4 Moreover, China recently imposed an import ban on plastic waste, generating a shift in the trade directions to poorer countries and creating a new discussion about waste management in developed countries (Wang et al., 2020). As the global peak of waste is not supposed to be reached until at least 2100 due to increased population and consumption patterns, long-term solutions need to be designed. Here, developed and rich countries have a huge responsibility, specially towards some Asian and African countries, were the harmful effects of the traded plastic from rich regions are being substantial (Browning et al., 2021; Jambeck et al., 2015). As seen, the issue of plastic pollution is not only an environmental and public health one, but also an economic, social, and political problem, and its consequences can be as catastrophic as uncertain. Therefore, when searching for solutions to the plastic/climate crisis, several vectors need to be considered, mostly in a bioeconomic framework. Here, integral solutions can be provided by the application of circular economy approaches. 1.2 ALTERNATIVE BIO-BASED PLASTICS AND BIOMATERIALS Around 50% of the produced plastics are SUP (Galloway et al., 2017), which constitute the 49% of the marine litter in the EU (EC, 2018b). Moreover, it is estimated that about 95% of the economic value of plastic packaging materials is lost due to the single-use life cycles (EC, 2018a). Therefore, the EC recently addressed this issue by publishing a directive on the reduction of the impact of plastics on the environment (EC, 2019b), and by generating a new legal framework in the European Green Deal (EC, 2019) and the new Circular Economy Action Plan for a cleaner and more competitive Europe (EC, 2020); including actions to develop a framework on sourcing, labelling and use of bio-based, biodegradable and/or compostable plastics. While biodegradable plastics can have petrochemical origin, bioplastics include a broad number of renewable, biodegradable and bio-based polymers that are expected to substitute conventional plastics, as they have similar physicochemical, thermal and mechanical properties (Mannina et al., 2020). Nowadays, the annual production of biobased and/or biodegradable plastics is less than 1% of the global plastic production (Changwichan et al., 2018), being polyhydroxyalkanoates (PHA), polylactic acid (PLA) and polybutylene succinate (PBS) the biomaterials with more potential in the market. The three of them are produced involving microbiological processes. For PLA and PBS, the building blocks of the polymer (the monomers, lactic, succinic acid and 1,4-butanediol, respectively) are produced by the activity of microorganisms, but monomer polymerization is not a natural process. However, for PHA, polymerization occurs naturally by the microorganisms, so the production is totally biological. Recently, PHA is gaining more attention, but its sustainability needs to be ensured before entering the value chains. Therefore, the replacement of petrochemical materials by biomaterials could be very promising to solve the climate crisis and stop relying on oil. This is the main motivation for PHA production, but also for the generation of other bioproducts, like triacylglycerydes (TAG), which are lipids that can be used as a feedstock for biodiesel production (Pinto-ibieta et al., 2021). TAG can be stored by microorganisms as intracellular granules and could replace petrochemical materials in a broad number of applications besides biodiesel, such as the pharmaceutical, cosmetical or food-processing industry (Argiz et al., 2020b; Tamis et al., 2015). However, not much is currently known about its production using biotechnology, CHAPTER 1: INTRODUCTION 5 so other research works in the frame of TREASURE-TECHNOSALT1 project aimed to produce it through the valorisation of waste streams. Although this introduction includes TAG, as its production process was one of the assessed ones in this thesis, this Chapter will focus on PHA, as it was the main biomaterial produced and evaluated in the present research work. In summary, it is clear that, in the transition from linear to circular economy where value-added products from waste are generated to replace petrochemical materials, the strong commitment for bioplastics is one of the main targets of institutions like the EU (Di Bartolo et al., 2021). 1.2.1. Technological development in PHA production and mixed microbial cultures PHAs are produced by some microorganisms subjected to stress conditions. If feedstock and nutrients are available transiently, some microbes respond by accumulating extra carbon that can be used in the case of feeding deprivation (Valentino et al., 2017). That extra carbon is stored inside the cells in the form of polyesters, which, after extraction, can be employed as base materials to replace petrochemical plastics. Therefore, in PHA production processes, carbon-rich feedstocks are transformed into value-added products by microorganisms, so they are an opportunity to fit the circular economy action plans by EU. First studies on PHA production were based on pure substrates, mostly coming from dedicated crops such as sugar cane or palm oil, to feed pure cultures of microorganisms consisting in one single strain of bacteria able to store PHA. Under this circumstance, accumulating capacities are usually very high, reaching up to 90% of the cell content, with biomass concentrations that can be of 10 – 50 g/L (Marciniak and MozejkoCiesielska, 2021). However, the associated costs were high, reaching 45% of the total expenses due to high purity feedstocks use (Kourmentza et al., 2017). Besides, pure culture processes require sterile conditions, which implies high energy costs for reactor sterilisation (Bengtsson et al., 2017b). Finally, the use of dedicated crops involves an ethic conflict, as a global production of bioplastic based on agricultural feedstocks could alter the word’s food supply chain (Sabapathy et al., 2020). Research efforts are currently oriented to replace pure feedstocks by unexpensive substrates, such as waste streams. They are carbon-rich, and sometimes nitrogen and nutrient-rich, so they can also be used to feed the microorganisms in charge of PHA production. Moreover, waste streams are not competing at all with the food-chain supplies or land use, and they are environmentally burden-free (Yadav et al., 2020). Accordingly, in the last decade, a wide amount of waste effluents were tested to produce PHA: cheese whey, sewage sludge, oil mill wastewater, glycerol, municipal wastewater, food processing effluents, molasses, paper mill wastewater, hardwood liquor, or Kraft mill effluents are just examples of the used streams reported in literature (Sabapathy et al., 2020). Moreover, the feasibility of using saline effluents has been also recently assessed (Argiz et al., 2020a; Pedrouso et al., 2020). 1 TREASURE – TECHNOSALT: Facing the treatment/recovery of saline wastewater to assure future water availability. CTQ2017-83225-C2-1-R. From 2017 to 2021. National Spanish Agency of Research. https://biogroup.usc.es/treasure. ALBA ROIBÁS ROZAS 6 When waste streams are used for PHA production, pure cultures are often replaced by mixed microbial cultures (MMC), where a high diversity of microorganisms are present. Consequently, a selection strategy is required to favour the growth of the desired strains and disfavour the thrive of the non-storing populations of microorganisms (Kourmentza et al., 2017). In any case, On the other hand, although some processes using waste streams and MMC reached accumulations of 70 – 80%, most of the systems result in storages of 20 – 50% with biomass concentrations significantly lower than the ones of pure cultures (Mannina et al., 2020; Sabapathy et al., 2020). However, for axenic-based PHA productions, the benefits of high biopolymer production might be buffered by the costs of sterilisation, so MMC-based productions represent a big opportunity. Therefore, although full-scale PHA production is currently dominated by pure culture systems (Sabapathy et al., 2020), which are generally crop-based (Yates and Barlow, 2013), the pilot-scale projects reported in literature and operated for PHA production are nowadays working with MMCs (Rodriguez-Perez et al., 2018), showing a switch trend in the culture type used. As seen in Figure 1.1, this production pathway is generally performed in a three stage system (Figure 1.1) (Valentino et al., 2017): i) The first step is a pre-treatment where the organic sources present in the feedstock (like carbohydrates or proteins) are transformed into suitable forms of carbon for PHA production through anaerobic fermentation (i.e. volatile fatty acids (VFA)), which will be naturally esterified and accumulated by bacteria (Kosseva and Rusbandi, 2018)); ii) Once the feedstock is fermented, the VFA-rich stream will be fed transiently to an open MMC that will be enriched in microorganisms with the capacity of accumulating PHA. This step is generally performed in a Sequencing Batch Reactor (SBR). iii) The outgoing stream, rich in storing biomass, will be fed with the same VFArich stream where the MMC accumulating capacity will be maximized (Sabapathy et al., 2020). This step is generally performed in a Fed-Batch Reactor (FBR) and generates PHA-rich biomass. Figure 1.1. Flowchart of a typical MMC-based PHA production process Depending on the substrate characteristics, this three-stage strategy has been also used to produce other biomaterials, like TAG (Fra-Vázquez et al., 2018; Tamis et al., 2015), achieving high accumulating capacities around 80% (Argiz et al., 2020b). However, when the feedstock employed is oily, there are other process features that need to be considered. For example, substrate concentration needs to be usually low, as oil normally presents high concentrations of Chemical Oxygen Demand (COD) per gram of substrate. This naturally leads to low biomass productions, so the productivity of the process is low although PHA accumulation is high. Nevertheless, this type of substrates also presents some advantages. As oily streams are composed by long chain fatty acids, the TAG production process can skip the acidification step by performing fatty acid CHAPTER 1: INTRODUCTION 7 hydrolysis and MMC enrichment in the same reactor (Argiz et al., 2021, 2020b), providing new opportunities for the generation of this biomaterial In any case, after microorganisms transformed the organic matter present in the waste stream into PHA or TAG, it is necessary to extract the stored material from the microbial cells (Figure 1.1). So, the Downstream Process (DSP) involves the activities required to recover and purify the PHA (or any accumulated biopolymer) from the biomass and make it available for its transformation into useful materials (i.e., compounding and shaping, see Figure 1.2). For PHA, studies still allocate 30 – 50% of the total production costs to DSP (Colombo et al., 2020), being, economically and environmentally, the most expensive stage of the whole MMC-based production process (Fernández-Dacosta et al., 2015). However, room for environmental and economic improvements are also present for low cost processes aiming to generate low grade PHA (Saavedra del Oso et al., 2020). Nowadays, the existing DSPs are divided in two categories: the ones that separate biomass from PHA by dissolving the polymers, and the ones that dissolve or mechanically disrupt the cellular material to release the PHA (Jiang et al., 2018; Kosseva and Rusbandi, 2018; Yadav et al., 2020) (Figure 1.2). Generally, the process starts with the concentration of the PHA-rich biomass (Pérez-Rivero et al., 2019). Then, different treatments are used to weaken the cell membrane before it is disrupted. Although the most applied pretreatment involves freeze-drying the biomass (Rodriguez-Perez et al., 2018), it presents technical and economic difficulties for its full-scale application, so other techniques, like thermal dehydration (Kourmentza et al., 2017) or osmotic shocks (Koller et al., 2013a; Kourmentza et al., 2017; Rathi et al., 2013) have been researched. Once the cell is weakened or broken up, the methods directly dissolving the biopolymeric material usually employ solvents that change the permeability of the cell membrane. It provides a high-quality final product, but the use of chlorinated solvents is economically and environmentally detrimental (Fernández-Dacosta et al., 2015; Saavedra del Oso et al., 2020). The alternative to PHA direct solving is cell digestion, which is normally chemical: acids, alkalis, or, recently, surfactants (Fernández-Dacosta et al., 2015; Mannina et al., 2019; Rathi et al., 2013). Here, the sodium dodecyl sulphate (SDS) is the most employed one (Pérez-Rivero et al., 2019). This detergent enters the lipid membrane and increases the volume of the cell envelope until it breaks down, so the PHA, is released. This detergent enters the lipid membrane and increases the volume of the cell envelope until it breaks down, so the PHA is released. This method obtains good results when combined with other treatments, like alkali addition (Fernández-Dacosta et al., 2015; Rathi et al., 2013), so at present surfactant treatment is, economically and environmentally, the most promising alternative for low-grade PHA extraction (Fernández-Dacosta et al., 2015; Saavedra del Oso et al., 2020). In fact, the costs of solvent-based PHA extraction, normally aiming to produce high grade biopolymer to be used in applications like pharmaceutical, are estimated to range 0.25 – 2.25 €/kg PHA. On the other hand the cost of PHA obtained by cell digestion/mechanical disruption, which produces low grade biopolymer to be used in low value applications like packaging costs around 0.25 – 1.50 €/kg PHA (Saavedra del Oso et al., 2020). Once the polymer is separated, its final uses are multiple and depend on its properties (Roohi et al., 2018). These vary with the feedstock composition and the extraction method ALBA ROIBÁS ROZAS 8 (Bengtsson et al., 2017b; Melendez-Rodriguez et al., 2018), as monomer distribution and chain length are two of the main issues defining these final properties (Raza et al., 2018). Although much is still to be known until the tailoring of the PHA materials is perfectly understood, it seems likely that polymers only composed by butyrate monomers (polyhydroxybutirate (PHB)) can substitute some polypropylene (PP) and polyethylene terephthalate (PET) applications (Lamberti et al., 2020; Melendez-Rodriguez et al., 2018; Roohi et al., 2018). Moreover, the presence of valerate monomers (so polyhydroxyvalerate or PHV) can also affect the final properties of the polyester, modifying the crystallinity or the melting temperature in comparison to polymers composed only by PHB (T Palmeiro-Sánchez et al., 2016). Common PHA manufacturing techniques are moulding and injection (Bengtsson et al., 2017b; Changwichan et al., 2018) as well as electrospinning (Melendez-Rodriguez et al., 2018) (Figure 1.2). Besides, it has been stated that the combination of PHA and PLA fibres can produce materials with improved final properties (Bengtsson et al., 2017b; Lamberti et al., 2020), and composites formed by PHA, PLA and PBS reinforced with non-woven flax fibres achieve the necessary requirements to be employed in the car industry (Pantaloni et al., 2020). Finally, poor-grade PHA (that did not achieve the market specifications) or wasted (after several use cycles) PHA films can be further valorised in feedstock recycling to produce crotonic acid and its oligomers (Figure 1.2), which are used as building blocks in the production of some chemicals, like paints or adhesives (Fernández-Dacosta et al., 2016; Lamberti et al., 2020). In any case, after consumer use, these bio-based materials need to be properly disposed and managed at their EoL (Figure 1.2). Here, the waste management options are landfilling, incineration, anaerobic digestion, composting and recycling (Yates and Barlow, 2013). Although PHAs are biodegradable, their degradation is complex and hard to model, as it is dependent of many factors (Laycock et al., 2017). Figure 1.2. Life cycle of a biopolymer in the frame of circular economy CHAPTER 1: INTRODUCTION 15 3. Validate the environmental profile of the designed process and of the generated product by LCA, acknowledging the considerations extracted from the review (Chapter 5). 4. Prove the economic validation of the proposed systems, ensuring that the economic vector is acknowledged when technologies are evaluated in a bioeconomic framework (Chapter 6). 5. Fill some of the methodological gaps found when environmentally assessing bioprocesses and biomaterials, particularly focusing on the ones using saline effluents. In this case, a new methodology to assess the effects of salinity variations in aquatic environments was proposed (Chapter 7). As seen, this thesis aims to contribute to the readiness level increase of emerging technologies based on circular economy by providing a systematic structure to validate the performance of the assessed technology. Therefore, the methodology shapes the structure of the thesis itself (Figure 1.3): Chapter 1 and Chapter 2 provide an introduction and a methodological overview of this work: - Chapter 1. provides the hypothesis to be tested in this document. Here, the necessity of generating new biomaterials in the context of the plastic and environmental crises is exposed. The concept to be tested (a MMC-based PHA production system fed by MCW) is presented. - Chapter 2 describes the methods used to perform the current research work: experimental protocols for technical validation and assessment methodologies for the environmental and economic validation. Therefore, the tool for the design of the MMCbased system and this methodology itself are presented. Then, Chapters from 3 to 7 address the results obtained. - Chapter 3. First, the concept presented in Chapter 1 (PHA production process under saline conditions) was tested by subjecting it to technical feasibility validation. Therefore, the system of bioreactors was first designed. Then, the design was put into practice by the laboratory-scale operation of about two years. - Chapter 4. Here, the challenges faced to evaluate the environmental profile of MMC-based processes were assessed in a methodological state of the art review. Therefore, the research works addressing this issue were critically analysed hoping to stablish a robust methodological framework to perform the environmental validation of the studied process. - Chapter 5. Once identified the existing tools to evaluate the environmental profile of the developed process, the data generated in Chapter 3 was used to perform an environmental evaluation of the proposed MMC-based PHA production process through a case study. This chapter aims to contribute to fill one of the general gaps found in Chapter 4: the lack of data for environmental verification of PHA production systems based on circular economy. - Chapter 6. Once the technical feasibility and the environmental benefits of introducing these materials in the value chains are proved, it is necessary to provide ALBA ROIBÁS ROZAS 16 economic validation of the assessed processes as well. Therefore, techno-economic analysis (TEA) and Life Cycle Cost Assessment (LCCA) was performed. - As Chapter 5 aimed to contribute to fill a general gap found for PHA production systems, Chapter 7 aims to improve to the assessment of technologies using saline streams by contributing to methodology development in LCIA. Therefore, a new methodology to assess the effects of salinity variations in aquatic environments was stablished, addressing both negative and positive effects of emission-related impacts. Chapter 8 sums up the main conclusions of the work. CHAPTER 2: METHODOLOGY 1 2. METHODOLOGY SUMMARY This chapter provides an overview of the materials and methods used for the research done. It is then divided in two sections, acknowledging the work carried out experimentally and in the field of process evaluation. They cover: i) the experimental protocols and analytical methods used to follow up the bioreactors operation, and ii) the evaluation methodologies used for the processes assessment to increase technology readiness level, including Life Cycle Assessment (LCA) for the environmental evaluation, with special focus on the Life Cycle Impact Assessment (LCIA) stage, and Life Cycle Cost Assessment (LCCA) and Techno-Economic Analysis (TEA) for the economic evaluation. ALBA ROIBÁS ROZAS 2 2.1. ANALITICAL METHODS 2.1.1. Characterization of the liquid phase 2.1.1.1 Chemical Oxygen Demand (COD) The COD is defined as the amount of oxygen required to oxidise all the organic matter present in a sample (like wastewater) using a strong chemical oxidant (in this case, potassium dichromate) in acidic conditions. The catalyst silver sulphate is used to ensure the total oxidation of all organic compounds during the digestion of the samples. The global reaction of the process is: Cr2O7−2+ Organic Matter+ H+↔Cr3++CO2+H2O Hence, this measurement is performed to determine the concentration of organic matter in an analysed sample. After digestion, the remaining unreduced K2Cr2O72is titrated with ferrous ammonium sulphate (FAS) to determine the amount of potassium dichromate consumed, being the amount of oxidable (reduced) matter calculated in terms of oxygen equivalents. As the presence of Clinterferes with the reaction, the method described in Soto et al. (1989), that acknowledges the presence of salinity in the water samples, was applied when they had a concentration of salt that generated a ratio Cl-/COD above 2 (g/g). This method was an adaptation of the one stablished by APHA-AWWA-WEF (2017), which was used for the remaining samples. Total COD and soluble COD (referred in the text as CODT and COD, respectively) were measured by these two methods. The reagents are: - Potassium dichromate digestion solution: 10.23 g of K2Cr2O7, 33 g of HgSO4 and 167 mL of concentrated H2SO4 are diluted in 1 L of distilled water. - Sulphuric acid catalytic solution: 10.7 g of Ag2SO4 are dissolved in 1 L of concentrated H2SO4. The solution can be used after 2 days of preparation. - Ferroin indicator solution: 1.49 g of C18H8N2·H2O (phenanthroline monohydrate) and 0.69 g of FeSO4·7 H2O are dissolved in 100 mL of distilled water. - Standard potassium dichromate solution 0.05 N: 1.23 g of K2Cr2O7, previously dried at 105°C for 2 hours, are dissolved in 500 mL of distilled water. - Ferrous ammonium sulphate titrant (FAS) 0.035 N: 13.72 g of Fe(NH)4(SO)2·6 H2O and 20 mL of concentrated H2SO4 are dissolved in 1 L of distilled water. Determination procedure: A volume of 2.5 mL of each sample is placed in 10-mL Pyrex® glass tubes. Then, 1.5 mL of digestion solution and 3.5 mL of sulphuric acid reagent are added to the tube with the sample. A blank sample using distilled water is prepared in the same way to provide a reference for the zero value. The glass tubes are sealed with Teflon® and covered with Bakelite® caps to avoid the leakage of gases and liquids due to possible boiling. Then, the tubes are shaken and placed for 2 h in the thermodigester (VELP ECO16) preheated to 150 °C. After digestion, the tubes are cooled to room temperature. Then, the content of the tubes is transferred to a beaker and, after the addition of a couple of drops of ferroin indicator, the solution is titrated while stirring with standard FAS. The FAS solution is standardised as follows: 5 mL of distilled water are put into a small beaker and 3.5 mL of sulphuric acid reagent are added as well. Then, the solution is cooled to room temperature and 5 mL of standard potassium dichromate solution are added together CHAPTER 2: METHODOLOGY 3 with several drops of ferroin indicator to enable the titration with standard FAS. The endpoint is visualised by a sharp colour change from blue green to red. The COD concentration (in mg COD/L) of the different wastewater samples is calculated with the Equation 2.1: COD (mg L ⁄)=(A−B)∙8000∙NFAS V Eq. 2.1 Where A is the volume of FAS consumed by the blank (mL), B is the volume of FAS consumed by the sample (mL), 8,000 represents the milliequivalent weight of oxygen in 1,000 mL/L, V is the volume of sample used in the titration procedure, equal to 2.5 mL, and NFAS is the normality of the FAS solution (N), calculated as in Eq. 2.2: NFAS=VK2Cr2O7∙NK2Cr2O7 VFAS Eq. 2.2 Where 𝑉𝐾2𝐶𝑟2𝑂7 is the added volume of the standard K2Cr2O7 solution, equal to 5 mL. NK2Cr2O7 is the normality of the standard K2Cr2O7 solution, equal to 0.05 N. VFAS is the volume of FAS solution consumed in the titration procedure. 