Beyond the combustion motor: a MCDM-based approach to analyse the alternative fuel vehicle decision from the customers' point of view

Beyond the combustion motor: A MCDM-based approach to analyse the alternative fuel vehicle decision from the customers’point of view David Boix-Cots a,* , Alessio Ishizaka b , Albert de la Fuente a,** , Pablo Pujadas c,d a Department of Civil and Environmental Engineering, School of Civil Engineering of Barcelona (ETSECCPB), Polytechnic University of Catalonia, UPC, C/ Jordi Girona 1-3, Barcelona, 08034, Spain b Department of Information Systems, Supply Chain and Decision Making, NEOMA Business School, 1 Rue du Mar´ echal Juin - BP 215, Mont-Saint-Aignan Cedex, 76825, France c Department of Project and Construction Engineering, School of Industrial Engineering of Barcelona (ETSEIB), Polytechnic University of Catalonia, UPC, Av. Diagonal 647, Barcelona, 08028, Spain d Group of Construction Research and Innovation (GRIC), C/ Colom, 11, Ed. TR5, Terrassa, Barcelona, 08222, Spain ARTICLE INFO Handling Editor: Giovanni Baiocchi Keywords: Sustainable mobility Low-carbon transportation Emission reduction MCDM MIVES Policy recommendations ABSTRACT In the transportation sector, one of the most polluting industries, the adoption of alternative fuel vehicles is essential for reducing carbon emissions and achieving sustainability goals outlined by international bodies, such as the United Nations Framework Convention on Climate Change. Despite their critical role, there is a significant lack of scientific studies that focus on direct consumer opinions regarding vehicle selection by fuel type, even with the upcoming European regulations poised to significantly change the vehicle market. This article presents a novel framework to study consumer perceptions regarding the choice of vehicles with different fuel types, considering their perspectives on all dimensions of sustainability. The framework is adaptable to any city and incorporates the opinions of local citizens to determine relevant indicators and their weights. To account for the inherent uncertainty within these indicators, Monte Carlo simulations and discrete profiles were carried out. The proposed framework is applied to the city of Barcelona as a case study. By actively involving citizens in the decision-making process, valuable insights are gained, leading to actionable recommendations for both the private sector and public administrators. These recommendations are aimed to address the primary concerns of potential customers and promote the adoption of alternative fuel vehicles, aiding in the formulation of policies and strategies that can effectively mitigate emissions and support the transition to a low-carbon economy. The results show that over the three period, conventional fuel vehicles (CFV) are the most preferred, even when only second-hand CFV are permitted. 1. Introduction The world is grappling with one of its most formidable challenges: climate change. Unanticipated floods, extended droughts resulting in water restrictions and hampered food production, or soaring heat waves elevating the risk of wildfires are just a few of the critical situations the world is currently confronting. It has long been recognized that a primary driver of this climate change is the excessive CO 2 emissions stemming from human activities (Kozera et al., 2024;Mac Dowell et al., 2017), functioning as a potent greenhouse gas (GHG). Among the contributors to GHG emissions, the transportation sector holds the rather unenviable distinction of being the foremost contributor, responsible for 27% of total emissions (US EPA, 2022). Therefore, it comes as no surprise that numerous automobile manufacturers and scientists have dedicated substantial resources and time to the research and development of novel alternative fuels designed to curtail CO 2 emissions (Seol et al., 2022). For instance, certain variants of traditional fossil fuels, such as oil and gas, have been engineered through processes based on liquefaction and compression, resulting in liquefied petroleum gas (LPG), liquefied natural gas (LNG), and compressed natural gas (CNG). The use of these fuels in alternative fuel vehicles (AFVs) has been extensively examined, primarily due to their potential to reduce both costs and GHG emissions compared to traditional fuels (Brzezi´ nska, 2019;Dimaratos et al., 2020; * Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (D. Boix-Cots), [email protected] (A. Fuente). Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro https://doi.org/10.1016/j.jclepro.2024.144564 Received 4 September 2024; Received in revised form 3 December 2024; Accepted 21 December 2024 Journal of Cleaner Production 486 (2025) 144564 Available online 24 December 2024 0959-6526/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). Savickis et al., 2021). However, despite these advantages, fossil fuels and their derivatives fall short of aligning with the long-term environmental sustainability goals set by governments and society, which highlights the need for solutions that deliver substantial and lasting reductions in GHG emissions (Díaz-Trujillo et al., 2019;Swanson et al., 2020). Among other alternatives, biofuels derived from biowaste have attracted significant attention, although their environmental impact varies depending on life-cycle analyses (Hoekman, 2009). Despite this, some of them, such as hydrotreated vegetable oil (HVO), have been rigorously evaluated for their suitability in AFVs (Dimitriadis et al., 2018). Nevertheless, two alternative fuels have long captured the attention of governments, society, and researchers due to their ability to produce zero emissions at the point of consumption. The first is electricity, initially incorporated into combustion vehicles as hybrid vehicles (ˇ Ceˇ rovský and Mindl, 2008a;Hollins, 2008;Karim and Shahid, 2018). This development led to full electric vehicles, which achieved substantial emission reductions (Michael et al., 2022;Petrovi´ c et al., 2020). The second is hydrogen, which produces only water as a by-product when used as an AFV fuel (Babadzhanova