2.1.1.2 Total Organic Carbon (TOC), Total Carbon (TC), Inorganic Carbon (IC), and Total Nitrogen (TN) Organic carbon in liquid samples may include a variety of organic compounds. Measurement of Total Organic Carbon (TOC) is performed according to the method 5310 of the Standard Methods for the Examination of Water and Wastewater (APHA-AWWAWEF, 2017). To determine the quantity of organically bound carbon in a liquid sample, the organic molecules must be broken down, so the amount of carbon is quantified. The TOC concentration is determined by a Shimadzu analyser (TOC-L CSN) as the difference between the Total Carbon (TC) and the Inorganic Carbon (IC) concentrations. A curve comprising calibration points in the range of 0.5 to 1,000 mg C/L is used for the quantification. Potassium phthalate (C8H5KO4) and a mixture of sodium carbonate and bicarbonate (Na2CO3/NaHCO3, 3:4 w/w) are used as standards for TC and IC determination, respectively. The TC concentrations is determined by the amount of CO2 produced during the combustion of the sample at 720 °C, using a platinum catalyst. The IC concentration is obtained from the CO2 produced from the reaction at room temperature with a hydrochloric acid (HCl) 1 N reaction at room temperature. High purity air is used as carrier gas with a flow of 150 mL/min. The streams are cooled and dried at room temperature. The CO2 produced is optically measured with a nondispersive infrared detector (NDIR), which generated a peak with an area related to the concentration of the compounds. The Total Nitrogen (TN) is determined in the same Shimadzu analyser (TOC-L CSN). The sample is injected in the combustion tube where the oxidative pyrolysis occurs at 720 ˚C, to convert all nitrogenous compounds into nitric oxide gas (NO). The flow of gas is cooled to eliminate any possible condensation. Then, it reaches a detector where NO reacts with ozone (O3) to obtain nitrogen dioxide in an unstable excited state (NO2*). The NO2* reaches its elemental state by emitting a photon in the range of 590 - 299 nm. Then, it is detected by a chemiluminescence detector (CLD). The calibration is carried out by a standard commercial solution of ammonium (Fluka) in the range of 0.5 - 777 mg N/L. ALBA ROIBÁS ROZAS 4 2.1.1.3 Volatile Fatty Acids (VFAs) The VFAs are those monocarboxylic acids that contain 2 to 5 carbon atoms, from acetic to n-valeric acid. They are intermediate products in the anaerobic digestion for the production of biogas and generally necessary to produce Polyhydroxyalkanoates (PHAs) from wastewater using Mixed Cultures. As such, measuring their concentration is of key importance when monitoring the PHA production processes. VFAs were determined by gas chromatography (GC) (AGV-DB1 method). To a volume of 1 mL of sample 10 μL of H3PO4 (85%) are added. The sample (1 μL) is injected to a gas chromatography (GC) (6850 Series II, Agilent Technologies). The instrument is equipped with a flame ionisation detector (FID) and an automatic injector. The determination is performed using a glass column (30 m x 0.250 mm x 0.25 μm). Nitrogen is the carrier gas at a flux of 45 mL/min. The injector and detector temperatures are maintained at 250 and 300 °C, respectively. The oven is programmed to begin at 70 °C and increase the temperature at a rate of 15 °C/min to 130 °C; increase at a rate of 3 °C/min to 150 °C; increase at a rate of 10 °C/min to 180 °C and remain there for 1 min; increase at a rate of 65 °C/min to 250 °C, and hold at 250 °C for 5 min. Air and H2 are used as auxiliary gases with flows of 400 and 40 mL/min, respectively. The VFAs are separated in the column according to their molecular weights. A commercial VFA mixture dissolved in deionised water is employed as standard (Supelco, USA). The obtained chromatogram gives information about the area of the obtained peaks corresponding to each analysed VFA. Each VFA is identified by the retention time of its corresponding peak. This is made with a calibration curve for each acid in the range of concentrations of 0 - 0.6 g/L, which is related with the area of each peak. 2.1.1.4 Ammonium Ammonium concentration is determined spectrophotometrically using the method proposed by Bower and Holm-Hansen (1980). Here, indophenol blue is produced by the reaction of the ammonium ions, at pH 12.6, with salicylate and hypochlorite ions, in the presence of sodium nitroprusside as catalyst. The characteristic blue colour produced by increasing concentrations of ammonia makes the assay useful for the direct, visual estimation of ammonia in culture systems. The reagents are the following ones: - Reagent A: solution of 0.28 g/L of sodium nitroprusside (C5FeN6Na2O) and 440 g/L of sodium salicylate (C7H5NaO3). - Reagent B: solution of 18.5 g/L of sodium hydroxide (NaOH) and 120 g/L of trisodium citrate (Na3C6H5O7). - Reagent C: standard commercial solution of sodium hypochlorite (NaOCl) with 4.00 to 4.99 % active chlorine. - Reagent D: solution prepared by mixing 7 parts of reagent B and 1 part of reagent C. Reagent D is stable for 1 hour after preparation. Determination procedure: Reagent A (600 μL) and reagent D (1 mL) are added to 5 mL of sample. This mixture needs to be performed under a gas extraction device due to possible leakage of hazardous vapour. Samples with reagents are shaken and stored in obscurity, to react for more than CHAPTER 2: METHODOLOGY 5 2 hours but less than 3 (otherwise procedure is not valid). The measurements of the coloured samples are done with a spectrophotometer (Shimadzu, UV-1800) at a wavelength of 640 nm. The quantification is done with a calibration curve in the range of 0 – 0.9 mg NH4+-N/L using a commercial solution of NH4Cl. 2.1.1.5 Proteins The protein determination is done according to the Lowry’s method (Lowry and Randall, 1951). This procedure is based on two reactions. First, Biuret reaction, in which the alkaline cupric tartrate reagent complexes the peptide bonds of the protein. Then, the reduction of the Folin and Ciocalteu’s phenol reagent, which yields a purple colour, indicates the presence of proteins. Note that the reagents must be stored at 4 °C. Solution A and B must be prepared the same day of the analysis. Therefore, the reagents are: - Copper sulphate solution (1%): 1 g of CuSO4 is dissolved in 100 mL of distilled water. - Sodium tartrate solution (2%): 2 g of C4H4O6Na2 are dissolved in 100 mL of distilled water. - NaOH 1 N solution. - Solution A: 2 g of Na2CO3 and 1 mL of CuSO4 solution and sodium tartrate solution are dissolved in 100 mL of distilled water. - Solution B (Folin and Ciocalteu’s phenol reagent): 1 mL of commercial Folin and Ciocalteu’s phenol reagent are dissolved in 10 mL of distilled water. Determination procedure: A sample volume of 0.5 mL is rapidly mixed with 0.5 mL of NaOH 1 N solution and 5 mL of solution A and left for 10 minutes at room temperature. Then, 0.5 mL of solution B are added and mixed, and after 30 minutes the absorbance is ready to be measured. A blank with distilled water and the same reagents as the samples is also measured as reference for zero value. The measurements of the coloured samples are done with a spectrophotometer (Shimadzu, UV-1800) at a wavelength of 700 nm. The quantification of protein is done with a calibration curve using bovine serum albumin (BSA) as standard (Sigma), in a range of 0 - 0.6 g BSA/L. 2.1.1.6 Carbohydrates The carbohydrate concentration is measured with the Dubois et al. (1956) method and expressed in equivalent glucose. Carbohydrates react with the anthrone reagent under acidic conditions to yield blue-green colour. The reagents are: - Anthrone solution (0.2%): 0.2 g of anthrone are dissolved in 100 mL of sulphuric acid (98%). This solution is not stable, and it is necessary to prepare it the same day of the analysis. Determination procedure A volume of 2 mL of samples (diluted if necessary) is placed in 10-mL Pyrex® glass tubes, and 4 mL of anthrone solution are slowly added. A blank sample using distilled water is prepared in the same way, acting as reference for zero value. Then, the tubes are shaken and placed for 10 minutes in the block digester (VELP ECO16) preheated to 100 ALBA ROIBÁS ROZAS 6 °C. To stop the reaction, the tubes are rapidly removed from the bath and cooled immediately with ice. Then, the measurements of the coloured samples are done with a spectrophotometer (Shimadzu, UV-1800) at a wavelength of 625 nm. The quantification of carbohydrates is done with a calibration curve using glucose (Sigma), in a range of 0 to 1 g glucose/L. 2.1.1.7 Inorganic Ions Several inorganic ions were measured in the wastewater samples by chromatography. The anions are nitrite (NO2-), nitrate (NO3-), chloride (Cl-), bromide (Br-), phosphate (PO43-), sulphate (SO42-), and thiosulphate (S2O32-). The cations are sodium (Na+), ammonium (NH4+), potassium (K+), magnesium (Mg2+), and calcium (Ca2+). They are determined using an 861 Advanced Compact ion chromatography system equipped with a CO2 suppressor (MCS 853, Metrohm) and an 838 Advanced Sample Processor (Metrohm, Switzerland). Table 2.1 shows the calibration ranges for the different inorganic ion concentrations. In this chromatographic method, the sample goes through a column where cations and anions are separated according to their retention time in the resin column (Methrohm, Switzerland). Anions are determined with a Metrosep A column (250 x 4.0 mm) and a carbonate-bicarbonate mobile phase (3.2 mM Na2CO3 and 1.0 mM NaHCO3) at a flow rate of 0.7 mL/min. Table 2.1. Ranges of detection for the chromatographic method of ion detecting (in mg/L) Anion Lower Limit Upper Limit Cation Lower Limit Upper Limit Cl1.0 100.0 Li+ 0.05 5.0 NO20.05 5.0 Na+ 1.5 150.0 NO30.5 50.0 NH4+ 0.1 10.0 Br0.2 20.0 K+ 0.5 50.0 PO40.5 50.0 Mg2+ 0.5 50.0 SO41.5 150.0 Ca2+ 0.5 50.0 S2O31.5 150.0 Cations were determined with a Metrosep C3 resin column (250 x 4.0 mm) and nitric acid 3.5 mM is used as mobile phase. Then, the sample passes through a conductimetric detector after the separation, where the obtained signal corresponding to each retention time is registered. Therefore, each ion has a characteristic retention time. The resulting chromatograms identify each measured ion by its retention time position together with the amount of ion, which is related to the area. Then, the obtained areas are compared with the ones obtained in the standard calibration. 2.1.2. Characterization of the solid phase 2.1.2.1. Solids The measurement of Total Suspended Solids (TSS) and Volatile Suspended Solids (VSS) are performed according to standard methods (APHA-AWWA-WEF, 2017). For the TSS, a certain known volume of the sample is taken with a pipette. Then, the sample is poured through a fibre glass filter (Whatman, GF/C, 4.7 cm of diameter, 1.2 μm of pore size) that was previously tared. To do so, the filter is placed in an oven (Memmert) at 105 °C for half an hour, so the humidity is removed, and the filter has a constant weight. After achieving room temperature in a desiccator with silica, the filter can be weighed. CHAPTER 2: METHODOLOGY 7 The filter containing the retained solids is then dried for at least 2 hours at 105 °C, until reaching a constant weight. After that, it is again located inside a desiccator till it achieves room temperature. Finally, it is weighed, so the increase in the weight of the filter represents the TSS. TSS concentration is calculated as in Eq. 2.4: TSS (g L ⁄)=(W1−W0) V Eq. 2.4 Where W0 is the weight of the dried filter (g), W1 is the dry weight of the filter containing the solids (g), and V is the sample volume (L). Finally, for VSS determination, the filter previously dried is burnt inside a muffle furnace at 550 °C for half an hour. Then, it is located inside the desiccator to reach room temperature for another half an hour. Then, it is weighed, and the weight lost during ignition corresponds to the VSS content. VSS concentration is calculated as in Eq. 2.5, where W2 is the dried weight of the burnt filter: TSS (g L ⁄)=(W1−W2) V Eq. 2.5 2.1.2.2. Polyhydroxyalkanoates (PHA) The biopolymer content in the samples was determined according to Smolders et al. (1994) for the quantification of the monomer propyl esters present in lyophilized samples. Therefore, unfiltered biomass samples were centrifuged, and then the supernatant removed. After that, the remaining pellet was frozen and freeze-dried in a tube sealed with Parafilm. The sample tube was weighed three times (empty, full of sample and after freeze-drying) to calculate the total solids in the pellet (TS). The salt content of the samples was acknowledged to calculate the PHA content in the cells as g PHA/g VSS, as a total solids to total volatile solids ratio (TS:TVS) is stablished according to the TSS/VSS previously measured of the sample. The reagents are: - A digestion solution, where 1 L of a mixture in a ratio of 4:1 is prepared with 800 mL of 1-propanol (99.9 vol%) and 200 mL of hydrochloric acid (37 - 38 vol%). It is stored in the fridge and preserved from light. - Commercial 1,2-dichloromethane (HPLC quality). - Commercial standard of PHA (The ratio of Hydroxybutyric (HB) to Hydroxyvalerate (HV) is 88:12 wt.%). - Benzoic acid is used as internal standard in a concentration of 20 g/L. Here, 1 g of benzoic acid is dissolved in 50 mL of 1-propanol. This solution is preserved from light and stored in the fridge for a maximum of 12 months. - Commercial sodium sulphate anhydrous (99 wt.%). Determination procedure: Approximately 10 mg of dried samples are weighted in a high-precision scale. Then, they are located into 10 mL glass tubes and 50 μL of the benzoic acid standard solution is added in each tube. Then, 1.5 mL of the digestion solution and 1.5 mL of 1,2dicloromethane are added under a gas extraction system to prevent the inhalation of hazardous vapours. It is important to make sure that the glass tubes are completely sealed to prevent gas leaks during the following steps of the determination procedure. Therefore, the tubes are placed in an oven at 100 °C (Memmert) for 4 hours. During this time, the ALBA ROIBÁS ROZAS 8 hydrochloric acid breaks the cell walls, releasing the microbial material and the PHA. The strong acid hydrolyses the PHA, that will be later esterified by the propanol. Finally, the non-polar esterified compounds are dissolved in the organic phase solvent (dichloromethane), which will be analysed in the gas chromatography unit. After cooling the samples to room temperature, 3 mL of distilled water are added to the digested samples, and all the tubes are energetically shaken in a vortex (IKA MS1, Germany). Samples are then centrifuged at 2500 rpm for 5 minutes to separate the two liquid phases: organic and aqueous. The PHA is then present in the organic fraction, while the water contains the remaining cell material, like proteins. A sample of 1.5 mL is taken from the organic phase with a glass pipette. Then, the sample is immediately filtered and dried by making the liquid go through a pipette tip of 1 mL filled with glass wool (which acts as a filter) and sodium sulphate anhydrous (which acts as dryer agent). Finally, the filtered and dried samples are quickly located into GC vials, which are sealed with a septum. Samples are analysed by a gas chromatographer (6850 Series II, Agilent Technologies) equipped with a flame ionisation detector (FID) with a mixture of H2/air at 40:400 mL/min; an automatic injector, that injects 1 μL of sample at 250 °C. The used software is Chemstation®. The chromatographic column employed is an HP-INNOWAX 30 m x 0.25 mm x 0.25 μm (Agilent, USA) with Helium as the carrier gas, being supplied at a constant velocity of 1 mL/min. The oven is operated at a temperature ramp from 60 °C to 230 °C with a velocity increase of 10 °C/min. The column can measure the monomers present in the polymer, and the identified ones are HB and HV. Therefore, a calibration curve is done by using the PHA standard (Sigma) that contains HB and HV, so the retention times of the monomers in the column can be detected and identified. Then, the areas of the peaks represent mass of HB and HV in the samples. Therefore, when compared the areas of the samples with the values of the calibration curve, the concentration of the monomers is obtained. The calibration curve is done by plotting the weight of the standard against its corresponding ratio value. This ratio is obtained by dividing the area corresponding to the peak of PHA (HB or HV) by the area of the peak of the benzoic acid used as internal standard. 2.1.3. Other Parameters 2.1.3.1 pH The pH was measured with an electrode (52‐03, Crison Instruments, USA) equipped with an automatic compensatory temperature device (21‐910‐01, Crison Instruments, USA) and connected to a digital pH-meter. The uncertainty of the instrument was 0.01 pH units. The electrode was calibrated at room temperature with two standard buffer solutions of pH 7.02 and 4.00. 2.1.3.2. Alkalinity The alkalinity of a sample is linked with its capacity of neutralizing acids, so it is intimately related with wastewater pH. Therefore, this parameter is important for effluents that go through an acidification process, as it might limit the system capacity for VFA generation. Although alkalinity is linked to the presence of salts and weak acids, at near neutral pH, the most common buffer substances are carbonate, bicarbonate and hydroxides. However, other substances (like borates, phosphates, silicates and/or other bases) might CHAPTER 3: STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM INTO PHA 15 3. STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM FROM THE FISH CANNING INDUSTRY INTO PHA SUMMARY Complex saline Mussels Cooking Wastewater (MCW) was valorised to produce polyhydroxyalkanoates (PHA) with Mixed Microbial Cultures (MMC) in a typical three stage system with acidification, enrichment, and accumulation units. Due to the high organic matter content in the form of proteins (protein Chemical Oxygen Demand or CODPROT), which ranged 1.8 – 5.7 g CODPROT/L, PHA accumulating capacity was below 10 %. Therefore, several strategies were tested with the goal of improving this storing rate. In the acidification unit, Na(HCO3) was added to manage pH and alkalinity, increasing protein conversion into Volatile Fatty Acids (VFA) from 10.3 % to 69.2 % and subsequent PHA accumulation from 6.9 to 14.7 %. Then, the proteins/VFA ratio was varied by modifying the MCW in the feeding, so MMC could gradually acclimate to the high protein concentration of the feedstock. This allowed to increase PHA accumulation to 41.5%. Finally, in the enrichment unit, the incorporation of a settling stage after the feast phase provoked a shift in the proteins’ oxidation from the feast to the famine phase, where the nitrogen released during the starvation period due to protein uptake is used by the MMC for growth. This increased the biomass concentration from 1.6 to 4.2 g VSS/L, and the tolerated COD from 2.2 to 4.38 g COD/L, yielding in increased productivities. The content of this chapter was published as A. Roibás-Rozas, Á.V. del Río, A. Hospido, A. Mosquera-Corral (2021), Strategies for the valorisation of a protein-rich saline waste stream into polyhydroxyalkanoates (PHA), Bioresource Technology. https://doi.org/10.1016/j.biortech.2021.124964. All authors’ affiliation is CRETUS, Department of Chemical Engineering, Universidade de Santiago de Compostela, 15782 Santiago de Compostela, Spain ALBA ROIBÁS ROZAS 16 3.1 INTRODUCTION PHA are renewable, biodegradable and bio-based polymers produced by bacteria which are expected to gradually replace conventional petrochemical plastics (Kourmentza et al., 2017). In the last decade, the use of MMC and waste streams for PHA production has been explored to reduce overall costs and environmental impacts (RoibásRozas et al., 2020; Yadav et al., 2020). However, the use of waste streams for PHA production is challenging due to the complexity of the substrate, containing substances that can be inhibitory for the microbial activity (Korkakaki et al., 2016a; PalmeiroSánchez et al., 2019). The Spanish north-western region of Galicia is responsible for the 90 % of the production of mussels in the country, which is also the main mollusc producer of Europe (FAO, 2020). This activity generates more than 100 M€ of incomes and more than 10,000 jobs in the region (Rey-Méndez et al., 2016), and it is highly dependent on the water quality in the estuaries where seafood is harvested. Because of this, a proper treatment of the effluents generated in the fish canning and fish-processing industries, normally located at shore areas, is essential for the environmental and economic preservation of the region. Nevertheless, the treatment of these effluents is challenging due to the presence of potentially inhibitory compounds, such as sodium chloride or ammonia. More precisely, Mussels Cooking Wastewater (MCW) contains high sodium chloride concentrations (between 17.9-19.0 g NaCl/L), as well as sulphate and nitrogen (organic and in the form of ammonia) in ranges that can be inhibitory for conventional processes, like the anaerobic digestion (Palmeiro-Sánchez et al., 2013). In fact, several attempts were made to valorise MCW in Galicia through biogas production. They were unsuccessful possibly due to the complexity of the waste stream (Barros et al., 2009; Omil et al., 1995), so the great concern regarding MCW treatment is reflected in local legislation (Consellería de Medio Rural e Mar, 2015). Because of this, and its large biodegradable organic matter concentration, the valorisation of MCW by PHA production was defined as a possible alternative to treat these effluents. The COD of MCW, ranging between 9.8-26.5 g COD/L, is about 95% soluble and mainly composed by proteins and carbohydrates. Carbohydrates are rapidly consumed in fermentation processes, but proteins are consumed slowly. Furthermore, their presence can provoke problems in the subsequent aerobic PHA production unit, so several strategies, as uncoupling the carbon and nitrogen feeding or introducing a settling stage in the process have been proposed (Argiz et al., 2020a; Oliveira et al., 2017). The aim of this chapter is to evaluate how the presence of proteins affects the MMCbased PHA production process and to assess the feasibility of using protein-rich and saline wastewater as a substrate. To do so, several long-term strategies were tested in a threestage system for PHA production, using a MMC fed with MCW: i) alkalinity increase in the acidification unit, ii) a settling stage in the PHA enrichment reactor, and iii) increasing the MMC capacity to consume proteins by modifying the proteins/ VFA ratio. Then, these strategies were assessed and compared to choose the optimal operational conditions. 3.2 MATERIAL AND METHODS 3.2.1 Experimental set-up A three-step system was utilized to produce PHA (see Appendix 1) as described in the following sections: a CSTR for acidification of the MCW; SBR for enrichment of the MMC; and a FBR for the accumulation of PHA. The experimental stages tested in the CHAPTER 3: STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM INTO PHA 17 continuous operation are summarized in Figure 3.1 and named as: A1-A6 for the acidification CSTR, E1-E5 for the enrichment SBR. 3.2.1.1 CSTR for acidification A CSTR was fed with MCW coming from a facility in Pontevedra, NW of Spain. Although the main compounds of the MCW did not change throughout the operational time, their concentrations fluctuated highly (see Appendix 1), so the Organic Loading Rate (OLR) for CSTR stages ranged between 1.01 and 1.76 g COD/(L·d). It had a volume of 5 L and it was operated as described in Fra-Vázquez et al. (2020). This study started in the operational day 400, where VFA were produced steadily at a pH of 4.6 ± 0.6 with about 43 % of the effluent COD in the form of VFA. Figure 3.1: Operational periods in the three-stage PHA production system. Continuous lines represent the operational stages and dashed lines represent the mass balances for each one of them. The boxes filled with diagonal grey lines for the enrichment stages represent the periods where the volume of the reactor was reduced from 2 L to 1 L. The effluent of the CSTR is referred in the figure as fMCW (fermented MCW) to facilitate the comprehension of the scheme. The Solid Retention Time (SRT) was fixed equal to the Hydraulic Retention Time (HRT), with a value of 6.25 days. The operational conditions by (Fra-Vázquez et al., 2020) were first maintained (stage A1), but then, different amounts of sodium bicarbonate were added (20 - 2700 mg NaHCO3/L of MCW in stages A2 to A6, see Table 3.1) to increase the pH and improve protein solubility in the CSTR. The effluent produced in the CSTR was stored at 4 ºC until it was used in the following units. This feature was linked to the operation scale, and it was used for a better load control in the next units, so operational stages for the CSTR and Sequencing Batch Reactor were not synchronized. In fact, this provoked a change in the operational volume of the subsequent SBR (see Figure 3.1, and Table 3.1). 3.2.1.2. Enrichment SBR The inoculum of the SBR was enriched in PHA accumulating biomass (accumulation capacity of 25 %), as it came from a similar bench-scale reactor that was fed with the same MCW for a year, but this feeding was diluted with tap water to decrease the salt concentration to approximately of 5 g NaCl/L (Pedrouso et al. (2020)). ALBA ROIBÁS ROZAS 18 For the present work, salinity was gradually increased until achieving the salinity of the MCW (17-18 g NaCl/L) in 100 days by replacing the tap water by the effluent of the same SBR as dilution agent, which means that the SBR effluent was recirculated to the influent. Then, two different operational strategies (Cycles 1 and 2) were evaluated (see Appendix 1): conventional and modified Aerobic Dynamic Feeding (ADF and M-ADF, respectively). Table 3.1: Operational Stages throughout the operation of the acidification (CSTR) unit. Stage Na(HCO3) (mg/L of feeding) Days OLR (g COD/(L·d)) Stage in SBR(1) A1 0 0-263 1.01 ± 0.21 E1-E2 A2 20 264-332 1.27 ± 0.13 E2-E3 A3 80 333-395 1.44 ± 0.23 E4 A4 320 396-466 1.76 ± 0.21 E4 A5 640 467-502 1.75 ± 0.21 E5 A6 2700 503-621 1.48 ± 0.42 E5-E6 (1) CSTR effluent was stored until used, so the operational time is not always coincident. This column shows the link between the effluents obtained in each acidification CSTR operational stage (employed as SBR feeding) and the SBR operational stage. When the conventional ADF strategy (Cycle 1) was applied, air was continuously supplied while, with the M-ADF one (Cycle 2), a settling and discharge intermediate phases were included. In the case of Cycle 1, the exchanged volume was 50 % so the SRT and HRT of operation were equal to 1 day. For Cycle 2, no oxygen was supplied during the settling and supernatant discharge phases, and aeration was provided throughout the rest of the cycle phases. After settling, 25 % of the total reactor volume (VR) was withdrawn (supernatant discharge), so during the rest of Cycle 2, the reactor operated with 75 % of the initial volume. As the initial feeding corresponded to 50 % of VR and the final discharge was of 25 % of VR, the SRT and HRT were of 2 and 1 days, respectively. The feeding of the SBR consisted of centrifuged CSTR effluent, where the dilution agent was the settled (without solids) effluent of the same SBR. The recirculation ratio (RR, i.e., volume of effluent from the CSTR / volume of effluent from the SBR used as feeding) was established according to the composition of the effluent produced in the CSTR and the organic load required to feed the SBR (Table 3.2). The operation of the SBR was divided in six stages (from E1 to E6) depending on the percentage of COD present as proteins (RPROT) at the beginning of the cycles and on the cycle type (ADF or M-ADF) (Figure 3.1). Allylthiourea was added to the feeding of the SBR at 10 mg/L to avoid nitrification, antifoaming (0.5 mL/L of feeding) to prevent spume problems and NaOH or HCl to maintain the pH value in the feeding approximately neutral in stages from E3 to E6, but pH inside the reactor was not controlled. For all the stages, the reactor operated at 30 ºC by using a thermostatic bath (Techne Inc., USA), and air was supplied (6 L/min) through a ceramic air diffuser located at the bottom of the reactor. As the CSTR effluent was not enough to achieve the desired loads from stages E4 to E6, the SBR operation volume was reduced from 2 L (used from stages E1 to E3) to 1 L, maintaining all the operational conditions. CHAPTER 3: STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM INTO PHA 19 Table 3.2: Operational Stages throughout the operation of the enrichment (SBR) unit. Stage Days Cycle CODVFA,0 (g CODVFA/L) RVFA(1), % (g CODVFA/g COD) SRT (days) RR(2) E1 0-205 ADF 0.50 ± 0.26 34.98 ± 16.79 1 1:4 E2 206-263 M-ADF 0.71 ± 0.11 32.02 ± 7.32 2 1:5 E3 264-335 M-ADF 0.72 ± 0.32 33.19 ± 14.22 2 1:6 to 1:1 E4 336-425 M-ADF 2.69 ± 0.75 61.36 ± 9.97 2 1:1 or 1:0 E5 425-498 M-ADF 1.75 ± 0.48 51.28 ± 8.84 2 1:1 to 1:3 E6 499-536 ADF 0.94 ± 0.37 36.98 ± 4.73 1 1:4 (1) As the COD at the beginning of enrichment cycles was mainly VFA and proteins (CODCARB was negligible), RPROT can be estimated as 100 – RVFA. (2) RR: Recirculation ratio to the SBR (CSTR effluent volume: SBR recirculated effluent volume). 3.2.1.3. FBR An FBR was employed to evaluate the maximal PHA accumulation capacity of the enriched MMC. As inoculum, it was used biomass from the SBR, and as feeding, centrifuged effluent of the CSTR (containing VFA and proteins), which was added in pulses. Analysis of the liquid phase (data no shown) showed that VFA were consumed first and proteins after, being possible to use the dissolved oxygen (DO) concentration value as indicator. For this reason, each pulse of feeding in the FBR was supplied when the DO concentration increased to intermediate values (approximately 4.0 - 4.5 mg O2/L), which marked that VFA was depleted, and protein oxidation started. To make assays comparable, the amount of VFA supplied in every pulse was set as 15 CmmolVFA/(LREACTOR·pulse), except on day 388, when pulses of 70 CmmolVFA/(LREACTOR·pulse) by increasing the volume of CSTR effluent in each pulse. Therefore, in the accumulation assays, the FBR volume was variable according with the CSTR effluent acidification degree and the number of pulses. Each experiment lasted a maximum time of 12 hours, which was equal to the enrichment cycle length. The reactor temperature was maintained at 30 C by using a thermostatic bath (Techne Inc., USA), and air was supplied (6 L/min) through a ceramic air diffuser located at the bottom of the reactor. 3.2.2. Analytical methods Analytical methods are described in detail in Chapter 2 of this thesis. Conductivity and pH were measured with glass electrodes (Hach-Lange 50-60 and Crisson GLP22 respectively), while DO concentration and temperature were determined in the SBR and FBR with a probe (model HQ40d, Hach-Lange, USA). Alkalinity, TSS, VSS and total COD (CODT) concentrations were analysed in bulk samples, and the dissolved matter was characterised after filtering with a cellulose-ester filter of 0.45 μm pore size (Advantec, Japan), according to the Standard Methods (APHAAWWA-WEF, 2017). Measuring included the following parameters: soluble COD (noted as COD in the text) carbohydrates, proteins, Total Organic Carbon (TOC), Total Nitrogen (TN), VFA, ammonium and other ions (Na+, Cl-, SO42-). COD was generally quantified according to APHA-AWWA-WEF (2017), and based on the methodology described by Soto et al. ALBA ROIBÁS ROZAS 20 (1989) when samples had concentrations below 5 g COD/L. Carbohydrates were measured as indicated in Dubois et al. (1956) and expressed as COD considering that 1 g of glucose (used as standard) corresponds to 1.07 g of COD. Proteins were analysed by the Lowry method (Lowry and Randall, 1951) and expressed as COD considering that bovine serum albumin (used as standard) contains 1.32 g of COD and 0.15 g of N per gram of protein. TOC and TN concentrations were measured by catalytic combustion in the TOC-L CNS analyser with the TNM-1 module (Shimadzu, Japan). VFA concentrations were determined in a gas chromatograph (Hewlett Packard 5890A, USA) equipped with a flame ionization detector (FID) and an automatic injector (Hewlett Packard 7673A, USA) and with ChemStation Rev. A. 10. 02 (1757) Agilent Technologies software. Ammonium was quantified by the Bower/Holm Hansen method (Bower and Holm-Hansen, 1980) and ions (Na+, Cl-, SO42-) by ion chromatography (861 Advanced Compact IC, Metrohm, Switzerland). PHA content was determined in biomass samples. To do so, unfiltered biomass samples were centrifuged, and the supernatant removed. Then, the remaining pellet was frozen and freeze-dried. The sample tube was weighed three times (empty, full of sample and after freeze-drying) to calculate the total solids in the pellet (TS). The PHA content (as g PHA/g VSS) was measured following the method described by Smolders et al. (1994) for the quantification of the monomer propyl esters present in a lyophilized sample. A commercial PHA standard (Sigma-Aldrich, USA) containing 88 % of hydroxybutyrate (HB) and 12 % of hydroxyvalerate (HV) and benzoic acid as internal standard were used. The propyl esters were analysed by means of gas chromatography in a HP innovax column equipped with an FID (Agilent, USA). Biogas composition (N2, CH4, CO2, H2S) was analyzed in a gas chromatograph (Hewlett Packard, USA) equipped with a packed column and a thermal conductivity detector (TCD) using helium as carrier gas and the percentage of each compound was calculated by analyzing 1 mL of biogas injected with a syringe. 3.2.3. Calculations 3.2.3.1. CSTR mass balances For the CSTR, the main parameters calculated were the acidification degree (Ac., Eq. 3.1), protein and carbohydrates removal percentages (REMCARB and REMPROT), Eq. 3.2) and VFA production (CODVFAprod, Eq. 3.3). In Eq. 3.1, CODVFA is the concentration of VFA in the effluent (g COD/L). Subindex i represents whether the equation refers to carbohydrates or proteins. “INF” and “EFF” subindexes represent influent and effluent streams, respectively. Ac.(%)= CODVFA COD ·100 Eq. 3.1 REMi(%)=CODi,INF−CODi,EFF CODi,INF ·100 Eq. 3.2 CODVFAprod=CODVFA,EFF−CODVFA,INF CODINF Eq. 3.3 CHAPTER 3: STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM INTO PHA 21 3.2.3.2. SBR mass balances For the SBR, the COD concentration at the beginning of an enrichment cycle (COD0, in g COD/L) was calculated according to Eq. 3.4 because it has a 50 % of exchange volume ratio. Accordingly, the COD concentration of proteins or VFA at the beginning of the cycle (CODi,0, g CODi/L) was calculated as indicated in Eq. 3.5. Analogously, this mass balance can be applied to calculate the concentration of the nitrogen forms in the cycle (substituting COD by TN or NH4+-N, in mg N/L). Here, “INF” refers to influent (feeding stream) and “EFF” to the effluent of the previous cycle, which coincides in concentration with the remaining volume inside the reactor. The proportion of CODPROT over the total COD at the beginning of the SBR cycles (RPROT, % (g CODPROT/g COD)) is calculated as stated in Eq. 3.6, where the COD is mainly in the form of proteins and VFA (CODCARB was negligible). COD0= CODINF+ CODEFF 2 Eq. 3.4 CODi,0= CODi,INF+ CODi,EFF 2 Eq. 3.5 RPROT= CODPROT,0 COD0·100 Eq. 3.6 A mass balance was performed to calculate protein consumption during the cycle. For conventional ADF, protein removal is only due to oxidation. However, for Cycle 2 (M-ADF), protein removal also occurs due to the withdrawal of the supernatant. Therefore, the protein removal rate due to oxidation (REMPROT,OX, %) and to supernatant removal (REMPROT,SUP, %) are calculated according to Eq. 3.7 and Eq. 3.8, respectively, where VSUP is the volume of the supernatant withdrawal and VR is the reactor volume at the beginning of the cycle after feeding addition. REMPROT.OX 100 ⁄= = CODPROT,0·VR−CODPROT,SUP·VSUP−CODPROT,EFF·(VR−VSUP) CODPROT,0·VR Eq. 3.7 REMPROT,SUP= CODPROT,SUP·VSUP CODPROT,0·VR·100 Eq. 3.8 For Cycle 1, VSUP is zero, so REMPROT,OX is (CODPROT,0 – CODPROT,EFF)/CODPROT,0. These mass balances can be applied for nitrogen forms by substituting CODPROT by TN or NH4+-N. Finally, the protein variation during the cycle phases due to oxidation (ΔPROTOX in g PROT/L) is calculated as indicated in Eq. 3.9: ∆PROTOX=∆PROTF+∆PROTFAM Eq. 3.9 Where ΔPROTF and ΔPROTFAM (both in g PROT/L) are the variation in the concentration of proteins during the feast and the famine phases, respectively, and they are calculated as indicated in Eq. 3.10 and 3.11: ∆PROTF=PROT0−PROTMID Eq. 3.10 ALBA ROIBÁS ROZAS 22 ∆PROTFAM=PROTEFF−PROTMID Eq. 3.11 Being PROTMID (g PROT/L) the protein concentration in the middle of the cycle (the point where feast finishes and famine begins). Accordingly, the percentage of CODPROT oxidated during feast or famine (REMPROT,F or REMPROT,FAM, in %) can be calculated by dividing ΔPROTF or ΔPROTFAM by ΔPROTOX 10. 3.2.3.3. PHA content and kinetic parameters For the SBR and the FBR, PHA content in the cells is calculated according to Eq. 3.12. Specific consumption and production rates (qVFA, qPROT and qPHA, in mg/(g X·h)) are estimated from the maximum slopes of the curves obtained from the corresponding experimental data, divided by the average active biomass (X, calculated as the subtraction of the mass of PHA to the mass of VSS measured). The active biomass composition was considered CH1.8O0.5N0.2 for all the calculations. The biomass yield (YX) was estimated dividing the biomass production rate (g CODX/h) by the VFA consumption rate (g CODVFA/h). Finally, the ratio r (g VFA/g proteins) is calculated by dividing qVFA/qPROT. PHA (wt.%)=g TSL ⁄·g PHA in sampleg TS ⁄ g VSSL ⁄·100 Eq. 3.12 3.3. RESULTS AND DISCUSSION 3.3.1. Protein fermentation in the acidification reactor (CSTR) The CSTR operated for 621 days in 6 operational stages according to the Na(HCO3) added in the feeding, which was increased from 0 to 2.7 g/L to improve protein solubility and promote their better degradation. Changes in the applied load (1 - 1.76 g COD/(L·d)), are due to the variability of the MCW (Table 3.1). From stages A1 to A5, pH in the CSTR was always below 5.0 due to VFA formation and low alkalinity, because the added amount of Na(HCO3) was still not enough to clearly increase the pH value. For stage A6, pH was around neutrality (pH = 6.95 ± 0.57) due to the highest added alkalinity dosage (2,7 g Na(HCO3)/L), so A6 corresponds to a neutral-pH stage. Carbohydrates were almost fully consumed (over 90%) for every stage, but no proteins (Figure 3.2). For acidic-pH stages, protein removal ranged 10.30 – 21.94%, while, for neutral-pH stage A6 it increased to 42.45 ± 19.56%. The higher protein removal promoted an increase in the CODVFAprod and acidification from 0.21 – 0.28 g CODVFA/g COD and 45.7 – 60.6% in A1-A5 to 0.43 ± 0.12 g CODVFA/g COD and 71.6 ± 13.1 % in A6, respectively. Therefore, the increase in CODVFAprod and acidification percentages for the neutral-pH stage can be attributed to the increase in the protein consumption associated to the modified pHenvironmental conditions. Successful acidification of the MCW was achieved and no methane was detected during the whole operational time. For stages A1-A5 (acidic-pH), the inhibition of methanogenic activity can be attributed to the low pH of operation (below 5) caused by VFA accumulation (Fra-Vázquez et al., 2020), but this cannot be the reason for the inhibition for the neutral-pH stage (stage A6). Although high 10 Note that REMPROT,F and REMPROT,FAM (%) can be expressed as g/g or g CODPROT/g CODPROT, to obtain the same result. CHAPTER 3: STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM INTO PHA 23 concentrations of salt, and in special sodium ion, can inhibit methanogenic activity, methane can appear in long-term operations despite high salinity (Palmeiro-Sánchez et al., 2013). Figure 3.2: Protein Removal (REMPROT, ), carbohydrates removal (REMCARB, ◼), and Acidification Degree (Ac, Δ), expressed as percentages, and VFA production (CODVFAprod ○, g CODVFA/g CODINF) in the CSTR. However, when sulphate is present in the wastewater and the COD/SO4-2 ratio is below 10 g/g, methanogens are inhibited due to their competition with sulphate reducing bacteria. In this case, the complex matrix of MCW makes this waste stream ideal for VFA production, as it has high concentrations of sodium due to salinity, and COD/SO4-2 ratio below 10 g/g, which inhibits methanogens. Therefore, the compounds present in the MCW caused the absence of methanogenic activity during all the CSTR operation. The transformation of proteins into VFA in anaerobic mixed culture fermentations has generated increasing interest in the last years (Bevilacqua et al., 2020; Jin et al., 2016; Liu et al., 2015; Regueira et al., 2020a, 2020b; Yang et al., 2015), as they are degraded slower than carbohydrates. Hydrolysates of proteins and carbohydrates are peptides/amino acids and glucose, respectively. As the presence of glucose represses protease formation, the uptake of glucose/carbohydrates happens first (Yang et al., 2015). Moreover, pH is one of the most relevant factors regarding VFA production from proteins (Regueira et al., 2020a). When pH is low and VFA are formed, undissociated acids cross the membrane, releasing protons inside the cell and inhibiting the generation of new acids. Furthermore, as the concentration of protons in the solution increases, the interaction of hydrogen bonds becomes stronger. Thus, the solubility of proteins decreases, as they aggregate and even coagulate at low pH values (Liu et al., 2015). This can explain the results of the CSTR operation, where carbohydrates are always consumed, and protein uptake only happens at the neutral pH. 0.0 0.1 0.2 0.3 0.4 0.5 0 20 40 60 80 100 A1 (4.47) A2 (4.42) A3 (4.84) A4 (4.41) A5 (4.37) A6 (6.95) CODVFAprod (g CODVFA/g CODINF) REMPROT, REMCARB, Ac., (%) Stage (CSTR pH) ALBA ROIBÁS ROZAS 24 3.3.2 Enrichment (SBR) and accumulation (FBR) reactors The SBR was operated during 540 days, and the experimental time was divided in six stages (Figure 3.1), selected according to: a) CSTR operational conditions (generating an acidic-pH or neutral-pH feeding), b) proportion of CODPROT in SBR cycles (if RPROT is higher than 50%, it is a high-CODPROT stage; if RPROT is below 50%, it is a lowCODPROT stage and, consequently, a high-CODVFA stage), and c) enrichment strategy (ADF or M-ADF). A recirculation was performed in five of the stages (Table 3.2) with the aim of reaching stable salt concentrations avoiding the use of tap water for dilution and load control. The maximum concentrations of 28.3 ± 0.7 g NaCl/L in the SBR-FBR corresponded to the use of saline wastewater (19 g NaCl/L), the addition of alkalinity as Na(HCO3), and NaOH for feeding pH control, so the raise was gradual (data not shown) to let the biomass adapt to the highly saline conditions. 3.3.2.1. SBR operation using acidic-pH feeding Proteins were present in the SBR feeding, although their proportion in the COD (RPROT) varied (Figure 3.3). To unravel the effect of proteins on the enrichment process, two operational strategies were tested by applying the two previously defined cycle distributions (ADF and M-ADF) to two levels of protein (high and low RPROT). - High RPROT and conventional ADF The enrichment of PHA-accumulating organisms was initiated applying an ADF regime (stage E1). Along with the 100% consumption of VFA (data not shown), approximately 34.7% of the proteins were removed in each cycle during the feast period (Table 3.3, Figure 3.4). Therefore, the MMC consumed proteins simultaneously to VFA. The presence of non-VFA COD in the enrichment process (like proteins) can be harmful for the selection, as carbon is used by non-storing populations to grow and proliferate in the system (Argiz et al., 2020). This issue was explored by Korkakaki et al. (2016b) for a system with methanol and VFA, where they proposed the parameter r (qVFA/qMeOH) as a control for the enrichment success. They found that r needs to be equal to or higher than 4.7 g HAc/g MeOH (or 5 Cmol HAc/Cmol MeOH) for MMC selection to be effective. Here, r (qVFA/qPROT) had values of 0.81-2.60 g VFA/g protein (Table 3.3), indicating poor MMC selection, which in certain periods consumed proteins even faster than VFA. Hence, maximal PHAaccumulation capacities of the MMC from E1 in the FBR were always below 10% (Table 3.3). - High RPROT and M-ADF Later, the cycle distribution of the SBR was modified to M-ADF (Stage E2), so a settling phase was introduced to improve MMC selection. After withdrawal, the removed volume (which was 25% of VR) was not substituted with any other stream due to the inexistence of a current with the same composition of the acidified wastewater but without proteins, thus only 75% of the initial volume remained inside the reactor during the rest of the cycle. CHAPTER 3: STRATEGIES FOR THE VALORISATION OF A PROTEIN-RICH SALINE WASTE STREAM INTO PHA 31 Summing up, the enrichment strategy did not directly affect the storing capacity (it is only boosted by the load increase) but it does indirectly, as a load raise is only possible if a proper enrichment strategy (M-ADF) is applied. Therefore, suitable selection of the culture (where non-VFA forms are present) is only possible if two selective pressures are imposed. Again, the maximum COD0 tolerated by the system was 2 – 3 g COD/L, so RPROT increased due to dilution. Moreover, a change in the enrichment strategy reversed the SBR behaviour, which presented similar r values and biomass concentration than in stage E1, while protein oxidation occurred again during famine (as in E1, Table 3.3). Therefore, for the same operational strategy but lower protein availability, the system performance was not improved in E5 although protein consumption was lower with respect to stage E1. 3.4 CONCLUSIONS Protein conversion into VFA in the acidification unit was enhanced from 10.3% to 69.2% with the addition of Na(HCO3) and neutral pH. In the enrichment unit, proteins can be beneficial (around/below 30% of COD) if the enrichment strategy includes a settling stage to promote their consumption in the famine, because the nitrogen released is used for biomass growth. Therefore, as biomass concentration increases, the load tolerated and the PHA produced by the system are higher. Finally, varying the proteins/VFA ratio to acclimate the MMC to proteins allowed increasing the PHA accumulation from 8.8 to 41.5%. ALBA ROIBÁS ROZAS 32 CHAPTER 4: HOW CAN WE VALIDATE THE ENVIRONMENTAL PROFILE OF BIOPLASTICS? 33 4. HOW CAN WE VALIDATE THE ENVIRONMENTAL PROFILE OF BIOPLASTICS? TOWARDS THE INTRODUCTION OF PHA IN THE VALUE-CHAINS. SUMMARY Petrochemical plastics are threatening the environment, human health, and economy. Biodegradable plastics produced from renewable sources like polyhydroxyalkanoates (PHA) are a promising alternative to solve this crisis. However, the fact that a material is biodegradable, or renewable, does not automatically make it environmentally friendly, so a validation (normally though life cycle assessment (LCA)) is needed. Thus, the present review studied the research works published in the last 20 years about this issue. It focuses on the methodological features defining the outputs of the LCA, like the functional unit and allocation choices, but also on the data sources of the existing studies, as these issues need to be addressed by the LCA community to properly evaluate the environmental profile of biomaterials like PHA. As found in this review, the few LCA studies for PHA production obtained contradictory results, mainly due to the lack of actual data, but also to the absence of a consensus methodological framework. As one of the main results of this review, we found that many current references still use inventories generated twenty years ago. This is necessarily going to generate conflicting results and is not going to provide information in concordance to nowadays processes, so LCA practitioners need to create new inventories reflecting nowadays PHA production processes if accurate results are wanted. Part of the content of this chapter was published as A. Roibás-Rozas1, M. Saavedra del Oso1, G. Zarroli1,2, M. Mauricio-Iglesias1, A. Mosquera-Corral1, S. Fiore2, and A. Hospido1 (2022). How can we validate the environmental profile of bioplastics? Towards the introduction of Polyhydroxyalkanoates (PHA) in the value-chains. Chapter 20 in: C. Teodosiu, S. Fiore, A. Hospido (Eds.), Assessing Progress Towards Sustainability, Elsevier. https://www.elsevier.com/books/assessing-progress-towardssustainability/teodosiu/978-0-323-85851-9 a CRETUS, Department of Chemical Engineering, Universidade de Santiago de Compostela, 15782 Santiago de Compostela, Spain b DIATI (Department of Engineering for Environment, Land and Infrastructures), Politecnico di Torino, 10129 Torino, Italy ALBA ROIBÁS ROZAS 34 4.1 INTRODUCTION Petrochemical plastics are threatening the environment, human health, and economy. Biodegradable plastics produced from renewable sources like PHA are a promising alternative to solve this crisis. However, the fact that a material is biodegradable, or renewable, does not automatically mean a better environmental profile, so a validation (normally though LCA) is needed. The few LCA studies for PHA production obtained contradictory results, mainly due to the lack of actual data, but also to the absence of a consensus methodological framework. The present review studied the research works published in the last 20 years about this issue. It focuses on the methodological features defining the outputs of the LCA, like the functional unit and allocation choices, but also in the data sources of the existing studies, as these issues need to be addressed by the LCA community to properly evaluate the environmental profile of biomaterials like PHA. 4.2. REVIEW STRATEGY: CURRENT LITERATURE REGARDING LCA OF PHA This review was performed using the Scopus search engine, where the following key words and their acronyms were selected: Life Cycle Assessment (LCA), Polyhydroxyalkanoates (PHA), bioplastics, sustainable process index (SPI), waste reduction algorithm (WRA), techno-economic analysis (TEA) and sustainability. The results of the review will be discussed according to: i) LCA methodological issues, ii) process issues, and iii) the studies’ outputs. Regarding LCA methodological issues, a standardized protocol to perform LCA studies of PHA and bioplastics has been recently proposed by the EC and the Joint Research Centre (JRC) of the EC to conduct comparative LCA studies for bioplastics and fossil-based polymers (Nessi et al., 2020), which is based on several widely recognised guidance documents, such as ISO or EN standards. The main principles of this guidelines framed the revision and discussion presented later. Finally, besides the discussion of the main outputs, the referred process issues classification is linked to relevant PHA production features, explicitly: - Substrate employed: ad hoc produced material vs. waste stream. Substrate selection has economic, social, ethical, and environmental implications, but it also partially defines the PHA production process (wastes are generally preferred for MMCbased systems). Moreover, the relevant environmental categories for each system type are linked to the substrate employed (in crop-based systems, land use or eutrophication are of special importance). Therefore, reviewed articles will be classified as waste streams or dedicated crops. - Culture type: mixed culture vs. pure culture. Although there is not explicit correlation between culture type and feedstock selection, these two issues are often linked. Besides, the culture type defines the process configuration and type, so the presented studies will be also categorised regarding the microbiological features of the process, as mixed and pure cultures. 4.3 CRITICAL REVIEW OF THE CURRENT LCA STUDIES FOR PHA PRODUCTION A selection of 60 papers where the environmental profile of PHA production was assessed resulted from the literature search. Most of these papers will be discussed in the following sections. Thus, the references can be found in the literature section of this CHAPTER 4: HOW CAN WE VALIDATE THE ENVIRONMENTAL PROFILE OF BIOPLASTICS? 35 thesis. However, the research works not explicitly addressed but included among the 60 documents are included in Appendix 2. The results for LCA methodological issues and process issues are quantitatively summarised in Table 4.1, and further discussed in the consequent sections of this Chapter. The first document addressing LCA of PHA was published back in 1998 and, since then, two studies were published per year on average, with an important growth in the last two years (i.e., 15 documents in 2019/2020). About 70% of the documents were research papers, and around 20% were reviews where LCA of biopolymers was the main issue, or where the PHA state of the art was assessed and LCA was included. The rest of the documents were 4 books or book chapters (Martinez-Hernandez et al., 2018; Gupta et al., 2019; Medina-Martos et al., 2019; Plackett, 2011), 1 report (Bengtsson et al., 2017b), and 2 methodological documents: one assessing carbon displacement of bioproducts in comparison to conventional materials (Lynd and Wang, 2003), and the JCR-EC report (Nessi et al., 2020). Nevertheless, Heimersson et al. (2014) addressed methodological issues in their review, which also included a case study. Table 4.1 Main quantitative results of the critical review. Results are expressed as the total number of studies (NS) and, therefore, some of values include more than one of the considered features (for example, some studies compared MMC and pure culture performance, thus, they are accounted for within both culture types). 1 The data is expressed as studies only based on previous literature works (secondary sources) and studies based on data generated explicitly for the assessment (primary sources). 4.3.1. Review Outputs regarding methodological issues 4.3.1.1. Motivation of the studies, Functional Unit, and System Boundaries In the first years when LCA-PHA was researched, studies aimed to compare the environmental profile of PHA with the profiles of polystyrene (PS), polyethylene (PE) or PP. Nonetheless, in the last years, studies aim to investigate the most optimal routes to treat industrial streams, so the obtained results provide information about the routes with lower energy requirements or GHG emissions. ALBA ROIBÁS ROZAS 36 On the other hand, the question of the SB is linked to the FU definition. The most common SB were the cradle-to-gate type1 (about 65% of the studies), which means that the EoL of the product is not considered. Thus, only the production process is included in the study, as very few data is yet available about bioplastics EoL. However, as seen previously, JCR-EC method recommends cradle-to-grave2 SB for LCA of PHA, as, when EoL of the product is not included, biodegradability is disregarded, and the obtained results are just partial. It is also truth that fair comparison cannot be established between PHA and petrochemical plastics with the current LCA methodology, as the actual effects of plastic pollution are not yet modelled. Thus, biodegradability and energy recovery from bioplastics are disregarded in the study if the LCA has Cradle-to-gate SB, and the devastating effects of marine and terrestrial plastic pollution are dismissed if Cradle-to-grave approach is applied. This means that, possibly, most of the studies performed to date suffer from bias in favour of petrochemical plastics, so results might not show the actual net benefits of PHA. Moreover, the scientific community must orient their efforts to gather data about the EoL of PHA and the effects of plastic leaks in the environment, as addressed in the work by Changwichan et al. (2018) or in projects like MarILCA (2020). About 75% of the studies considered a FU linked to the final product (normally, mass of polymer or mass of final product, as plastic bags, or boxes). As stablished by Heimersson et al. (2014), Cradle-to-gate SB are enough if the studied systems have the same gate (a certain mass of PHA), and the aim of the study is to assess different bioplastic production routes. However, for assessments with a FU considering the mass of polymer, and where the aim of the evaluation is to compare the environmental performance of biobased and petrochemical materials, the issue of the SB becomes relevant and hard to solve. Moreover, PHA properties are different than the ones of petrochemical plastics, and they depend on the extraction protocol, the monomer proportion distribution and so on. Therefore, the assumption generally made that 1 kg of PHA will directly replace 1 kg of PE, PP or PET is not very accurate and can lead to wrong results. Unfortunately, most of the current studies reviewed follow this approach, and this might be one of the reasons why the obtained results are highly variable and sometimes conflicting. This question needs to be considered and properly addressed, at least until more information about PHA use and EoL and plastic pollution is provided. Meanwhile, the lack of data can be managed by applying conservative replacement ratios of PHA to other polymers (Vega et al., 2019) and/or by performing sensitivity analysis (Roibás-Rozas et al., 2020). On the other hand, a FU referred to the feedstock appears to be the most suitable option to compare different treatment/valorisation systems to produce the same polymer. This means assessments where the aim is not to compare the environmental profile of biopolymers to petrochemical polymers, but to study the performance of different process schemes. Although Gurieff and Lant (2007) considered for the first time the influent as a 1 Cradle-to-gate refers to an LCA that considers the life cycle of a product or process, from raw material acquisition only to the factory gate. Thus, it excludes the use and disposal phases. 2 Cradle-to-grave refers to an LCA that considers the whole life cycle of a product or process, from raw material acquisition through production, use, end of life, and final disposal (ISO, 2006). CHAPTER 4: HOW CAN WE VALIDATE THE ENVIRONMENTAL PROFILE OF BIOPLASTICS? 37 FU more than a decade ago, the standard approach was the opposite for almost each study reported until Morgan-Sagastume et al. (2016) (except for assessments only considering the DSP). This is linked to the process approach in the studied system, which normally was a pure culture fed by a pure substrate, so the aim was to compare the performance of the produced PHA to a conventional material. Consequently, when more process approaches appeared (not only MMCs, but also new production or extraction methods) and wastes started to be the main considered option for PHA production, the trend switched, and the chosen FU was referred to the feedstock, as the aim was not only to produce PHA but to treat a waste. In fact, 5 out of 10 studies reported in the last two years considered the influent as a FU, in opposition to the approach followed in the past decades, also indicating that the goal of these novel process approaches is not only to produce a polymer, but to establish processes that fit circularity by valorising waste streams. 4.3.1.2. Impact categories The most assessed impact category is Global Warming Potential (GWP), as it is expected that bioproducts can mitigate the effect of climate change. Non-renewable Energy Use (NREU) or Fossil Resource Depletion are also commonly evaluated as bioplastics would replace petrochemical materials. Acidification and Eutrophication Potential were also commonly studied, and just a few papers reviewed an important number of the categories included in the ReCiPe impact methodology (Fernández-Braña et al., 2019; Harding et al., 2007; Roibás-Rozas et al., 2020; Vega et al., 2020; Vogli et al., 2020). Again, the lack of data regarding EoL affects to the selection of impact categories, as it is not completely clear how the effects linked to polymer production, use and disposal are transferred on the environmental compartments. On the other hand, the JRC-EC method states that the whole 16 categories addressed in the Product Environmental Footprint Guidelines (Manfredi et al., 2012), so each of them must be considered from now on. In any case, it has been stated that, when the LCA is performed for polymers processed from crops, categories as Land Use, Acidification and Eutrophication must be mandatorily considered (Heimersson et al., 2014). 4.3.1.3. Assessment type More than 60% of the assessments used pure LCA (meaning according to ISO 14040 and 14044), where the most employed software was SimaPro (Gupta et al., 2019). Around 20% combined environmental assessment (i.e. LCA) and cost assessment by adding any economic evaluation (Changwichan et al., 2018; Gurieff and Lant, 2007; Leong et al., 2017, 2016; Martinez-Hernandez et al., 2018) or implementing TEA (Fernández-Dacosta et al., 2015; Medina-Martos et al., 2019; Saavedra del Oso et al., 2020; Vega et al., 2020). Almost 5% of the reviewed papers used Territorial Metabolism-LCA (TM-LCA), (Vega et al., 2020, 2019)which considers regional aspects of the environmental assessments, and almost 10% used SPI (Koller et al., 2013b; Kourmentza et al., 2017; Rathi et al., 2013), which is an index that normalizes impacts according to the availability of planet surface (it calculates the necessary area for industrial processes). Recently, studies started including other sustainability aspects by linking environmental impacts with social and economic effects (Chen et al., 2020; Pérez et al., 2020a; Talan et al., 2020). In fact, the most recent findings have pointed out that the ALBA ROIBÁS ROZAS 38 assessment of bioproducts cannot be linked only to environmental aspects, as suggested by the triple-bottom line of sustainability, which includes environment, economy, and society. It makes sense that, if one wants to evaluate the feasibility of bioeconomy implementation, the concept of sustainability needs to be linked not only to the environmental issues, but also to economic aspects, which are necessarily related to society. Finally, the remaining studies combined LCA with green design principles by scoring products according to proposed Ecodesign metrics (Tabone et al., 2010) or with WRA algorithm, which links waste minimization with a decrease in ecotoxic impacts (Leong et al., 2016). 4.3.1.4. Source of data. Is there enough information to perform accurate assessments? Yates and Barlow (2013) published the first literature review of PHA sustainability assessments (also including PLA and starch-based polymers). They reported that the studies published until that date showed unclear data sources, as some LCAs were performed using confidential information or personal communications (Hermann et al., 2007; Kim and Dale, 2005; Pietrini et al., 2007) and the reader cannot know the features of the studied system. Moreover, assessments performed in the last decade are relying on data provided by studies carried out twenty years ago. To see some examples, see Figure 4.1, where the studies reported on Table 4.1 based on secondary data (dashed lines) were tracked to find the source of information (continuous lines) LCA practitioners need to stop relying in data generated twenty years ago to provide results referring today’s processes. New inventories need to be generated adapted to the novel strategies for PHA production (MMCs, waste streams, or strategies to improve PHA yield). Figure 4.1 Relationship among some of the published LCA-PHA studies and their sources. References framed in continuous lines represent studies presenting original inventories (using primary and/or secondary data), and dashed lines represent the studies derived from these inventories. Otherwise, LCA of PHA is halted and results are going to repeat the same outputs generated in the past, reproducing the same biases. Fortunately, some of the most recent studies are based on the inventories generated explicitly for the assessments, and based on up-scaled laboratory experiences (Nitkiewicz et al., 2020; Pérez et al., 2020b; Righi et CHAPTER 4: HOW CAN WE VALIDATE THE ENVIRONMENTAL PROFILE OF BIOPLASTICS? 39 al., 2016; Roibás-Rozas et al., 2020; Vogli et al., 2020) or pilot-scale experiences (Bengtsson et al., 2017b; Morgan-Sagastume et al., 2016; Vega et al., 2020, 2019). 4.3.2. Classification of the studies regarding process type Among the research works published until 2010, only few considered the use of any waste feedstock. However, in the last decade the trend switched, and most of the studies focused on the valorisation of waste streams, where only two studies addressed PHA production from dedicated crops (Changwichan et al., 2018; Kookos et al., 2019) using secondary data from literature (Akiyama et al. (2003), and Khoo and Tan (2010), respectively, see Figure 3) for the production of the substrate. Regarding culture type, only Gurieff and Lant (2007) and Fernández-Dacosta et al. (2015) considered MMC-based processes until 2015. Nevertheless, and considering the recent trends in PHA production pathways, most of the papers published in the last years evaluated processed based on MMC and/or waste streams. Moreover, it has been pointed out that data sources need to be clear and transparent (Yates and Barlow, 2013), as sometimes data sources are unclear and it is hard to understand features as important as the culture type or the feedstock employed. This becomes an important issue specially for old references and works relying on them (see Figure 3), as some of these studies were based on personal communications or confidential information provided by the company. 4.3.3. Main outputs As mentioned, the results obtained from PHA-LCA studies were sometimes contradictory or conflicting among them, and the reasons might be linked to biases related to the lack of information for the last stages of polymers life cycle (biobased and petrochemical) and allocation choices (Heimersson et al., 2014; Kookos et al., 2019). However, studies agreed at indicating that the environmental performance of the process can be improved by introducing waste streams in the PHA production process (Chanprateep, 2010; Kim and Dale, 2008, 2005; Koller et al., 2013b; Shahzad et al., 2013), and Kendall (2012) proved that, as expected, using them is more favourable than employing dedicated crops. Moreover, the energy source was also important, as green energy usage can switch the environmental profile of the bioproduct (Khoo et al., 2010; Kurdikar et al., 2000; Zhong et al., 2009), suggesting that impacts might be regionally dependent of the energy mix employed (Shahzad et al., 2013). Finally, the most recent outputs highlight specially that the success of the process can only be complete if several aspects of sustainability (economic, environmental and social) are considered (Chen et al., 2020; Koller et al., 2017; Kookos et al., 2019; Pérez et al., 2020b; Talan et al., 2020; Yadav et al., 2020). Results point out that this success will be reached if an integral biorefinery approach is applied (Medina-Martos et al., 2019; Nitkiewicz et al., 2020; Yadav et al., 2020), where all streams are valorised (for example, the generated sludge is converted into biogas by anaerobic digestion). In fact, the latest outputs proved that economic and environmental viability of PHA production facilities can be reached if part of the waste carbon used for biopolymer generation is derived for on-site for energy generation, like the case of anaerobic digestion (AD) (Dietrich et al., 2017; Martinez-Hernandez et al., 2018; Kookos et al., 2019; Vega et al., 2019, 2020; Pérez et al., 2020; Vogli et al., 2020) ALBA ROIBÁS ROZAS 40 4.3.3.1. Full value-chain vision. What do we know about the use and EoL stages? As mentioned, little is known about the last stages of PHA life cycle. On the one hand, it has been considered that impacts derived from the use stage are the same as for petrochemical plastics, so they can be disregarded for comparative LCA with conventional materials (Hermann et al., 2007). On the other hand, it is especially urgent to generate information about the EoL as just 9 out of the 40 papers revised considered it and only 4 papers compared different management routes for the (bio)plastic materials once used. Heyde (1998) concluded that waste bioplastic management might determine the global sustainability of the process, while Khoo and Tan (2010) highlighted that this stage usually has a low impact in relation to the full PHA life cycle. Both research works considered landfilling, composting and incineration, where landfilling showed the worst results and the success of the remaining options depended on factors like energy origin. Hermann et al. (2011) compared AD, incineration, and home/industrial composting to find that AD had generally the best performances and industrial composting the worst. Finally, Changwichan et al. (2018) performed a comparative study which included landfilling, composting, incineration and mechanical recycling, where recycling clearly showed the best results. These outputs indicate that new studies including modern approaches and recent findings are needed, as composting (the most typically considered EoL route) might not be necessarily the optimal alternative. 4.3.4. Lessons learned PHAs are a promising alternative to solve the serious challenges we are currently facing regarding plastic pollution and climate change. However, PHA full-scale production is hindered by the economic feasibility and the lack of consistency concerning its environmental validation, which, as most studies indicated, are linked to i) the use of waste streams as feedstock, ii) employing clean or renewable energy, and iii) applying an integral biorefinery vision, where each stream is valorised (energy production on-site is encouraged). In this sense, the main aspects that need to be urgently solved for this validation are: a) The lack of data. The uncertainties regarding plastic leaks in the environment for petrochemical plastics and EoL pathways for biopolymers are generating a bias in favour of conventional materials, as cradle to gate SB exclude the benefits generated from valorising wasted bioplastics and cradle to grave SB disregard the serious effects of uncontrolled plastic leaks in the environment. These gaps generate conflicting or contradictory results depending, among other factors, on the allocation choices and the hypothesis formulated when trying to fill those data requirements up. b) Sources of information. Although the number of studies performed based on new inventories is increasing, there is still a big number of assessments carried out relying on old information, as seen in Section 4.1.4 (about 60 % of the research works used only secondary data sources from previous studies). This provokes results that are stuck in old technologies and process approaches, as novel strategies for PHA production are not considered, where the reproduction of biases is aggravated. c) Process scale. There is still no full-scale market for PHA, so studies are based on up-scaled laboratory data or pilot-scale systems. This is another issue that can CHAPTER 5: ENVIRONMENTAL ASSESSMENT OF COMPLEX WASTEWATER VALORISATION BY PHA PRODUCTION 47 Table 5.1 Characterization of MCW (VFA is Volatile Fatty Acids, COD is Chemical Oxygen Demand, TOC is Total Organic Carbon, TSS is Total Suspended Solids and VSS is Volatile Suspended Solids) Parameter Units Range pH --- 7.5 ± 0.7 Conductivity mS/cm 37.9 ± 4.7 Alkalinity mg CaCO3/L 225.5 ± 74.2 Total COD g/L 16.6 ± 5.6 Soluble COD g/L 15.3 ± 4.4 VFA g CODVFA/L 0.025 ± 0.045 Carbohydrates g CODCarb/L 12.8 ± 4.2 Proteins g CODprot/L 3.2 ± 1.2 TOC g C/L 6.10 ± 3.52 Inorganic Carbon mg C/L 40.6 ± 32.5 Ammonium mg NH4+-N/L 134.0 ± 75.2 Total Nitrogen mg N/L 987.2 ± 110.0 Sodium Chloride g NaCl/L 18.52 ± 3.94 Sulfate g SO42-/L 2.44 ± 0.70 TSS g/L 3.0 ± 1.6 VSS g/L 1.8 ± 0.8 Regarding wastewater production, a processing facility of high capacity generates about 5 m3/t of mussels processed, of which a 6% is considered of high load (Bello Bugallo et al., 2012). As 100 - 105 kt of mussels are assumed to be processed by the three facilities per year, the centralized full-scale plant for the valorisation and treatment of MCW would have an input flow of 85 m3/day, considering 365 working days per year (Secretaría xeral de Calidade de Avaliación Ambiental, 2013a)3. This is a value significantly lower than the threshold level reported above for economic feasibility. Nevertheless, and considering the expected evolution of this novel technologies, this flow is chosen as the FU for this study, as it is the available flow of the researched feedstock in the target area. This centralized scenario will be compared with the current one, in which MCW is treated in the industrial wastewater treatment plant (WWTP) of each processing facility before being released. When the fish canning facility is located in an industrial area, the effluents are usually disposed to the sanitation network with low requirements regarding discharge limits (Dirección Xeral de Calidade e Avaliación Ambiental, 2008a). One of the problems linked to mussels processing industry is that facilities are usually located in coastal areas, away from industrial zones. Therefore, after wastewater treatment, the effluent is usually disposed directly into the estuaries through a sewage pipe, with variable discharge limits. 5.2.2. System boundaries and scenarios definition The system starts with the MCW generation, so the upstream processes, i.e., the harvesting and canning of the molluscs, are left out the system boundaries. Besides, the treatment and discharge of LLW is also left out as it is a common process for all the scenarios under evaluation. The system boundaries of the present study include all the processes involved in wastewater treatment and valorisation: water line from the influent pumping to the final discharge of the effluent in the receiving water body, sludge 3 The feedstock quantification is based on an annual production so the number of working days will neither change the annual wastewater generated or the total PHA production. ALBA ROIBÁS ROZAS 48 management and disposal, and bioplastic production (extraction of the PHA from the biomass). Sludge is produced from the MCW treatment and valorisation processes and, according to the Spanish legislation, biowaste has to be treated through anaerobic digestion or composting, encouraging administrations to promote the use of the resulting organic fertilizers in agriculture (Jefatura del Estado, 2011). A total of 19 facilities in Galicia are authorised for sludge valorisation, and only three of them work with anaerobic digestion, while the rest of them are mainly compost producers (Xunta de Galicia, 2020). Therefore, transport to the nearest of these facilities and sludge composting is considered for the conventional scenario, while anaerobic digestion at the centralized facility for the valorisation scenario is considered as the most feasible choice. However, due to the lack of information concerning the sludge characteristics, the final application of the compost and the digestate are excluded from the evaluation. For all the stages, only the environmental impacts associated with operation have been considered in the study, while construction and decommissioning of the required infrastructures have been excluded due to lack of data. 5.2.2.1. Baseline scenario: the linear economy approach High strength fish canning wastewater (like MCW) is usually treated in two stages: primary and secondary. In primary treatments, solids and oily matter are removed from the wastewater, while, in the secondary one, biological systems are usually employed to remove dissolved organic matter and nutrients. The choice of the best available technique (BAT) ensures the minimum environmental impact without compromising the economic performance of an installation. BAT selection, which is based on technical feasibility, environmental benefits, and economic profitability, establishes the following treatment sequence for fish canning effluent: first, a primary treatment (equalization, screening, sedimentation, pH adjustment, flocculation, flotation, or microfiltration) and then, a secondary treatment (aerobic or anaerobic). Therefore, the up-scaling of the linear economy scenario will be based on the treatments recommended in the BAT for fish processing (Tomczak-Wandzel et al., 2015), as it is commonly the base for WWTP design in MCW treatment (Mejillones Ría de Arosa, 2016). For primary treatment, the use of Dissolved Air flotation systems (DAF) is widely spread due to their high efficiency and simplicity, and they are used to treat MCW (Barros et al., 2009). When coagulants are employed, its removal efficiency increases to 80 – 95%, since the DAF unit will be able to separate not only solids, but also soluble organic compounds that can precipitate and be removed (Tomczak-Wandzel et al., 2015). In the fish canning industry, it is recommended to operate at a pH approximately of 5 to decrease protein solubility. As MCW have high protein concentration, the use of this unit is justified (Figure 5.2), and its aim is to remove not only solids, but also organic nitrogen associated to proteins. Concerning secondary treatment, biological processes are the preferred option (activated sludge, anaerobic digestion, granular treatments, etc.). Anaerobic digestion has been employed, but the experience has only been fully successful in facilities with low salinity wastewater (Secretaría xeral de Calidade de Avaliación Ambiental, 2013b), as the presence of sodium, ammonia and sulphate can hinder anaerobic treatments. For MCW, these ions are normally present in inhibitory concentrations, and methane production is very difficult. Despite this fact, several full-scale experiences were carried CHAPTER 5: ENVIRONMENTAL ASSESSMENT OF COMPLEX WASTEWATER VALORISATION BY PHA PRODUCTION 49 out in the past in Galician industries trying to valorise MCW anaerobically, but with unsuccessful results where the low amounts of poor quality biogas produced were permanently burned in a torch (Barros et al., 2009). Therefore, most of the industrial plants for fish canning processing employ activated sludge processes for the wastewater secondary treatment, as it is the conventional treatment defined here. Each canning facility is supposed to have a small WWTP for the treatment of their effluents (Figure 5.2); so, assuming equal production capacities, the capacity of each WWTP for the linear economy scenario is supposed to be 1/3 of the total MCW generation, i.e., 28.3 m3/day. Sludge is thickened and stored in an industrial container. Once the container is full, it is transported to the closest industrial composting facility. Figure 5.2. Linear economy scenario. Dashed-line square represents the system boundaries. Conventional treatment process flowchart (bottom part): P1 and P2 represent the first and second aerobic ponds, respectively. Dashed lines represent sludge streams. 5.2.2.2. Circular economy approach The valorisation scenario was designed from the knowledge acquired in previous research projects (see Appendix 3) and the experimental results after two years operating a bench-scale three-stage system for PHA production employing MCW (Figure 5.3). For more information about the lab-scale three-stage system for PHA production, see Appendix 3 and the description provided in Chapter 3 and in Roibás-Rozas et al. (2021a). First, in the acidification stage, COD present in the MCW is transformed into VFA. The acidification reactor and its start-up is described in Fra-Vázquez et al.(2020). After biomass separation by centrifugation, VFA are fed to two reactors: a SBR and a FBR. The SBR is an enrichment reactor for the MMC selection, while the FBR is employed to maximize PHA content in the MMC. Several operational strategies were tested throughout the experimental time (Roibás-Rozas et al., 2021). Optimal conditions were found when a mild alkalinity addition was performed by the supplementation of 80 mg Na(HCO3)/L to the MCW fed to the acidogenic reactor. The enrichment reactor ALBA ROIBÁS ROZAS 50 operated with M-ADF, in which Feast-Famine regime was combined with a settling stage as explained Chapter 3. Figure 5.3. Circular economy scenario. MCW is Mussels Cooking Wastewater. Dashed-line square represents the system boundaries. Innovative process flowchart (bottom part): dashed lines represent high solids content streams. CAS is Conventional Activated Sludge, FBR is Fed-Batch Reactor, SBR is Sequencing Batch Reactor and VFA means Volatile Fatty Acids. Initially, biological nitrification was inhibited in the SBR by allylthiourea addition (20 mg/L of feeding). When its addition was stopped, this biomass was able to produce nitrite during the famine period, with a nitritation rate of 30 mg NH4+-N consumed per g Volatile Suspended Solids and day (NH4+-N/(g VSS·d)). Denitritation occurred during the anoxic periods of feast, where all nitrite was consumed with no significant worsening of the enrichment performance (Fra-Vázquez et al., 2019). In addition to the desired stream, the PHA-rich one, other streams are produced. They are unified in a single stream to be treated together and safely released to the environment. Therefore, a primary settler is required to remove solids present in this waste stream and to work as a homogenization unit to change from sequential operation to a continuous mode. Then, a biological treatment will be employed to remove nitrogen and organic matter in a conventional activated sludge system (CAS), in which aerobic nitrification and anoxic denitrification are performed in a step-fed type system with four anoxic/aerobic sections. Sludge generated in every stage of the process is treated anaerobically in a sludge digester. CHAPTER 5: ENVIRONMENTAL ASSESSMENT OF COMPLEX WASTEWATER VALORISATION BY PHA PRODUCTION 51 The experiences carried out at lab-scale comprise only the valorisation steps, in which an effluent is generated for treatment and disposal, and PHA-rich biomass is produced for extraction. Therefore, experimental data developed in the frame of TREASURE-TECHNOSALT were employed for the up-scaling of the biological valorisation line. For the DSP, data from USABLE PACKAGING project4 were used. Environmental and techno-economic analysis showed that purification of PHA employing alkaline treatment with SDS is a promising choice for PHA extraction from MMC (Saavedra del Oso et al., 2020), so data from the provided inventories are used in the present study. For the selected treatment (SDS and sodium hypochlorite), 99.9% purity was stablished according to Fernández-Dacosta et al. (2015) and an ideal recovery yield was assumed. Concerning the avoided product, PHA properties showed similar specifications to PET, PE, PS and PP (Koller et al., 2013b). PP is similar to PHB polymers, although when other monomers are included in the chain, like hydroxyvalerate (HV), plastic properties change (Palmeiro-Sánchez et al., 2016). Biopolymer composition in TREASURETECHNOSALT project was approximately 25:75 HV:HB. Recent studies show that this type of biopolymer could substitute PET for packaging applications (MelendezRodriguez et al., 2018) so PET is the conventional plastic selected as avoided product from the PHA production. Note that neither the use nor the end-of-life stages for both conventional and bio-based plastics are included in the present study. Finally, the design of the anaerobic reactor for the sludge valorisation was performed according to the results obtained in the frame of PLASTICWATER5 project, where the anaerobic digestion of sludge under brackish conditions and high ammonia and sulphate concentrations was successfully assessed (Palmeiro-Sánchez et al., 2013). 5.2.2.3. Mass balances and stream compositions Table 5.2 summarizes the composition of the main streams of every scenario and compares the achieved effluent with the discharge limits. As there is little information available regarding these limits in mussels boiling facilities, they were established according to the specifications imposed on a fish canning company discharging its effluent directly to the Arousa estuary or ría de Arousa (Dirección Xeral de Calidade e Avaliación Ambiental, 2008b), which is the area where the majority of the mussels boiling facilities are located (Fig 5.1). More information about the design of every process unit and the mass balances performed is supplied in Appendix 3. 5.2.3 Life Cycle Impact Assessment: impact categories and method selection The choice of the relevant impact categories and impact assessment methodology is essential to provide a consistent evaluation of the environmental performance of the scenarios under study. As this research has a circular economy approach, several review papers and articles were examined to identify the most common impact categories assessed in the different areas covered by this study: WWTP operation and management, plastic production/waste and PHA production. Here, the lack of studies regarding MMCbased PHA production is noticeable, and most research works focus on pure culture processes. 4 https://www.bbi-europe.eu/projects/usable-packaging 5 Recycling of wastewater and sludge to produce bioplastic materials (PLASTICWATER). Supported by Spanish Government (CTQ2011-22675) ALBA ROIBÁS ROZAS 52 Table 5.2 Flows and compositions of the most relevant streams in every scenario. (1)For conventional treatment. the system is not centralized. Data shown here refer to the characteristics of each one of the three WWTP. (2)Stream directly to biological reactors for nitrification/denitrification after primary settler. Table 5.3 shows the result of this review. All of them will be covered in this study, paying special attention to the four first ones. For eutrophication, both freshwater and marine will be assessed (EP-F and EP-M, respectively). Besides, and considering the particularities of this study, Marine Ecotoxicity (ME) will also be considered. The Hierarchist ReCiPe 2016 Midpoint method (v1.13) has been selected to evaluate all the selected impact categories except GWP, which was assessed by the last update of the IPCC method (v 1.03 100a). SimaPro v8.3 was used for the computational implementation of all the inventories and impact categories calculations. 5.3. LIFE CYCLE INVENTORY Mass balances were employed to estimate the composition and flows of every stream and complemented with literature data when needed. Conventional treatment units were designed following wastewater treatment and chemical engineering handbooks (Coulson and Richardson, 1999; Metcalf & Eddy, 2014). Theoretical power equations (described in Appendix 3) were used to determine the energy requirements of every stirrer, blower, centrifuge, and pump. Chemicals` dosage was estimated using laboratory results and literature data. Results of all the calculations are summarised in Table 5.4, and detailed information is given in the following sections. 5.3.1. Inventory of the linear economy scenario As already stated, three facilities were considered for boiling mussels. Therefore, the design was individually performed considering a third of the total flux. Then, the linear economy scenario was built up putting together the three facilities. - Design of the process units: The DAF unit removes 85% of solids and, with the use of coagulants at low pH, about 40% of the COD. A volume of 1 L HCl per m3 of wastewater is assumed to be added to lower pH to 5 (and equimolar amounts of NaOH are supplied after DAF to raise pH to neutrality). The removed COD is 80% of coagulated proteins, eliminating approximately a 35% of total nitrogen, and approximately 30% of carbohydrates. After DAF unit, COD/N ratio in the wastewater is high (approximately 12), so two aerobic CHAPTER 5: ENVIRONMENTAL ASSESSMENT OF COMPLEX WASTEWATER VALORISATION BY PHA PRODUCTION 53 ponds are employed to ensure the removal of organic matter, while nitrogen is supposed to be consumed mainly due to cell growth. Mass balances were performed to determine Solids Retention Time and Hydraulic Retention Time (SRT and HRT, respectively) and the number of ponds needed to achieve discharge limits. Short SRT (5 days) guarantees avoiding nitrification in the first pond, so ammonia and COD consumption is calculated by mass balances according to the SRT. A second pond with extended aeration (SRT = 40 days) is needed to achieve COD levels according to legal requirements considering the available ammonia. Due to the long SRT in the second pond, some ammonia oxidation can take place, and it was estimated from mass balances (see Table 5.2). Table 5.3 Categories assessed in LCA studies of WWTP operation, plastic production/waste and PHA production. and its incidence in literature. Light colour indicates relevant category and in dark colour indicates highly relevant categories. Impact Category Field of Interest WWTP operation Plastic/Plastic Waste PHA with MMC Global Warming Potential (GWP) Acidification Potential (AP) Eutrophication Potential (EP) Photochemical Oxidation Potential (POCP) Fossil (abiotic) Depletion Potential (DAR) Ozone Layer Depletion (ODP) Terrestrial Ecotoxicity (TAETP) Human Toxicity (HT) Reference (Meneses et al.. 2015; Zang et al.. 2015) (Antelava et al.. 2019; Koch and Mihalyi. 2018) (Heimersson et al.. 2014; Zarroli. 2020) - Direct emissions to air: Biogenic CO2 emissions related to COD consumption were calculated based on the emission factor reported by Campos et al. (2016): 0.08 kg CO2/kg COD consumed. As COD/N ratio after DAF is about 12, nitrogen is consumed for biomass assimilation with no nitrification. According to the latest IPCC update for greenhouse inventories (IPCC, 2019), N2O emissions are generally expected for aerobic systems. Depending on the stream characteristics and treatment system employed, they can vary from negligible to significant, with an emission factor range of 0.00016-0.045 kg N2O-N/kg N in influent. In the linear economy scenario, as the main consumption of ammonia is related to cell growth, a low emission factor of 0.012 kg N2O-N/kg N is chosen. - Sludge production and management: Sludge generated in DAF units usually has 4% of solids concentration (Metcalf & Eddy, 2014). With 85% solids removal efficiency, sludge generation is 1.8 m3/day. Secondary sludge production was calculated according to SRT and decanters performance, producing 7.7 m3/day of sludge with 7.43 g VSS/L. Sludge is thickened in a gravity thickener to reach a concentration of 45 g Total Suspended Solids/L (g TSS/L) (Metcalf & Eddy, 2014) and then transported to external facilities for further management. Chemically clarified supernatant is mixed with LLW, and as its contribution is less than 1.5% of its flow (6.33 versus 430 m3/day), the influence of this stream on the LLW treatment can be disregarded. After that, sludge is composted together with other waste streams (41% average of sewage sludge as feedstock (TEN, 2020) was considered). ALBA ROIBÁS ROZAS 54 Table 2.4 Reference flows and Life Cycle Inventory for the linear and circular economy scenarios. Reference flows (m3/day) Stream Innovative Conventional(1) Influent 85.00 28.33 Effluent 67.66 22.23 PHA-rich stream(2) 8.42 - PHA (kg/day) 64.30 - Downstream wastewater 3.02 5.64 Primary and centrifuge sludge 2.32 1.80 CAS sludge 6.82 4.30 Thickened Sludge 4.75 2.88 Life Cycle Inventory (units/FU)(3) Inputs Chemicals (kg) Triethylene glycol (L) 15.91 - Ferrous Oxide 6.47 - Sodium Bicarbonate 6.80 - Sodium Hydroxide 80.0 42.00 Sodium Dodecyl Sulphate (DSP) 1.29 - Sodium Hypochlorite (DSP) 221.8 - Hydrochloric Acid - 100.30 Electricity Consumption (kWh) 855(4) 1250 DSP Energy (kWh)(4) 459.5 - Transport (tkm) 510.50 3.1(6) Outputs Sludge Production (m3) 4.75 8.63 Liquid Effluents (m3) 67.93 66.68 COD (kg) 0.03 0.03 Nitrogen (kg) 2.8 1.9 NO3 (kg) 1.5 0.0 NH4+ (kg) 1.3 1.9 Gaseous Emissions (kg) Biogenic CO2 50.05 39.10 Non-Biogenic CO2 1.8 - N2O 1.62 0.91 (1) Results for individual facility (2) Containing 18.40 g Volatile Suspended Solids with 41.50% PHA (3) Overall results for the three facilities together (4) Electricity consumption not considering DSP (5) Energy consumption includes electricity and cooling energy (6) Transport of sludge is not accounted because it is included in the Ecoinvent composting process (only transport of chemicals to the plant in included) 5.3.2. Inventory of the circular economy scenario The centralised scenario that treats the total MCW flow was up scaled from the data obtained in the lab. Theoretical equations were used to calculate power needs, the operating time of the equipment, etc (see Appendix 3). - Design of the process units: In the valorisation line, the up scaling was performed considering the obtained results. The effluent of the acidification reactor, which operated at SRT = HRT of 6.5 days, had 0.6 g CODVFA/g COD. The SBR consumed 1.21 g CODVFA/g biomass to reach 4 g VSS/L, and FBR needed 2.98 g CODVFA/g biomass to achieve 41.5% accumulation capacity. Overall, process efficiency was of 0.2 kg CODPHA/kg CODMCW. In the treatment line, to ensure nitrifiers growth, SRT in the step-fed CAS was set at 15 days and HRT in the anoxic volume was 58.83 m3. Antifoaming agent (0.5 mL triethylene glycol/L feeding) was supplied in the SBR and NaOH was used for pH adjustment in the SBR and FBR (0.75 and 1 g NaOH/L, respectively), based on laboratory-scale dosages. CHAPTER 5: ENVIRONMENTAL ASSESSMENT OF COMPLEX WASTEWATER VALORISATION BY PHA PRODUCTION 55 - Direct emissions to air: Direct emissions of the acidogenic unit were estimated from lab measurements and literature data. Waste mixtures containing carbohydrates and proteins in acidogenic fermenters generated about 279 mL biogas/g VSS, with 47% of CO2 (Alibardi and Cossu, 2016). No methane was detected in several years of operation, and 2.4% of the reactor headspace was H2S because of the high sulphate concentrations in MCW. For this reason, ferrous oxide pellets, which are by-products of metallurgic industry, are used to remove H2S with a ratio of 0.6 g H2S/g pellet (Allegue and Hinge, 2014). When pellets cannot be used anymore, they are employed as raw material for roads and bricks. Fossil CO2 is produced due to alkalinity addition, and CO2 emissions were calculated for operational temperature and pH. For the aerobic reactors of the valorisation line, biogenic CO2 generation due to COD consumption is 0.08 kg CO2/kg COD (Campos et al., 2016) and due to PHA generation is 30% (C-mol PHA/C-mol CO2) (Jiang et al., 2011). Regarding N2O emissions, as nitrification and denitrification take place in the circular economy scenario, more N2O is supposed to be generated. According to the latest IPCC update for greenhouse gas emissions (IPCC, 2019), the average value for the emission factor is 0.016 kg N2O-N/kg N in influent for aerobic systems (as emissions for anaerobic units are not expected). This factor will be considered for the circular economy approach, assuming that nitrogen removal takes place not only in the anoxic tank of CAS but also during SBR feast. - Sludge production and management: Sludge is produced in the centrifuge after the acidification unit, in which most of the solids are removed and 1 m3/day of sludge with 25% TSS is generated (Elías, 2012). Primary settler after the valorisation line removes 55.75% of solids and 33.96% of COD, according to theoretical equations (Metcalf & Eddy, 2014), and produces 1.3 m3/day of sludge with 5% TSS. CAS unit generates sludge according to SRT and decanter performance, producing 4.0 m3/day with 6.90 kg VSS/m3, which is thickened in a gravity thickener to reach a final concentration of 45 g TSS/L (Metcalf & Eddy, 2014). The clarified stream from the thickener is diluted and treated together with the LLW6, and the concentrated stream is treated together with the other high-solid streams in an anaerobic digester. Regarding biogas production, sludge valorisation under high concentrations of sodium, sulphate and ammonia is feasible after proper biomass acclimation and process control. Moreover, sludge digestion generated 280 L of biogas/kg VSS (with 58% of methane) when the digester feeding composition contained 6.7 g Na+/L and 1.5 g SO42per liter (Palmeiro-Sánchez et al., 2013). As these conditions are like the ones that could be found in this digester, the obtained results will be employed for the design of the digestion unit. Finally, biogas valorisation is performed considering electric and heat efficiencies of 35% and 45%, respectively (Metcalf & Eddy, 2014): i.e. 133.9 kWh/day of heat energy and 104.2 kWh/day of electricity. The former was used to maintain the temperature of the digester at 38 ºC and the latter was sent to the grid and considered then as avoided product. An amount of 1.5% of the produced biogas is leaked, while the presence of N2, H2S and NH3 in the biogas is 6 As the supernatant only represent less than 0.5% of the LLW flow (5.64 m3 versus 1290 m3), this addition is considered negligible and stays out the system boundaries. ALBA ROIBÁS ROZAS 56 estimated in 1%, 0.05% and 0.01%, respectively (Rodriguez-Verde et al., 2014). Emissions linked to biogas combustion are calculated from literature data (Paolini et al., 2018), and ferrous oxide pellets are employed for sulphide removal before valorisation. - Downstream Process: DSP design is indicated in Fernández-Dacosta et al., 2015 (surfactant-hypochlorite method). Wastewater generated in DSP is diluted with LLW for its safe disposal (contribution of less than 0.5% in volume) and therefore left out the system boundaries7. - Energy integration: Following a circular economy approach, energy integration is also applied. The temperature of MCW leaving the boilers is 125 ºC, and the bioreactors in the valorisation line operate at 38 ºC and 30 ºC, for the acidogenic reactor and the SBR-FBR respectively. Currently, this heat is dissipated naturally in the facilities, so an estimation of the available heat was done (1,000 kWh/day). It can cover the heat requirements of the biological reactors (estimated to be around 500 kWh/day) as well as the heat demand of the DSP (about 100 kWh/day). Therefore, the heat contained in MCW after the boilers is enough to cover the heat used in the process units. 5.3.3. Background processes Data for the processes of the background system (production of electricity, chemicals, transport, plastic materials, steam and composting process) come from the Ecoinvent v3.3 database (Weidema et al., 2013). The Spanish electricity mix has been adapted with the most recent update reported, referred to 2018 (Red Eléctrica Española, 2019). No environmental burdens are allocated to the antifoaming agent (triethylen glycol) as it is a by-product of ethylene oxidation, neither to the ferrous oxide pellets as they are considered waste from the metallurgic industry. Therefore, only the transport of these products is considered. In the case of ferrous oxide, as it is employed as a component in the production of roads and asphalt, transport to the recycling facility (asphalt producer) is also included. As the SDS used at the DSP is not available in the database, its background production process has been assimilated to the one of alkylbenzene sulfonate. 5.4 LIFE CYCLE IMPACT ASSESSMENT 5.4.1. Individual environmental performance of each scenario The contribution of the different elements to the set of impact categories for each scenario is presented in Figure 5.4. For the linear economy scenario (Figure 5.4a), electricity has an important contribution for all the categories (between 12 – 68%) except for EP-M. Besides, chemicals consumption (which also includes the transport from the suppliers) also has a relevant impact (13 – 87%, except for EP-M), mainly due to hydrochloric acid production. This process affects HT, TAETP and ME, and direct emissions of chlorine impact on GWP. Composting (sludge management) process has significant effects on GWP, AP and POCP due to the expected emissions of CH4, NH3 and N2O. 7 Same criteria than the one applied for the supernatant of thickener, with a DSP wastewater flow of 3.02 m3/day CHAPTER 5: ENVIRONMENTAL ASSESSMENT OF COMPLEX WASTEWATER VALORISATION BY PHA PRODUCTION 63 The present research work focuses on one single type of wastewater, the MCW. However, as stated in section 1, Galicia is responsible for about 80% of the Spanish fish canning production. Once the production of MMC-based PHA under high salinity is validated from a technical and environmental point of view, the possibilities are huge. One single fish canning facility can produce about 3,300 m3 of wastewater per day (Dirección Xeral de Calidade e Avaliación Ambiental, 2008a), so assuming 6% of high strength effluent each facility could produce about 50 m3/day of a PHA-rich stream (containing 380 kg PHA) on-site. Supposing 300 working days per year, and knowing that 7 of the top ten fish canning companies (in production and incomes) are located in Galicia (Eleconomista.es, 2018), the annual regional potential is of 911 t PHA/year, which is near the minimum value reported by Bengtsson et al. (2017) for economic feasibility. Four of these fish canning plants are separated less than 25 km, although the longest distance among facilities is of 100 km. Therefore, DSP can be centralised for an optimized environmental and economic performance of bio-based products, opening an unexplored scenario where technical, economic and environmental issues need to be studied. Finally, one challenge that is being faced by plastics and bioplastics industry is related to end of life options (Geyer et al., 2017). Due to the lack of information, this study has excluded them. Recent research works point out that mechanical recycling is preferable to composting, incineration or landfilling for bio-based materials (Changwichan et al., 2018). However, the effects of plastic litter in the water bodies, especially in the ocean, are still unknown, and the actual effects of plastic pollution worldwide are uncertain. Initiatives such as Plastic Leak Project8 or MarILCA9 will allow for better metrics to account for plastic leakage and to integrate potential environmental impacts of marine litter, especially plastic, in LCA results. 5.6. CONCLUSIONS The renewable origin of bio-based plastics does not convert them automatically in an environmentally sustainable products and, even when is clear that they are promising materials, the environmental assessments comparing conventional plastics and bioplastics do not always show explicit benefits. This study presents a detailed analysis of the environmental performance of PHA production from industrial wastewater, an area in which the LCA studies are still scarce. The current linear economy approach for mussel cooking wastewater management (generation – treatment – discharge) was compared with a circular economy approach, where PHA is produced, and wastewater is safely treated and discharged. A better environmental performance was reported for nine out of the ten impact categories under evaluation, which reflects the potential benefits of this novel approach. Process optimization through innovative systems (like anammox reactors for nitrogen removal) come along as preferred choices to improve overall performance (effluent quality and energy optimization). The potential of bioplastics production processes is huge, although case-by-case studies are needed, evaluating the local industries and waste streams availability, as well as PHA and biogas yield for every feedstock. This study showed that the feasibility of 8 https://quantis-intl.com/strategy/collaborative-initiatives/plastic-leak-project/ 9 http://marilca.org/ ALBA ROIBÁS ROZAS 64 producing value-added bioproducts under a circular economy approach is linked to the integral vision of the process, where all streams are valorised, and energy use is optimized. Finally, the potential of Galician region for bioplastics production was estimated near 1 kt PHA/year, near the threshold value for economic feasibility. Accordingly, centralized extraction facilities employing novel extraction protocols could be the key for the environmental validation and the economic viability of these products, stepping towards high-scale PHA production from mixed microbial culture. CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 65 6. ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNO-ECONOMIC AND ENVIRONMENTAL PERSPECTIVE SUMMARY As the last step of the technology readiness increase for the processes developed, economic performance was included in the assessment. Industrial effluents (Mussels Cooking Wastewater (MCW) and oily Fish Processing Wastewater (FPW)) were valorised by polyhydroxyalkanoates (PHA) and triacylglycerides (TAG) production using mixed microbial cultures (MMC). First, the strategies carried out at lab-scale (three years of operation) were up-scaled and compared, at environmental and economic level, to the systems currently used at full-scale by the industries in two stages. In this first stage only the wastewater treatment processes were compared (downstream process (DSP) for biomaterial recovery is not included) to choose the best strategy for the valorisation of each effluent. Then, the scenarios with the best performance are chosen for the second stage of the study. Here, a centralized biorefinery scheme is proposed where ten, twenty, and fifty of these small facilities are clustered to work as raw material (PHA and TAGrich biomass) providers for a centralized DSP factory. This scenario is compared with the current one, where factories only treat, but do not valorise waste effluents. For the first stage (i.e., excluding DSP), the cost of MCW treatment could potentially decrease from 0.53 €/kg COD (treatment only) to 0.48 €/kg COD (when PHA-rich biomass is produced). For FPW, cost would decrease from 0.52 (treatment only) to 0.30 €/kg COD (with TAG/PHA production). Environmental impacts would decrease about 30% for FPW in some relevant categories as climate change, while for MCW the environmental validation would be linked to an optimization in the chemical’s usage and in residual streams management, like treating waste solids through anaerobic digestion. Then, under the centralized biomaterial extraction approach (the second stage of the studio), PHA production cost could be 0.95 – 1.18 €/kg PHA for MCW, which is a competitive market value. For FPW, the cost of biomaterials production increases (1.56 – 3.35 €/kg PHA and 6.75 – 9.21 €/kg TAG) due to the lower biomass concentration achieved here. Concerning to the environmental performance, the replacement of petrochemical materials by the bioproducts results in environmental credits for MCW under a centralized biorefinery approach, while the validation is less clear for FPW, where the generation of biomaterials is again lower. This chapter proves that it is economically and environmentally feasible to manage the treatment of industrial effluents from the food industry under a circular economy approach, where centralized biorefinery schemes show results that yield in competitive product costs and decreased environmental impacts for PHA production systems based on wastewater valorisation. This chapter was written as a result of a research stay performed in the department of biotechnology at TU Delft (The Netherlands) under the supervision of Prof. John Posada. This research stay was financially supported by CRETUS grants for research stays 2021, which are funded by Collaboration agreement with the Ministry of Culture, Education and University for the development of strategic R&D actions in 2021. ALBA ROIBÁS ROZAS 66 6.1. INTRODUCTION Once process development and preliminary environmental evaluation have been carried out, it is necessary to assess the economic performance of the developed technology. Therefore, data from Chapter 3 on PHA production (Roibás-Rozas et al., 2021), and from other research activities carried out in the framework of TREASURETECHNOSALT project on PHA plus TAG production (Argiz et al., 2020b), will be used with the aim of validating the economic plus environmental performance of these circular economy-based processes to increase technology readiness level. Studies assessing the production of biomaterials normally identified the environmental and economic burdens of the process in the DSP and the cost of raw materials (Kumar et al., 2020). By introducing MMC-based processes that use waste streams, the burden of the raw materials acquisition is addressed. However, DSP still limits the feasibility of up-scaling these systems (Bengtsson et al., 2017a; FernándezDacosta et al., 2015). Most of the studies performed till date (Roibás-Rozas et al., 2022b) assessed the same type of process scheme, where raw materials, coming from pure or waste sources, are transformed into PHA-rich biomass in a big-scale biorefinery1. Of them, only some included a DSP on-site (Fernández-Dacosta et al., 2015; Roibás-Rozas et al., 2020) while others stopped at the biomass production stage (Morgan-Sagastume et al., 2016). However, DSP on-site is only feasible for factories of high capacity. Therefore studies including on-site DSP normally work with PHA productions of around 103 ton per year (Bengtsson et al., 2017a; Crutchik et al., 2020; Fernández-Dacosta et al., 2015; Gurieff and Lant, 2007), which needs large wastewater flow generations that are not representative of many food processing plants, yielding in costs of 1.5 – 2.5 €/kg PHA. This is, in fact, the case of Galician fish canneries, where many small factories are responsible for the main share of the production (González-López, 2018). Therefore, to increase the feasibility of implementing circular economy strategies in facilities with small or moderate production capacities, a different approach is needed. The present chapter aims to assess the techno-economic and environmental performance of a biorefinery scheme to produce biomaterials from fish processing wastewater that includes a centralized set-up for biomaterials extraction. Here, data generated in the framework of TREASURE-TECHNOSALT project to produce biomaterials are used to design and up-scale several scenarios where the organic matter present in waste streams (Mussels Cooking wastewater or MCW, and Fish Processing Wastewater or FPW) is valorised into PHA and TAG. Then, based on techno-economic and environmental criteria, three circular economybased configurations will be selected to be further analysed: one for PHA production from MCW, one for PHA production from FPW, and one for TAG and PHA production from FPW using MMC. Moreover, a scenario where biomaterials are not produced, but innovative technologies developed in the framework of the present project are used, is assessed as well. Here, the selected scenarios will be assessed to test the economic and environmental feasibility of a centralized biorefinery scheme where extraction of PHA 1 It is worth mentioning that, to the best of our knowledge, studies have not been carried out till date for processes producing TAG with MMC. CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 67 and/or TAG is not performed on-site but in a centralized plant. Therefore, the analysis is divided in two stages: Firstly, data from laboratory scale experiments are used to upscale different scenarios for PHA and TAG-rich biomass production, with two goals: i) to compare, at an environmental and techno-economic level, the current systems used by the industry to treat the complex wastewater with the innovative ones, and ii) to select the best configuration (i.e., best environmental and economic performance) to be used for the valorisation perspective at the next stage. Note that only wastewater processing (for treatment or for valorisation) is considered in the first stage of the study, so DSP is not included. Secondly, the selected valorisation configurations are used to assess the environmental and economic profile of a wastewater valorisation strategy where WWTPs act as raw materials providers for centralized biorefineries where DSP is carried out, as suggested in the previous chapter and elsewhere (Morgan-Sagastume et al., 2016). Therefore, industrial WWTPs produce PHA/TAG-rich biomass simultaneously to the wastewater treatment, and this biomass is treated in a centralized for biomaterials extraction. To do so and taking into account the characteristics of the sector in our region, ten, twenty, and fifty facilities for food processing have been defined as the three centralization scales to be evaluated for the centralized biorefinery scheme. Therefore, on the one hand, the environmental and economic cost of switching conventional by innovative treatments for wastewater will be assessed. On the other hand, the cost of biomaterials production under the centralized scheme will be estimated, to evaluate if these materials achieve competitive market prices. 6.2. MATERIALS AND METHODS 6.2.1. Discharged effluent quality and WWTPs design considerations The discharge limits of the effluents generated from all the evaluated treatment and valorisation alternatives are the ones stablished according to the local legislation (Boletín Oficial de la Provincia de A Coruña, 2016, see Appendix 4). Besides, chemicals are used in some solid-liquid separations, so when the use of chemicals is indicated, the dosage of coagulant is 50 mg FeCl3/kg TSS for decanters, and 3 mg polyelectrolyte/kg TSS for centrifuges, to generate sludge cakes with TSS concentrations of 10% and 25%, respectively (Elías, 2012; Metcalf & Eddy, 2014). 6.2.2. System and scenarios definition in the first stage Two long-term experiments (Chapter 3 and Argiz et al., 2020b) were carried out at lab scale with two goals: to develop more efficient treatment systems for complex industrial effluents, and to valorise the organic carbon in the waste streams into valueadded materials (PHA and TAG). The effluents from two industrial fish processing facilities located in Galicia (NW of Spain) were used. As they had different origin and composition, two typologies of residual streams were stablished: non-oily effluent (MCW) and oily effluent (FPW). The strategies applied for their management were also different, as well as the bioproduct(s) obtained. As a result, two systems were here defined and evaluated according to the feedstock used (Figure 6.1): ALBA ROIBÁS ROZAS 68 • System A (SA): mussel cooking wastewater (MCW), characterized by relatively high concentration of salt, nitrogen, and COD, which contained no oil (see the characterization of MCW in Table 5.1 in Chapter 5). • System B (SB): fish processing wastewater (FPW), that came from a facility where mainly tuna is cooked and contained relatively high fats and oil concentrations and moderate concentrations of salt, nitrogen, and COD (see the characterization of FPW in Appendix 4) Then, several scenarios were defined for both SA and SB according to the process approach and configuration (Figure 6.1). Here, the process approach refers to the treatment strategy: • Linear economy-based scenarios (i.e., conventional): SA0 and SB0, for MCW and FPW, respectively. • Circular economy-based scenarios (i.e., innovative): SA1 to SA4, where MCW is valorised into PHA, and SB1 to SB5, where the oil fraction of FPW is separated from wastewater and then valorised into PHA and TAG, while the remaining wastewater is treated using innovative treatments such as aerobic granular sludge (AGS). Finally, in scenario SB6 the oil fraction is not biologically valorised, but sent to biodiesel production, while the wastewater is treated using AGS (thus, SB6 performs an innovative treatment, but it is not based on circular economy). Figure 6.1. Studied scenarios at the 1st stage. The mussel represents the Mussels Cooking Wastewater for System A (MCW for SA), and the tuna fish represents the Fish Processing Wastewater for system B (FPW for SB). 6.2.3. Description of scenarios of System A (SA): PHA production from MCW - Linear economy-based scenario (SA0) SA0 (Fig. 6.2) represents the current treatment strategy applied by the industry. Its design is based on available information (Barros et al., 2009; Bello Bugallo et al., 2012) and it is similar to baseline scenario of Chapter 5 (Roibás-Rozas et al., 2020). CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 69 Figure 6.2. Layout of the WWTP for MCW and main streams involved. Dashed lines represent streams with high solids concentration and continuous lines represent liquid streams. Firstly, the primary settler removes solids, enhanced by the FeCl3 supply. Then, there is a homogenization tank prior to a CAS unit, which has one anoxic chamber and four aerobic ponds and a total volume of ca. 1,200 m3. The primary and secondary sludge are dehydrated in a centrifuge, and the resultant supernatant is returned to the homogenization tank. Dehydrated sludge is here composted, the most common sludge management alternative for food industrial sludge in the region (Roibás-Rozas et al., 2020). Fig. 6.2 represents the WWTP setup and labels the key stream flows that are considered in the study. - Circular economy-based scenarios The valorisation section of the scenarios SA1 to SA4 for MCW (Figure 6.3) was based on the lab-scale results described in Chapter 3, which used a typical three-stage process (wastewater acidification, MMC enrichment and biopolymer accumulation) for PHA production. It was up-scaled after the results of that chapter (Roibás-Rozas et al., 2021) where four out of the six operational strategies tested there resulted in stable experimental periods (so SA1, SA2, SA3, and SA4). Figure 6.3. Layout of the circular economy based WWTPs for MCW and main streams involved. The grey lines represent streams that may exist or not depending on the operational scenario (see Appendix 4 for details of the scenarios based on the operational stages defined in Chapter 3). Dashed lines represent streams with high solids concentration and continuous lines represent liquid streams. The names of the stream flows are used to perform the mass balances explained in supplementary materials. The scenarios were here classified regarding the strategy used to enrich the MMC in the SBR (Table 6.1), and the pH and alkalinity conditions in the acidification-CSTR (HRT ALBA ROIBÁS ROZAS 70 = 6.25 days), where the remaining waste streams are treated in a CAS. CAS and CSTR sludge are thickened in a centrifuge and treated externally by composting. Here, as the operational stages are defined according to the enrichment conditions for the obtainment of the MMC (the pH and the enrichment strategy), some streams are variable for the different stages (grey lines in Fig. 6.3). Process configurations, mass balances performed to design the reactors’ volume, streams’ flow, and information of each scenario are described in detail in Appendix 4. 6.2.4. Description of scenarios of System B (SB): PHA and TAG production from FPW - Linear economy-based scenario SB0 (Fig. 6.4) represents the current treatment of the oily effluent, according to the public records of the local government (Secretaría xeral de Calidade de Avaliación Ambiental, 2013a). The treatment starts at the DAF unit, that removes about 95% of the fat and 80% of the solids in the FPW. On the one hand, the remaining wastewater is treated in a 5,000 m3 CAS biological reactor divided in eight chambers (two anoxic and six aerobic) with internal recirculation. On the other hand, the DAF sludge is settled in 1 m3 tanks to remove the oil that naturally floats. Then, that oil is sent to an external facility where it is transformed into biodiesel. Then, the remaining non-oil DAF sludge and the centrifuged CAS waste solids are sent to another external facility that centralizes sludge treatment by anaerobic digestion. Figure 6.4. Layout of the linear economy based WWTPs for FPW and main streams involved. Dashed lines represent streams with high solids concentration and continuous lines represent liquid streams. The names of the stream flows are used to perform the mass balances explained in supplementary materials. - Circular economy-based scenarios The oily fraction of the FPW is separated in a DAF unit as in the linear scenario (Fig. 6.5). Then, the fat-free wastewater is treated in a sequencing batch reactor based on AGS (fully described in Appendix 4), and the oily fraction is valorised into PHA and TAG using an MMC, as it was recently proven by Argiz et al. (2020b) that mixed cultures feed with oil can store both compounds. Solids washed out from the AGS are decanted in a settler without additional coagulants, and then both AGS washout and DAF waste solids are concentrated in a decanter using chemicals, while supernatant is treated by the AGS. Then, the thickened solids are treated externally by anaerobic digestion. Fig. 6.5 represents the setup of the circular economy based WWTPs that valorise FPW and labels the key stream flows required to perform the mass balances. More details about mass balances and process configuration can be found in Appendix 4. CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 71 Figure 6.5. Layout of the circular economy based WWTP for FPW and main streams involved. Dashed lines represent streams with high solids concentration and continuous lines represent liquid streams. To facilitate the comprehension of the figure, waste streams that are returned to the main process line (the AGS feeding) are marked in red. The dilution water (recirculated from the outlet) is marked in green, and oil streams are highlighted in yellow. Table 6.1. Summary of all the scenarios analysed in the first stage of the evaluation, including circular and linear economy-based scenarios for both systems, and the main differences in the valorisation sections and the enrichment strategies. Scenario Approach Wastewater type Enrichment type Modification in the ADF and stage variation SA0 Linear MCW Does not apply - SA1 Circular MCW ADF No Low pH SA2 Circular MCW M-ADF1 Settling stage Low pH SA3 Circular MCW M-ADF Settling stage Low pH and high load SA4 Circular MCW ADF No High pH SB0 Linear FPW Does not apply - SB1 Circular FPW ADF No SB2 Circular FPW M-ADF Uncoupling C/N feeding Dilution water addition along with the nitrogen SB3 Circular FPW M-ADF Uncoupling C/N feeding Dilution water addition along with the carbon SB4 Circular FPW M-ADF Uncoupling C/N feeding Low pH SB5 Circular FPW M-ADF Uncoupling C/N feeding Neutral pH SB6 Innovative FPW Does not apply - 1 M-ADF: Modified Aerobic Dynamic Feeding As mentioned above, FPW was separated in two fractions, an oily fraction that was valorised, and remaining wastewater that is treated, so the scenarios SB1, SB2, SB3, SB4, ALBA ROIBÁS ROZAS 72 and SB5 are defined according to the process configuration in the valorisation section (Table 6.1) as described by Argiz et al. (2020b). Here, the hydrolysis of the fats into long chain fatty acids plus the enrichment of the MMC are carried out simultaneously in the enrichment SBR, and the hydrolysis plus the biopolymer accumulation are carried out simultaneously in the accumulation FBR, avoiding the need for the acidification step. Finally, when no PHA/TAG are produced (SB6), no biological valorisation or extraction section are required. Here, oil is managed externally by biodiesel production as in the conventional scenario SB0, but wastewater is treated by AGS instead of CAS, so a comparison between the innovative AGS process and the conventional CAS-based process can be performed. 6.2.5. Approach in the second stage of the evaluation As a result of the defined scenarios assessment, the scenarios that showed a better techno-economic and environmental performance in the first stage of the studio are selected for the second stage. The selected scenarios will be analysed under a centralized biorefinery approach at three different scales (ten, twenty, and fifty times bigger). These scenarios are: PHA production using each of the influents (so one for PHA production using MCW, and one for PHA production using FPW), and PHA plus TAG production using FPW. Moreover, as FPW is divided into two fractions (oil to be valorised, and wastewater to be treated), a scenario where biomaterials are not produced, but innovative technologies are used, is also assessed (scenario SB6). Under this centralized approach, the cost of wastewater treatment (while producing PHA or TAG-rich biomass) is the result of the economic analysis performed in the first stage of the studio. Consequently, the cost of PHA production is attributed only to the DSP. Therefore, in the second stage of the assessment, the cost of PHA/TAG extraction in a centralized facility where DSP is carried out will be evaluated, considering that the biomass provided is the one generated in the facilities selected in the first part of the assessment of this study. Here, the performance of a centralized DSP that extracts PHA/TAG from ten, twenty, and fifty factories that produce biomass is analysed. Finally, and regarding SB6 (where oil fraction is not biologically valorised, but remaining waste effluents are treated under an innovative approach using AGS reactors), its environmental performance is also studied, but its economic performance is not, as neither PHA nor TAG are produced here (the cost of wastewater treatment for SB6 is the result of the first stage of the assessment). 6.2.6. Features of the processes selected for the second stage 6.2.6.1. Waste treatment For the processes in the first stage of the evaluation, the waste solids generated in the WWTPs were treated by composting for SA scenarios, and by anaerobic digestion for SB, as it corresponds to the current reality in these factories. Here, and for the sake of cross-comparison, it is assumed that waste sludge is always treated by anaerobic digestion. Besides, it is assumed that the wastewater generated in the extraction process is treated at the DSP biorefinery by a standard primary treatment. 6.2.6.2. DSP for the systems extracting PHA only Data obtained from the TEA of different DSPs (Saavedra del Oso et al. (2020)) were used to choose the optimal extraction route: sodium dodecyl sulphate (SDS) at low CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 79 sludge for these facilities is treated suing this technique. As the conventional system SA0 produces more waste sludge, sludge management increases utilities costs. Note however that, although SA0 produces more sludge, the one with higher waste management costs is SA4 because, the produced sludge has a higher solids concentration due to the use of chemicals in the process. Therefore, the cost of treating MCW under a linear economy approach (so SA0) is 0.53 €/kg COD, while treating it using circular economy approaches yield in 0.48 – 0.61 €/kg COD (Figure 6.8). Note that, for these results, the benefits of PHA production are not accounted yet (DSP is not included). Therefore, the potential economic success of the circular economy-based process is not relying on the PHA value, but in the improved system configuration proposed. Excluding scenario SA4, where the use of chemicals for pH control in the acidification CSTR turns into high operational costs, the cost of treating wastewater under circular economy approaches that have a wise use of chemicals (SA1, SA2, and SA3) yields in 0.49 ± 0.01 €/kg COD. The scenario with the lowest cost is SA3, with 0.48 €/kg COD. 6.3.2.2. System B (SB) with Fish Processing Wastewater (FPW) For SB, only a fraction of the wastewater (the oil) is biologically valorised into PHA/TAG for scenarios SB1-SB5. Therefore, the non-oily wastewater is only treated, and not valorised, using an innovative treatment (AGS) for SB1-SB5, but also for SB6. For SB6 and SB0, where oil is not biologically valorised, it is transformed into biodiesel. Here, the reactors used under an innovative approach have a volume smaller than the one used under the conventional approach (see Table 6.3) due to the lower residence times linked to the operation of the innovative AGS reactors compared to the CAS ones for this industrial wastewater treatment (AGS has a residence time of 10 hours, while the CAS has a residence time of 7.1 days). However, scenarios SB1-SB5 have more process units and pumps. In this sense, there is no CSTR unit for SB, as the oil is hydrolysed in the SBR and FBR. However, there is an AGS that also requires of high amounts of aeration. Although it is considered that the AGS would have higher oxygen needs per volume of reactor than the CAS one, the CAS reactor has a volume of 5,000 m3, while the AGS has a volume of around 360 m3. Consequently, operating the AGS is still cheaper than operating the CAS, despite the higher oxygen needs per volume of reactor in the AGS than in the CAS. On the other hand, SB0 and SB6 have operational costs higher than the circular economy-based scenarios (Table 6.5). In the case of SB6, indirect costs are approximately the same as for circular scenarios SB1-SB5, since the cost of waste management (including oil treatment) is very high, but electricity needs are lower. Nevertheless, for the conventional scenario SB0, not only the costs of oil and waste sludge treatment are high, but also the energy demand is very high due to the aeration needs of the biological reactor. Circular economy-based scenarios are economically more attractive than the current ones (Figure 6.9). However, valorising oil on-site is still more expensive than treating it externally (normally, for biodiesel production as done by SB6), as the cost of the biological valorisation of the oil into PHA/TAG is around 1,100 €/m3, while the current ALBA ROIBÁS ROZAS 80 cost of oil management is around 500 €/m3 (Illukpitiya and de Kof, 2014) (thus, approximately 0.48 €/kg CODOIL for biological valorisation versus 0.22 €/kg CODOIL for biodiesel). Table 6.5 Equipment purchase cost of the WWTPs equipment according to their sizes, and main indirect costs for the scenarios with FPW. All direct costs are in € and indirect ones are in €/year. SB1 SB2 SB3 SB4 SB5 SB6 SB0 Equipment Purchase Cost (€)1 321,728 321,726 340,225 321,772 321,707 149,940 458,549 Reactors, 80,137 80,137 87,227 80,137 80,137 33,114 309,910 of which AGS 22,995 22,995 23,568 22,995 22,995 22,995 Pumping 72,403 72,403 73,301 72,403 72,403 33,448 21,593 Stirrers 4,185 4,185 4,185 4,185 4,185 4,185 10,736 Blowers 7,172 7,172 7,172 7,172 7,172 3,825 40,852 Compressor 75,367 75,367 75,367 75,367 75,367 75,367 75,367 Heat Exchanger 82,462 82,462 92,967 82,462 82,462 - - Main Indirect Costs (€/year) 96,261 107,374 121,616 107,373 107,366 122,048 259,009 Electricity 23,483 23,483 25,042 23,483 23,475 18,113 101,423 Heat 55,566 66,679 78,773 66,679 66,679 - - Sludge management2 17,212 17,212 17,800 17,212 17,212 103,935 157,586 1 The compressor is used in the DAF, and the heat exchangers are used to ensure that the temperature of the biological valorisation system maintains a temperature of 22.5 ºC to ensure that oil’s emulsion. 2 For SB1-SB5, the cost of sludge management includes the price of managing waste solids, but not oil (oil is valorised in the SBR and FBR system). For SB6 and SB0, it also includes the price of oil treatment by biodiesel production. Figure 6.9. Costs of the scenarios of system SB (FPW = Fish Processing Wastewater, AD = Annual Depreciation). SB0 represents the current linear economy-based system and SB1 – SB5 are the circular economy-based ones, while SB6 is the system where oil is not biologically valorised (only the current CAS reactor is replaced by an AGS reactor). CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 81 Treating FPW under a linear economy approach costs 0.52 €/kg COD (SB0), while treating it using circular economy approaches yield in 0.30 – 0.33 €/kg COD for SB1 – SB5 (Figure 6.9). Finally, treating wastewater under an innovative approach using AGS to treat the wastewater (but not biologically valorising it into PHA/TAG) yields in 0.21 €/kg COD (SB6). Here, other criteria, such as environmental or societal, could be considered to assist the decision making for oil valorisation into PHA/TAG. In this sense, elements like the heat and electricity requirements of the biological valorisation of the oil might hinder not only the economic feasibility (see Table 6.5) but also presumably the environmental one. The results presented then prove that replacing CAS units by AGS ones can clearly improve the economic performance of these industrial WWTPs, while they also indicate that further research and improvement might be needed before the biological oil valorisation into PHA/TAG is implemented at full-scale. 6.3.3. Environmental Analysis 6.3.3.1. Environmental impacts of System A scenarios with MCW The environmental validation of the systems based on circular economy strongly depends on the process configuration (Figure 6.10). In fact, the option with highest impacts is SA4, due to the use of chemicals for pH control in the acidification CSTR. Figure 6.10. Impacts of the assessed processes for SA at small scale (DSP and PHA are not accounted). GWP = Global Warming Potential, ODP = Ozone Depletion Potential, AP = Acidification Potential, F-EP and M-EP = Freshwater and marine eutrophication potential, respectively, ME = Marine Ecotoxicity, HT = Human Carcinogenic Toxicity, FRS = Fossil Resource Scarcity. The most profitable scenario economically (SA3) has higher impacts than the current strategy (process SA0) for some categories (about 5% for global warming (GWP), 15% for acidification (AP), 10% for freshwater eutrophication (F-EP), and 5% for human toxicity (HT)), also due to the chemical’s consumption. Nevertheless, the system was -40. -20. 0. 20. 40. 60. 80. 100. GWP ODP AP F-EP M-EP ME HT FRS Relative impacts (%) SA0 SA1 SA2 SA3 SA4 ALBA ROIBÁS ROZAS 82 upscaled using laboratory-scale dosages of chemicals where different ranges for chemical dosage were tested, but a proper dosage optimization was not performed. In fact, it was observed that the key feature enhancing process performance was linked to the enrichment strategy, and not to the pH or the alkalinity in the CSTR (Chapter 3), so it would be expected that, at full scale, the use of chemicals for pH control would be optimized and reduced. Moreover, sludge management by composting (the technique currently used to treat waste sludge in these factory types) generate noticeable impacts in climate change, ozone depletion and acidification, due to the process emissions (mainly ammonia, sulphur dioxide and nitrogen oxides, as well as some fossil carbon dioxide and methane). In fact, other waste management strategies (like the centralized anaerobic digestion of SB) could potentially validate the environmental profile of the process when compared to conventional techniques, as composting generates some impacts, and anaerobic digestion usually generates environmental credits. Finally, scenarios SA1 and SA2 perform better environmentally at this 1st stage of the analysis, but they had low PHA accumulations (Table 6.2), so they are not the best option from a techno-economic point of view. 6.3.3.2. Environmental impacts of System B scenarios with FPW For SB, the processes based on circular economy have less impact than the ones based in linear economy for all categories except for marine eutrophication (Figure 6.11) due to small differences in the effluent quality. Only SB1 performs slightly worse in some categories because of the higher use of chemicals for the enrichment of the MMC, as dosages were not optimized at this stage. Figure 6.11. Impacts of the assessed processes for SB at small scale (DSP and PHA are not accounted). GWP = Global Warming Potential, ODP = Ozone Depletion Potential, AP = Acidification Potential, F-EP and M-EP = Freshwater and marine eutrophication potential, respectively, ME = Marine Ecotoxicity, HT = Human Carcinogenic Toxicity, FRS = Fossil Resource Scarcity. -100. -80. -60. -40. -20. 0. 20. 40. 60. 80. 100. GWP ODP AP F-EP M-EP ME HT FRS Relative Impacts (%) SB0 SB1 SB2 SB3 SB4 SB5 SB6 CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 83 In fact, although the processes based on circular economy have a lower impact than the conventional one, the use of chemicals have important effects. As the enrichment process in the valorisation system had uncoupled nitrogen and carbon feedings and pH control, ammonium chloride and sodium bicarbonate were used, respectively. This yields in noticeable effects, representing about 50% of the impacts for global warming, acidification, and fossil resource scarcity. Moreover, the operation of the valorisation reactors for SB need to happen at mild temperature, and the steam needed yielded in environmental impacts (10 – 20% of the effects for global warming, acidification and marine ecotoxicity). Finally, the use of anaerobic digestion for sludge management results in environmental credits (specially for ozone depletion, marine ecotoxicity and human toxicity, but also for global warming), although it generates relevant impacts for freshwater eutrophication. Finally, managing oil externally (SB6) by esterification for biodiesel production results in the best environmental profile of all the studied scenarios. Therefore, replacing the current CAS by a system based on biofilms and AGS would yield in lower use of electricity, heat, and chemicals, so the process would be more profitable both economically and environmentally. Hence, it seems that biologically valorise oil into PHA/TAG is not yet the most attractive option from the environmental point of view. 6.4 RESULTS AND DISCUSSION – 2ND STAGE The results of the first stage showed that, for most process configurations, the economic performance of circular-economy and/or innovative approaches improves or has the potential to improve the current conventional approach. The analysis also showed that process configuration affects both the environmental and the economic performance, so the best alternative for each system was selected for the second stage of the study. For the case of MCW (SA), process configuration SA3 was chosen to upscale the facilities, as: i) SA3 has the lower costs at small scale, ii) lower environmental impacts are expected if waste sludge is treated by anaerobic digestion instead composting, and chemicals addition is optimized for full-scale operation, and iii) it is the scenario with the highest net PHA accumulation. For the case of FPW (SB), two alternatives based on two DSP approaches are considered, as two products can be generated (PHA and TAG). For PHA production only, the considered process configuration is SB3, as it has the highest production of PHA. For PHA+TAG extraction, SB2 was considered, as it is the only process approach that presented non-negligible concentrations of both products simultaneously. Besides, the environmental performance of SB6 is studied as well, as it presented the best economic performance in the first stage of the study. Note that it is not necessary to perform an economic evaluation of SB6 in the second stage of the study, as no biomaterials are produced in this scenario. Therefore, the cost of wastewater treatment was already assessed in the first stage of the evaluation, and no extra costs need to be calculated (thus, no extra benefits are included). Three scales will be tested for each of the selected scenarios. First, it was considered a scale ten times higher than the current one (x10), representing ten small facilities that provide PHA/TAG-rich biomass to be extracted in a centralised facility (plus SB6 case, where the performance of ten WWTPs treating AGS is compared to the same plants using CAS). Accordingly, scenarios for twenty (x20) and fifty (x50) WWTPs are assessed, where DSP would be centralized. Besides, for all the cases, it was assumed that waste ALBA ROIBÁS ROZAS 84 solids are managed by centralized anaerobic digestion. Moreover, it is important to notice that, in this 2nd stage of the evaluation where DSP is included, the environmental and economic burdens of PHA/TAG-rich biomass production (thus, wastewater treatment) are allocated to the WWTPs, while the burdens and benefits of biomaterials production (PHA/TAG extraction, and PHA/TAG replacement of fossil products) are allocated to the centralized facility where DSP is carried out. Finally, the use and end of life phases of the bioproducts life cycles are excluded of the study due to the lack of information. Note that transport costs are not considered in this study due to uncertainty, and because previous studies showed that the impact will not be very significant for short distances (Chapter 5), so it is assumed that raw materials (PHA-rich sludge) are supplied to the DSP plant directly. 6.4.1 Centralized biorefinery for PHA production: economic results As previously presented (Figure 6.8), the cost of wastewater management was 0.48 €/kg COD for SA3 (MCW). For the scenario x10 times the current one, this centralized biorefinery performs the DSP of the received biomass from ten facilities. The production cost of PHA under this approach would be 1.18 €/kg PHA (Figure 6.12), where costs come only from extraction operations (the cost of biomass generation is allocated to the WWTPs). Figure 6.12. Cost of PHA and TAG production under the centralized approach where WWTPs are raw materials providers for a centralized biorefinery where PHA is extracted from the biomass. Here, the cost of PHA is only due to the DSP. The costs of producing PHA with centralized extraction approaches for higher scales decrease accordingly. For a system with 20 facilities, costs are 1.05 €/kg PHA. For the scale of 50 facilities with centralized DSP, costs are 0.95 €/kg PHA (Figure 6.9). Considering that the cost of HDPE varied in the last four years in a range of 0.70 – 1.04 €/kg (statista, 2021), for the x20 and x50 scales, the cost of PHA produced from waste streams would yield in competitive prices, with promising results that could improve with process optimization and due to factors like the petrol availability. For FPW (SB3), the cost of treating wastewater for each facility is 0.33 €/kg COD (Figure 6.9). Then, the cost of the centralized extraction is 2.16 €/kg PHA, and it 0.00 1.50 3.00 4.50 6.00 7.50 9.00 10.50 12.00 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 Scale x10 Scale x20 Scale x50 TAG Cost (€/kg) PHA Cost (€/kg) PHA from MCW (SA3) PHA from FPW (SB3) PHA from FPW (SB2) TAG from FPW CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 85 decreases as scale increases, yielding in 1.84 €/kg PHA and 1.56 €/kg PHA for the scales x20 and x50, respectively (Figure 6.12). The reason why DSP is more expensive for SB than for SA is linked to the lower biomass concentration for the system that uses oil. Here, prices are close to the competitive cost of HDPE for the highest scale, although they are still about 30% higher. 6.4.2. Centralized biorefinery for PHA and TAG production: economic results To extract TAG, it is necessary to include distillation columns using hexane and methanol in the extraction process (Figure 6.7). Besides, the DSP is carried out at relatively high temperature (80 ºC, while DSP for low grade PHA only is at room temperature). Therefore, for scenarios where both TAG and PHA are produced, the cost of producing PHA increases when compared to scenarios where only PHA is extracted. The costs of a system with centralized extraction also decrease linearly with scale, but they are considerably higher than the market price of their petrochemical competition. PHA produced by centralized DSP when also TAG is extracted costs 3.35, 2.88, and 2.45 €/kg PHA for scales x10, x20, and x50, respectively (Figure 6.12). Therefore, PHA production cost for this combined DSP approach increases with respect to one obtained for PHA extraction only. Finally, TAG cost is 9.21, 7.90, and 6.75 €/kg TAG for the scales x10, x20, and x50, respectively. As there is still not a global market for TAG, it is difficult to stablish a threshold value to compare the results with other studies. Nevertheless, these numbers are considerably higher than the 1 €/kg that is traditionally assigned to petrochemical goods of common use. Moreover, the marketability of TAG is still uncertain, so more research is needed in this area. 6.4.3. Environmental performance of the centralized DSP approaches The environmental performance of ten, twenty, or fifty facilities that provide sludge for a centralized biorefinery where PHA and TAG are extracted is compared with the performance of the same facilities when they only treat wastewater (i.e. the current linear economy approach). Moreover, the performance of SB6 is included, where the approach is not circular, but innovative (as AGS is used to treat wastewater, instead of CAS). As happened with economic results, the scale had a linear effect on the results and that is why only the results of the scenario for 10 facilities are shown (Figure 6.13). Thus, only absolute values of emissions change, as they increase accordingly, but the relative impacts are the same between scales. Here, the avoided products (HDPE for SA3, and diesel with full petrochemical origin for SB2 and SB3) are accounted for, so the circular economy-based processes generally perform better than the conventional ones (Figure 6.13), generating environmental benefits. For MCW management, the circular economy-based process (SA3) performs better than the linear economy-based one (SA0) for all categories freshwater eutrophication potential, due to the use of chemicals, where a wide margin for improvement still exists. Here, the credits generated by the avoidance of petrochemical materials are smaller, so the impacts linked to the use of chemicals still provoke that the improvement for circular economy-based processes is not fully validated for this category. Positive effects are especially noticeable for global warming and fossil depletion categories, where the ALBA ROIBÁS ROZAS 86 environmental credits generated by the avoidance of a petrochemical plastic production are clearer. For FPW management, the conventional process (SB0) has impacts higher than SB3 for all categories except for marine eutrophication potential, due to small variations in the effluent quality between scenarios, while SB2 performs worse than SB0 for some categories due to the lower bioproduct generation. Finally, small impacts are linked to scenario SB6, where biological valorisation of oil is not performed. As the amount of bioproduct generated for systems using FPW is low, the potential benefits of avoiding petrochemical materials are not enough to buffer the burdens of operating the train of biological reactors, as it was for SA3. Figure 6.13. Impacts for a bigger scale considering 10 small facilities. The ones based on circular economy work as PHA/TAG-rich biomass providers for centralized DSP. GWP = Global Warming Potential, ODP = Ozone Depletion Potential, AP = Acidification Potential, F-EP and M-EP = Freshwater and marine eutrophication potential, respectively, ME = Marine Ecotoxicity, HT = Human Carcinogenic Toxicity, FRS = Fossil Resource Scarcity. On the other hand, biomass production is lower for oil-based systems than for MCW ones. Therefore, the final process biomass yield is lower than that for the MCW-based process: 0.019 kg PHA/kg COD for SB2 and 0.024 kg PHA/kg COD for SB3, versus 0.16 kg PHA/kg COD for SA3. This makes SB2 performance worse than SB3, and, for some categories, also worse than SB0. Therefore, the margin of improvement (the impact reduction) of circular economy-based systems with oily feedstock is narrower than the one for SA, where impact reduction is clearer. 6.5 DISCUSSION AND COMPARISON WITH THE AVAILABLE LITERATURE The use of biomass and wastes to generate biofuels and biomaterials could be a feasible solution to solve current threats faced by humans, as the climate crisis. However, different problems need to be addressed before these biotechnological processes are applied (Ahmed et al., 2021). Although there is a claim for biomaterials like PHA to be the environmentally friendly solution for the world’s future economy (Shahid et al., 2021), the true is that the lack of environmental and economic validation is hindering the introduction of these materials in the value chains. -100. -80. -60. -40. -20. 0. 20. 40. 60. 80. 100. GWP ODP AP F-EP M-EP ME HT FRS Relative Impacts (%) SA0 SA3 SB0 SB3 (PHA only) SB2 (PHA + TAG) SB6 CHAPTER 6: ASSESSING A CENTRALIZED BIOREFINERY APPROACH UNDER A TECHNOECONOMIC AND ENVIRONMENTAL PERSPECTIVE 87 The present study showed that applying strategies based on circular economy to manage industrial effluents could decrease the cost of wastewater treatment. Moreover, the economic feasibility of these processes is not necessarily relying on the production of biomaterials, but in the implementation of a more efficient treatment scheme combined with centralized and low-cost DSP approaches. It was proved here the economic and environmental feasibility of transforming industrial WWTPs into raw materials suppliers for biorefineries. Moreover, realistic models to achieve technical, environmental, and economic viability were explored. For example, in Galicia (NW Spain), around 70 companies generate about 85% of the total Spanish canned fish products and 50% of the total European canned tuna production (González-López, 2018). Here, a system where these existing small and medium capacity WWTPs work as biomass providers for a centralized downstream facility could improve not only economic and environmental benefits, but also the viability of a sector where conservative business models have in some cases hindered the evolution of this industry (González-López, 2018). The average price of PHA production for processes based on pure cultures reported in the las fifteen years is 3.46 €/kg PHA (Bhattacharyya et al., 2014; Leong et al., 2017; Martinez-Hernandez et al., 2018; Musonda et al., 2020; Pavan et al., 2019; Shahzad et al., 2017; Vega et al., 2020). Here, the use of mixed microbial cultures and waste streams to produce biomaterials claims to reduce those costs. When some of the first economic evaluations were made for MMC-based PHA, costs were estimated in 3.52 $/kg PHA (2.71 €/kg PHA) for a production of 700 t PHA/year (Gurieff and Lant, 2007). Later, Fernández-Dacosta et al. (2015) achieved 1.40 – 1.95 €/kg PHA for a production of 1500 t PHA/year depending on the DSP method, while Bengtsson et al. (2017) used for the first time pilot scale data to estimate 3.40 €/kg PHA for a production of 5000 t PHA/year. Crutchik et al. (2020) also assessed the effect of scale using data from real WWTPs, where production costs were 2.26 €/kg PHA for 90 t PHA/year and 1.26 €/kg PHA for 5700 t PHA/year. Pérez et al. (2020a) discussed on how PHA economic feasibility is variable with the geographical location, and costs ranged 1.5 – 6.9 €/kg PHA when part of the methane was used to generate energy on-site for systems using methanotrophic MMC. Therefore, the cost for MMC-based processes averaged 2.58 €/kg PHA in the last fifteen years, proving that it would indeed decrease PHA price with respect to pure culture biomaterials. Finally, Saavedra del Oso et al. (2020) investigated the cost of different DSP processes, and their cost ranged 0.2 – 2.25 €/kg PHA, where the lowest prices corresponded to low grade PHA (i.e. for low grade applications like packaging). Under the approach applied in this study, where WWTPs are raw biomass suppliers and the low-cost DSP is centralized, WWTP holders would not directly benefit from PHA production (they do not produce PHA on-site). Indirect benefits would be the lower cost of sequential reactors, as they would treat the same flow of wastewater using less reactor volumes (thus, less aeration is needed), and the savings in sludge management. Therefore, PHA production cost would only correspond to the extraction, being 1.18, 1.05, and 0.95 €/kg PHA for productions of 580, 1200, and 3000 t PHA/year, respectively, for wastewater feedstock (MCW). For waste oil feedstock (FPW), PHA costs are 2.16, 1.84, and 1.56 €/kg PHA for productions of 230, 450, and 1100 t PHA/year (if TAG is not recovered). Regarding the environmental effects, among the about 60 research works assessing the impacts of PHA production recently revised (Roibás-Rozas et al., 2022b) only 11 ALBA ROIBÁS ROZAS 88 considered MMC-based processes, and global warming potential was the most studied category. Although results vary depending on different methodological or process choices, global warming-related emissions generally ranged 0.5 – 5 kg CO2-eq/kg PHA (Vogli et al., 2020), which clearly show a great variability of the resultant impacts. For the present study, the emissions linked to the wastewater treatment generally decreased due to the less use of energy when compared to the conventional treatments currently applied, so the processes would be validated from a wastewater management perspective. When PHA extraction is centralized, the environmental burdens of PHA-rich biomass production are allocated to the WWTP, so PHA production’s burdens only account for the DSP. Here, the emissions of the extraction process are 0.25 kg CO2-eq/kg PHA for the system using MCW, 0.29 kg CO2-eq/kg PHA for the system using FPW with no TAG extraction, and 0.31 kg CO2-eq/kg PHA and 1.1 kg CO2-eq/kg TAG for the oil system that also recovers TAG, which is in the low range considering other literature studies. It is noticeable that the feasibility of the proposed system is also relying on new DSP approaches where the use of energy and chemicals is lower than in the extraction processes traditionally employed in PHA extraction. Generally, DSP was one of the process bottlenecks. However, with the use of these low-demanding processes, this might not be the case anymore if the aim is to produce low grade PHA (Saavedra del Oso et al., 2020). 6.6 FUTURE OUTLOOK The need of implementing more sustainable, yet economically profitable, production systems and growth models, is urgent. However, there are just a few PHA pilot-scale projects worldwide (Estévez-Alonso et al., 2021; Mannina et al., 2020), so studies on the economic and environmental validation of resource recovery processes are missing. As proved in this study, and stablished by Saavedra del Oso et al. (2020), DSP might not be a bottleneck anymore if low grade PHA is enough to meet the standards required. Therefore, special attention needs to be paid to the use of chemicals in the PHA-rich biomass production process, where pilot experiences are needed to stablish real thresholds. Regarding PHA production with wastewater, for the system using MCW, the presence of proteins was a major problem at laboratory scale (Chapter 3). Hence, the HRT of the acidification unit was high (6.25 days) to carry out their fermentation and remove them from the wastewater. Nevertheless, if other strategies to remove proteins were tested (like coagulation), the wastewater rich in carbohydrates can be fermented into VFA in approximately two days (Yang et al., 2015). This would decrease the CSTR volume (direct costs) and operational expenses (electricity of the stirrers) yielding in a higher efficiency and lower costs. Moreover, the obtained proteins could be used for other valueadded applications, like animal feeding. However, pilot and lab-scale data would be needed to confirm this. Besides, the potential benefits linked to the end of life of biomaterials or the detrimental (and hard to quantify) effects of plastic pollution of petrochemical materials (like plastic in the oceans) have been disregarded. As stated in chapter 1, ongoing projects like MariLCA (Marine impacts in LCA) or Plastic Leak, where these effects are trying to be quantified, could bring some light into the issue. CHAPTER 7: MODELLING THE EFFECTS OF SALINITY VARIATIONS IN AQUATIC ENVIRONMENTS 95 7.2.3. Effect Factor (EF) The EF quantifies the fraction of living species that are going to potentially disappear in the aquatic ecosystem by the release of a certain chemical (Fantke et al., 2018; Owsianiak et al., 2019; Rosenbaum et al., 2018). The available literature regarding salinity variation effects in aquatic ecosystems models these impacts according to ecotoxic methodologies, such as USEtox (Amores et al., 2013; Zhou et al., 2013). Under that approach, the EF is calculated employing a species-sensitivity distribution (SSD) curve which represents the sensitivity of an entire ecosystem to a substance (Rosenbaum et al., 2018). However, it is constructed based on the premise that increased exposures will lead to increased effects (Owsianiak et al., 2019). Indeed, if the salinity of a fresh aquatic environment increases, the organisms living in the system will be negatively affected (Böhme, 2011; Kefford et al., 2011; Piscart et al., 2011; Verbrugge et al., 2012; Ziemann and Schulz, 2011). However, negative effects in aquatic ecosystems are also reported when salinity decreases, such as massive mortalities reported after exposure to low salt concentrations due to climatologic events or anthropogenic actions (Du et al., 2021; Johnson et al., 2021; Parada et al., 2012). Moreover, some limitations of SSD curves to model ecotoxic impacts due to salinity have been pointed out (Fox et al., 2021), and other indicators, such as species richness, have been proposed for studies assessing the effects of salinity (Kefford et al., 2011). Therefore, variations in the concentration of essential substances can provoke effects in several directions, as an increase in the salt concentration can be potentially beneficial for an ecosystem and vice versa, meaning that emission-related impacts can be positive in some cases. Moreover, currently, it is considered that a substance provokes toxic effects if it enters in an organism and causes poisoning, endocrine disruption, or other lethal effects (Owsianiak et al., 2019). This approach might be partially accurate for effects of exposure to high salt concentrations, but it fails when defining the observed effects on ecosystems for decreases in salinity. This is the reason why the current applied approach (using ecotoxic methodologies) to model the impacts due to variations in essential substances (such as salts) need to be reconsidered. 7.3. EFFECT FACTOR FOR SALINITY VARIATIONS IN AQUATIC ENVIRONMENTS The proposed approach is based on the premise that the species in an ecosystem have an optimal range of salinity for living, and that detrimental effects will be observed if it varies below or above it (Smyth and Elliot, 2016). Figure 7.2 shows an ideal representation of a transitional water body, where there is an optimal concentration of salt at which no impact occurs (i.e., PDF = 0). Please note that this figure represents a Then, negative effects will occur if salinity increases above the optimal range or decreases below it. Moreover, positive impacts will occur for increments in the salt concentration for environmental concentrations below the optimal, and vice versa. Note that, for salinity increases in the high range, the function follows a distribution like the classic SSD curve. The first step is then to define the optimal salt concentration range by gathering data of chronic effects for the species of the ecosystem regarding salinity (also needed to quantify the EF itself). If the data are expressed as acute, an acute-to-chronic ratio (ACR), which is generally 2 (Fantke et al., 2018), can be used by dividing or multiplying the salt concentration by the ACR in the high range or in the low range, respectively. Then, according to our hypothesis (effects will happen above and below a favourable range), ALBA ROIBÁS ROZAS 96 that data representing the chronic effects linked to salinity will be shaped like in the examples provided in Figures 7.2a, 7.2b, or 7.2c. For water systems from oligohaline to hyperhaline (see Appendix 5), it is expected that the concentration-to-response curves have a shape like the ones shown in Figure 7.2a or 7.2b, where the optimal (environmental) salt concentration can be estimated as a range (the points for which the function has values y < 0, Figure 7.2a) or a point (the minimum value of y, Figure 7.2b). Figure 7.2. Expected (hypothetical) shape of the curve defining the optimal salinity range for aquatic ecosystems. The main plot is an ideal representation of the impacts of salt variations, and the graphs a, b, and c are theoretical representations of different types of ecosystems, where a and b shape the impacts of transitional or non-fresh compartments, and c is a freshwater system. For freshwater environments, it is more likely that the impacts in the ecosystem are due to salt concentration increases. For the part of the curve representing salinization, a classic SSD approach might be a fair approximation. Nevertheless, the whole system profile is only provided if both ranges (high and low) are included, where the optimal (environmental) concentration can be found as the intersection of the two functions (a lineal one for the low range and SSD-like for the high range, Figure 7.2c). After defining the optimal region, there are two EFs: EFLOW and EFHIGH, both in PDF·m3/kg. They represent the amount of a substance that generates a certain effect on the ecosystem, where the EF is the slope of the concentration-response curve. HC50LOW and HC50HIGH (both in kg NaCl/m3) represent the salt concentration that generates an CHAPTER 7: MODELLING THE EFFECTS OF SALINITY VARIATIONS IN AQUATIC ENVIRONMENTS 97 effect in 50% of the ecosystem species in the low and high range, respectively; thus, for effects in 50% of the species, EFs are expressed as in Eq. 7.5 and Eq. 7.6. HC50s are the geometric mean of the individual species EC50 (Fantke et al., 2018), which is the concentration of pollutant that generates an effect in the 50% of the individuals of a single species. In this sense, an SSD-like approach is still maintained. EFLOW=0.5 HC50LOW Eq. 7.5 EFHIGH=0.5 HC50HIGH Eq. 7.6 Assuming a classic SSD-like curve (i.e., log-logistic functions), EC50s could be calculated using to mirrored sigmoidal curves. Moreover, according to our hypothesis (Figure 7.2), EC50s could be estimated by fitting the chronic effect data to a quadratic function (see Appendix 5). Note that the log-logistic and the quadratic curves only converge at intermediate ranges of effects, so this approximation might not be accurate to calculate EFs based on HC10 or HC20, as new trends on ecotoxicity modelling suggest the use of EC10s or EC20s to model the factors (Owsianiak et al., 2019). Finally, the sign of each EF will be established depending on the direction of the change: increasing the salt concentration in the low range accounts for a negative EFLOW (impacts decline), and decreasing it accounts for a positive EFLOW (impacts increase), and vice versa for the high range. 7.4. APPLICATION OF THE METHODOLOGY TO A CASE STUDY 7.4.1 Fate Factor Calculation For the case of Arousa ría, Figure 7.1 is simplified: WG is neglected due to the insignificant contribution of the groundwater to the total flow of the ría (Ministerio de Agricultura Pesca y Alimentación, 2021), and WI-OUT is ignored since no infiltrations from the estuary to the river are expected (see Appendix 5). The remaining streams are quantified in cubic meters per month, covering from 2011 to 2018 according to the data availability, as detailed bellow. First, stream flows are determined: WI-IN corresponds to the flows of rivers Ulla and Umia, which are monthly measured and reported by local administrations (Augas de Galicia, 2021) in several locations along the river courses, so the closest point to the ría was chosen. WWW can be divided into fresh wastewater (WWW-FRESH) and salty one (WWW-SALT). It was estimated that the area of Arousa ría is responsible for about 6.6% of the total (WWW) Galician discharges (Ministerio de Agricultura Pesca y Alimentación, 2021; Ministerio para la transición ecológica y el reto demográfico, 2018) and that 82–92% of this amount (yearly variation in 2011-2018) corresponded to fish canning wastewater (Ministerio de Agricultura Pesca y Alimentación, 2021; Ministerio para la transición ecológica y el reto demográfico, 2018), the latter representing virtually all saline emitting sectors (WWW-SALT, in cubic meters per year). Then, to obtain monthly data, a distribution was defined based on the shellfish and seafood harvesting and processing seasonal pattern to estimate WWW-SALT in m3/month (see Appendix 5). On the other hand, WWW-FRESH (including urban wastewater and fresh industrial streams, in cubic meters per year) was estimated as the subtraction of WWW- ALBA ROIBÁS ROZAS 98 SALT from the total WWW. Then, to obtain monthly data, it was assumed that fresh discharges are evenly distributed along the months of the year (also considering that freshwater streams represent a flow considerably lower than salty streams). WANTH is not linked here to irrigation, as estuary water has high salinity and Galicia is an area with high precipitations, but the local fish canning industries use seawater for their processes with variable flows. Due to the lack of more precise data, we assumed that 75% of water input for fish canning processes was saline water and 25% was freshwater (Barros et al., 2009; Bello Bugallo et al., 2012; Dirección Xeral de Calidade e Avaliación Ambiental, 2008c; Tomczak-Wandzel et al., 2015) and that there are no water losses in the process, i.e., inputs = outputs, thus WANTH = 0.75·WWW-SALT. Evaporation rates were estimated using empirical correlations as indicated in Eq. 7.7 (Mcjannet et al., 2012; Risch et al., 2014), where W*EPT is stated in L/m2·time (i.e., monthly averages in this case), u2 is the wind speed (m/s), A is the evaporating area (m2), v*w is the saturated vapor pressure at the water surface temperature (kPa), and va is the partial vapor pressure in the air at 2 m height (kPa) calculated by Eq. 7.8, where v*a is the saturated vapor pressure at air temperature (kPa) and ɸ is the relative humidity (%). Finally, to obtain WEPT in equivalent units to the rest of the streams (m3/month), Eq. 7.9 is needed. W∗EPT=(2.36+1.72·u2)·A−0.05·(vw ∗−va) Eq. 7.7 va=va∗·ϕ Eq. 7.8 WEPT=WEPT ∗·A 1000 Eq. 7.9 Average monthly precipitations (WP) were obtained from the meteorological station located in Corón (Vilanova de Arousa) (MeteoGalicia, 2021). WR (runoff) was taken as the average values of the South West basins of Galicia (Fekete et al., 2002). Being the only data set available, the monthly distribution and expected runoff flow for 2002 were applied to the 2011-2018 period defined. At this point, all the flows, except WO-IN and WO-OUT, which represent the tidal, are quantified, so mass balances must be solved. First, Eq. 7.10 and Eq. 7.11 present the general steady state water and the salt balances, respectively, where Si is the salt concentration (kg NaCl/m3) of each stream Wi. WO−IN+WI+WR+WP+WWW=WO−OUT+WEPT+WANTH Eq. 7.10 WO−IN·SO−IN+WI·SI+WR·SR+WP·SP+WWW·SWW =WO−OUT·SO−OUT+WANTH·SANTH Eq. 7.11 Note that Eq. 6.10 could be written as WIN = WOUT, representing the total inflows and outflows in the estuary. Then, to accurately quantify the tidal streams, new mass balances will be stated, representing a system as natural as possible (Eq. 7.12 and Eq. 7.13), where no anthropogenic activity exists. WO−IN+WI+WR+WP=WO−OUT+WEPT Eq. 7.12 WO−IN·SO−IN+WI·SI+WR·SR+WP·SP=WO−OUT·SO−OUT Eq. 7.13 CHAPTER 7: MODELLING THE EFFECTS OF SALINITY VARIATIONS IN AQUATIC ENVIRONMENTS 99 To be able to proceed, the salt concentration of each stream, Si, needs now to be quantified: The salt concentration of the ocean (SO-IN) was considered constant at 36 g NaCl/L, as it is the average estimated sea surface salinity measured between 2004 and 2013 for this Atlantic area by satellite monitoring (Durack, 2015). The salt concentration of the estuary (SO-OUT, SANTH, but also 𝑆 in Eq. 7.3) is variable according to the spatial distribution and the tidal intensity. Water is fresher near the river mouth and saltier near the ocean, but the salt concentration also increases with depth because salty water has a higher density (Venâncio et al., 2019). Data collected from two buoys (Ribeira, in an intermediate location between the river mouth and the ocean, and Cortegada, near the river mouth; see Appendix 5) (MeteoGalicia, 2021) were used to determine the monthly average salinity of the estuary. Galician rivers are not salinized, presenting low salinity and conductivity (Estévez et al., 2019). Normally, chloride concentrations of Spanish rivers range between 0.010 and 0.030 kg Cl-/m3, and recent studies for Galician rivers near shore areas reported values of 0.0103 – 0.0222 kg Cl-/m3 (Dias et al., 2019). An average value of 0.020 kg NaCl/m3 is considered for SI and SR. Chloride concentration in rainwater is generally very low, although it varies worldwide depending on wind intensity and seawater proximity. Measurements of the chloride concentration in rainwater in Spain and Portugal averaged 0.020 kg Cl-/m3, and the value reported for the Galician station, located at around 40 km from the coast, was 0.003 kg Cl-/m3 (Alcalá and Custodio, 2008). A slightly greater salinity of 0.005 kg NaCl/m3 is applied for SP to consider the effect of sea salt aerosols. Eq. 7.14 and Eq. 7.15 are used to calculate SWW, where WWW-SALT is the yearly flow of industrial fish canning wastewater discharged into the estuary (m3/year), and 𝑆𝑡 is the average salt concentration of the estuary each year (kg/m3). It was assumed that SWWFRESH coming from drinkable sources was 0.5 kg NaCl/m3 (WHO, 2006), and that SWWSALT contained 75% of estuarine water and 25% of fresh (drinkable) water as previously indicated. The salinity of waste streams poured into the ría SWW can be estimated as an average value as shown in Eq. 7.15. SWW−SALT=0.75·WWW−SALT·St+0.25·WWW−FRESH·0.5 WWW−SALT Eq. 7.14 SWW=WWW−SALT·SWW−SALT+WWW−FRESH·SWW−FRESH WWW−SALT+WWW−FRESH Eq. 7.15 Finally, the FF can be calculated. As already stated, both monthly and yearly results can be obtained, as well as seasonal values, the dry season spanning June, July and August in the case study area, and the remaining months are the wet season. To see detailed monthly, seasonal, and yearly results, see Appendix 5. The results for the FF are shown in Table 7.1. Table 7.1. Results of the FF for Arousa ría (in units of month and year). The results are expressed as average ± standard deviation, and the confidence intervals are between brackets, where negative results were assumed zero. Wet (month) Dry (month) Annual (year) 6.84 ± 1.84 4.02 ± 1.52 4.51 ± 1.51 [1.32, 12.36] [0, 8.59] [0, 9.05] ALBA ROIBÁS ROZAS 100 7.4.2. Effect Factor calculation To calculate the EFs, data of chronic effects were gathered (for this study, only mortality was considered). Then, the collected data were represented together to find the ecosystem optimal (environmental) salt concentration (Figure 7.3). Note that this value is not going to be employed in any other further calculation, and it is just used to determine which data will be utilized to generate EFLOW and EFHIGH. In this sense, several fittings could be tested, as shown in Figure 7.2a and 7.2b. Different approaches were applied, evaluated, and discussed (see Appendix 3) to identify this cutoff point. These strategies pointed out a possible environmental optimal range of 24–36 g NaCl/L, where the average point obtained after testing different fittings is 31.9 ± 1.4, which will be the cut-off to define the low and high ranges of salt concentration comprising EFLOW and EFHIGH, respectively. Figure 7.3. Distribution of the data regarding mortality of different species present in Arousa ría. The data used to generate this plot are specified in Table 7.1 and in Appendix 5. Therefore, EC50 concentrations are calculated for each species at high and low range considering the estimated cut-off point. Most of the information gathered referred to effects at low range, so the EC50LOW for these species can be obtained directly using a sigmoidal curve with a profile mirrored to the classic SSD curve fittings. However, EC50LOW and EC50HIGH can be also estimated for each species by using the quadratic approach of our hypothesis, which provides effects at both ranges (see Appendix 5 for detailed information of how the quadratic approximation was used). The value of each EC50 (high and low) and both HC50s are shown in Table 7.2. For Table 7.2, the numbers in italic indicate that the concentration was obtained by the quadratic fitting using data from the low range of salinity (for V. senegalensis and S. polyschides, the power function had a convex profile, so the data were estimated using the inverse function, see Appendix 5). The results of EFHIGH and EFLOW (Table 7.3) fit with the expected system behaviour, as biotic diversity in estuaries starts to decline above a salinity of about 40 kg NaCl/m3, with most species unable to survive in salinities above 50 kg NaCl/m3 (Smyth and Elliot, 2016). 0 20 40 60 80 100 0 10 20 30 40 50 60 70 Mortality (%) Salt Concentration (g NaCl/L) Mollusk Annelid Algae CHAPTER 7: MODELLING THE EFFECTS OF SALINITY VARIATIONS IN AQUATIC ENVIRONMENTS 101 Table 7.2. Value of each EC50s and HC50s, and chronic data used to generate them. EC50s in italic represent the values estimated assuming quadratic distribution (see Appendix 5). The results are expressed as average ± standard deviation. HC50s and EC50s (kg/m3) Species EC50LOW EC50HIGH Ruditapes philipinarium (Japanese clam) (Carregosa et al., 2014; Parada et al., 2012) 16.0 34.4 Ruditapes decussatus (Grooved carpet shell) (Gharbi et al., 2016; Parada et al., 2012) 15.5 42.9 Cerastoderma edule (Common cockle) (Parada et al., 2012; Verdelhos et al., 2015) 20.4 61.7 Vanerupis corrugata (Pullet carpet shell)(Carregosa et al., 2014; Parada et al., 2012) 21.7 37.8 Saccharina latissima (Sea belt, brown algae)(Fiett, n.d.; Peteiro and Sánchez, 2012) 8.6 39.1 Diopatra neapolitana (Polychaete) (Freitas et al., 2015) 19.0 42.3 Donax trunculus (Wedge clam) (Reyes-martínez et al., 2020) 19.3 32.1 Mytilus galloprovincialis (Mediterranean Mussel) (Gaag et al., 2016) 9.8 38.2 Scrobicularia plana (Peppery furrow shell) (Verdelhos et al., 2015) 10.7 40.4 Saccorhiza polyschides (Furbellow, brown algae) (Norton and South, 1969) 30.1 40.1 Zostera noltei (Dwarf eelgrass, seagrass) (FernándezTorquemada and Sánchez-Lizaso, 2011; Salo et al., 2014) 1.3 49.6 HC50 (Geometric mean) 12.7 41.1 HC50 arithmetic mean 15.7 ± 7.8 41.7 ± 8.0 7.4.3. Characterization Factor calculation CFs were obtained according to Eq. (7.2). Here, the EF achieves different values according to the level of salinity (high and low), while the FF, which is supposed to be the same for both ranges of salinity (the water streams will have the same physical distribution regardless of the salt concentration), varies seasonally (wet/dry season). Therefore, there are six possible CFs to use (Table 7.4). Table 7.3. Value of each EF (PDF·m3/kg). The results are expressed as average ± standard deviation, and the confidence intervals are between brackets, where negative results were assumed zero. EFLOW EFHIGH 0.04 ± 0.02 0.01 ± 0.002 [0, 0.1] [0.005, 0.02] For the obtained FFs, the standard deviation acknowledges the discrepancy of steady state assumption (see Appendix 5). For the EF, the confidence intervals and the standard deviation of the factor were calculated by using the arithmetic average (see Table 7.4). For all factors, the uncertainty intervals were obtained considering three times the standard deviation, and negative lower bounds were considered zero. Table 7.4. Value of the CFs (PDF·month·m3/kg). The results are expressed as average ± standard deviation, and the confidence intervals are between brackets, where negative results were assumed zero. CFLOW (PDF·month·m3/kg) Dry Wet Annual 0.27 ± 0.21 0.16 ± 0.14 0.18 ± 0.15 [0, 0.89] [0, 0.57] [0, 0.62] CFHIGH (PDF·month·m3/kg) Dry Wet Annual 0.08 ± 0.04 0.05 ± 0.03 0.05 ± 0.03 [0, 0.20] [0, 0.13] [0, 0.14] ALBA ROIBÁS ROZAS 102 CFs are reported as absolute values, but they depend on the direction of the change (i.e., increase or decrease in salt concentration at low or high range). Therefore, to quantify the effects of a dam release during the wet season which provokes a decrease in the salt concentration, CF = 0.16 PDF·month·m3/kg (i.e. impact increases); but to quantify the impact of a saline effluent discharge in the same situation (wet season, low salinity), CF = -0.16 PDF·month·m3/kg should be used as salinity is already below the optimal, so the increase in salt concentration has a positive effect in the receiving waters. In this example, CFLOW is used instead of CFHIGH because Arousa ría has a salt concentration below the optimal (Estévez et al., 2019). 7.5 DISCUSSION When evaluating the effects of pollutants’ release, it is logic to assume that a rise in the chemical concentration generates an increase in the impacts. However, anthropogenic activities are affecting the planetary biogeochemical cycles, and impacts are now not only due to the release of harmful substances, but also to variations in the environmental conditions of ecosystems provoked by unnatural changes. There are some substances (e.g., salt, nitrogen, dissolved oxygen) to which the classic ecotoxic impact approach might just fit partially. Indeed, a drastic increase in the substance concentration will generate adverse impacts. However, at the same time, a certain concentration in the media is needed to support the ecosystem survival. In recent years, concern for salt variations in the different environmental compartments has increased, but the models available to describe their impacts are still scarce and the current literature uses ecotoxicity models to assess these effects (Amores et al., 2013; Berger et al., 2019; Núñez and Finkbeiner, 2020; Payen et al., 2016). For instance, the USEtox model, one of the most used methods to evaluate ecotoxic impacts in LCA, has not addressed effects of salt releases so far (Payen et al., 2016) and does not include the coastal seawater and brackish areas as environmental compartments (Owsianiak et al., 2019). Therefore, new approaches are necessary to include essential substances and these critical areas in the impact assessment models, and specific methodological choices must be implemented to develop CFs for essential elements in non-freshwater environments. Regarding the FF, it is expressed in this work in units of time (see the fate factor section in materials and methods). This provides a suitable framework to model the fate of essential substances. The few research works that have assessed the effects of salinity variations in aquatic environments solved the FF question from different approaches. In a first study, when evaluating the impacts of brine disposal (Zhou et al., 2013), the different elements present in brine were grouped. NaCl had a high residence time, so the FF was chosen as the residence time of the second most persistent element of the salinity group (Cu2+, 37 days). However, this approximation can have high errors due to the difference in the elements’ residence times (from days to millions of years). In another study, when steady-state mass balances were applied to the water streams and the salt in a coastal wetland (Amores et al., 2013), the units of the FF were g·year/L, yielding in a dimension of time per concentration. This factor was useful to directly link salt variations in the wetland not only with ecotoxic impacts, but also with social and economic effects as crop loss. However, the unit of that FF complicates the comparison of the results obtained there with the present study. Moreover, it hinders the evaluation of the impacts of salinity variations on the ecosystem linked to different impact categories as ecotoxicity. Therefore, to remain CHAPTER 7: MODELLING THE EFFECTS OF SALINITY VARIATIONS IN AQUATIC ENVIRONMENTS 103 coherent and consistent with other impact categories, it is necessary to provide factors that align with the standard units of the impact assessment stage and that are used by other methodologies recommended by the UNEP’s Life Cycle Initiative, such as the USEtox. This harmonization aspect also applies to the measure of the EF, which is PDF in this study because UNEP’s Life Cycle Initiative recommends to base ecosystem damage estimates on this metric (Owsianiak et al., 2019; Verones et al., 2017). As for the EF, the proposed approach copes with the fact that impacts might be due to increases but also decreases in the concentration of essential substances, and it acknowledges potential benefits linked to emission-related impacts. In this sense, a classic SSD-based methodology (typically used in ecotoxicity) appears as a practical and useful tool to predict the possible negative effects of a pollutant’s release on ecosystems. However, although some of the limitations linked to the use of SSD curves have been pointed out in the past few years, the principles shaping the methodology have remained unchanged for decades (Fox et al., 2021). Additionally, the use of classic ecotoxicity methodologies also have some limitations to model impacts due to variations in the concentration of essential substances, since the effects of these variations do not fit the classic definition of toxicity (linked to poisoning, endocrine disruption, etc.). For the present study, a quadratic function was used to shape the effects of salinity on ecosystems instead of the typical log-normal or log-logistic distribution applied in the SSD-based methods, by applying this approximation for the calculation of EC50s (and, therefore, the EFs). Nevertheless, log-logistic distributions were also applied to the chronic data gathered for each individual species to test the robustness of the approximation. The differences between the EC50s found applying log-logistic and quadratic fittings averaged 3.82% (data not shown), meaning that the quadratic approach is accurate to estimate effects at intermediate ranges of concentration in an LCA context (but might not be accurate for EC10s or EC20s). This approximation was used to estimate the EC50s and HC50s at both concentration ranges (low and high), but the SSD-based methodology fundamentals are maintained. Thus, the current study does not aim to question the SSD curves themselves, but to expand their scope to understand new ecological features. In fact, this quadratic approximation can be especially useful for systems from oligohaline to hyperhaline (see Appendix 5), but it might not be as accurate for freshwater systems. Here, the expected effects are linked to salinization, while impacts linked to a sharp freshening are unlikely. In fact, the most important impacts derived from anthropogenic activities and climate change in freshwater ecosystems are expected to be linked to salinization (Mosquera-Corral and Campos Gómez, 2021), where the SSDbased methodology can be a fair approximation to determine the effects of salinity increase if the effects of freshening are neglected. The truth is that the few studies considering the effects of salt variations in aquatic environments considered SSD approaches. In a wetland-related study (Amores et al., 2013), an SSD curve was constructed considering that anthropogenic activities (irrigation) were provoking salinization, which seems to be a fair approximation considering that the salt concentration in the water body had increased from 2.6 g/L in 1983 to 7.50 g/L in 2008. However, the salinity conditions of the wetland were oligohaline, so impacts related to freshening could also take place (due to rainfalls, for example). Therefore, the approximation was fair to evaluate salinization, but failed at evaluating the effects due to potential freshening. ALBA ROIBÁS ROZAS 104 In a brine disposal research work (Zhou et al., 2013), a concentration of 40 kg NaCl/m3 was chosen as EC50. Analogously to the wetland case, this approximation might be fair considering that desalination plants would discharge their briny effluents in marine waters, and the effects of this saline disposal are expected to be negative impacts linked to salinity increase. However, the modelling of the system is still not fully comprehensive if the low ranges of salinity are no included. In fact, a case where a transitional brackish ecosystem is subjected to freshening and where brine disposal is potentially beneficial could take place under this point of view (impacts would decrease due to brine disposal), so an approach where this is not considered is lacking some relevant information for the system. Finally, these divergent approaches hinder a discussion of the obtained CF. The wetland and brine CFs, in terms of potentially affected fraction of species (PAF), were 0.32 PAF·yr and 0.47 PAF·m3·day/kg, respectively, which are similar to the ranges obtained here (0.05 – 0.27 PDF·m3·day/kg). However, for the wetland, the units are not comparable, and, for both cases, the obtained CFs were at the midpoint level (measuring PAF, not PDF), so the discussion is not straightforward. Nevertheless, it is important to point out how the quantification of CFs to measure the effects of salinity variations can be relevant for the management of anthropogenic activities in sensitive ecological areas, such as transitional waters. In fact, the method described and applied here has the potential to support decision-making processes around effluent discharge, industrial stream management, brine disposal control, and dam flow regulation, by providing useful information about when and how to discharge these anthropogenic streams with s minimum or even a positive impact. To apply the CFs developed here, the life cycle inventory (LCI) shall record the mass of salts released to the aquatic environment per functional unit, acknowledging possible differences between the wet and the dry season when relevant and if information is available. 7.6. FUTURE OUTLOOK The presented novel approach proposes a model to shape the effects of variations in the concentration of an essential substance in the environment (Figure 7.4). Here, the effects in the ecosystem might not be always directly proportional to the pollutant concentration. Although the methodology was applied to salt (i.e., NaCl), the same principles can be implemented for other essential substances such as macronutrients, other salts, metals, and even resources such as water. Moreover, as the definition of salinity is broader than just sodium chloride, a comprehensive salinity assessment may shape the effects of varying the concentration of other substances. Furthermore, by including elements such as carbonate, nitrate and sulphate, a comprehensive study of the ecosystem salinity might extend the effects of the observed variations to other environmental categories (such as climate change, eutrophication, or acidification, respectively). In any case, the uniformization of the CFs for salinity impacts (expressed in the consensus units) opens a new pathway where the effects due to variations in the concentration of essential substances can be fully assessed for the first time. This pathway has a clear bottleneck, which might hinder its extensive application, and which is not new to the LCA community, i.e., regarding data availability. Fortunately, the access to information is becoming easier as science is becoming more accessible and