et al., 2022;JinKun, 2021). This diverse array of fuels has broadened consumer options, significantly influenced by the development of innovative fuel technologies, a choice that has expanded significantly with the development of innovative fuel technologies. However, the European Union (EU) regulations introduced in 2022 represent a turning point that amplifies this shift. The European Union (EU) executive body, the European Commission, proposed a 55% reduction in CO 2 emissions from new cars by 2030 compared to 2021 levels, and a complete elimination of CO 2 emissions from new vehicles by 2035, a proposal that was accepted by the European Parliament. These measures will fundamentally impact the decision-making process for all consumers, as traditional options like combustion-engine vehicles will gradually be removed from the market: by 2030, certain fossil fuel vehicles will no longer be available, and by 2035, the sale of new cars equipped with combustion engines will be prohibited (Commission, 2022). In parallel, these regulations are expected to drive the creation of new purchase incentives for AFVs and policies aimed at reducing the financial burden on consumers, such as tax reductions, to encourage the transition to more sustainable options. As a result, consumers are now compelled to actively engage with this decision-making process, which was once primarily driven by convenience or tradition. In this new scenario, understanding the decision-making process of consumers regarding the type of vehicle they will purchase has become an essential subject of study. Consumer opinions and perceptions play a crucial role in this process, as they can significantly influence market trends and the adoption of AFVs. By thoroughly studying these opinions, both businesses and governments can effectively promote the adoption of sustainable vehicles and identify key areas for further investment. For instance, as the phase-out of combustion engine vehicles progresses, consumers might lean towards acquiring AFVs, encouraged by governmental emphasis and support. However, consumer concerns and apprehensions, such as doubts about vehicle autonomy, the availability of charging infrastructure, and general uncertainties in the AFV sector, might push them towards the second-hand market instead. By understanding these consumer perspectives, stakeholders can address these issues more effectively, thereby facilitating a smoother transition to sustainable transportation and ensuring targeted investments that meet consumer needs and preferences. In this context, this paper endeavours to put forth a pioneering multiple-criteria decision-making (MCDM) approach rooted in a novel modified version of MIVES (Spanish acronym: Modelo Integrado de Valor para una Evaluaci´ on Sostenible, in English: Value Model for the Evaluation of Sustainability) to scrutinise, for the first time, potential consumer preferences regarding vehicle purchases in light of the evolving European regulatory landscape. The method incorporates the customer’s perspective on the three pillars of sustainability: economic, environmental, and social factors. In pursuit of this objective, various timeframes were examined to assist consumers in their decision-making processes while providing valuable insights to researchers, automobile manufacturers, and governmental bodies seeking a deeper understanding of consumer perspectives amidst the changing legislative framework. The rest of the paper is organised as follows: the next section presents a vehicle customer preference analysis literature review. Section 3defines the problem with its assumptions, limits and delimitations. Section 4explains in detail the proposed method. Section 5illustrates the technique with a case study. The final section gives a general conclusion. 2. Literature review In the literature, numerous analyses of alternative fuel-based vehicle systems are available. While many studies concentrate on the technical aspects of AFVs, such as enhanced batteries (Daems et al., 2024), comparisons of air toxic emissions (Winebrake et al., 2000;Winebrake et al., 2001), ozone reduction (C. C. Chang et al., 2001), decarbonization costs (Kim et al., 2024), or discrepancies between emissions and consumption in a laboratory and reality environment (Karabasoglu and Michalek, 2013), this manuscript’s objective is to examine customer behaviour. Thus, this section focuses on the factors influencing this behaviour, a topic that has received extensive attention from various perspectives. For instance, some studies have delved into the impact of public policy. Keith et al. (2020) analysed the diffusion of AFV technologies and customer opportunities, discussing both chances and existing barriers. These barriers encompass shared mobility, vehicle platforms, infrastructure deployment strategies, and policymaking. Mohammed et al. (2020) scrutinized this diffusion through the adoption of AFV fleets, exploring the barriers and enablers of adoption by firms. Other authors have concentrated on household vehicle choices in relation to various fuel taxes and how customers perceive them (De Borger and Rouwendal, 2014), the policy implementation cost of AFVs (Zeng et al., 2021), fuel efficiency (Schouten et al., 2014), or preference analysis through surveys (Batley et al., 2004). They concluded that AFVs require further infrastructure development and economic incentives to enhance their adoption. Regarding these incentives, Soto et al. (2018) examined public policies in Colombia to ameliorate customer attitudes and perceptions regarding AFV adoption, ultimately concluding that more infrastructure and awareness policies are imperative. Sang and Bekhet (2015) also analysed policies, obtaining a sensitive guideline for policy formulation and marketing recommendations to enhance AFV usage through a multiple regression model that employs private driver data. These findings align with broader literature reviews (Coffman et al., 2017;Liao et al., 2017), which emphasize the importance of infrastructure and policy incentives, such as tax reductions and charging infrastructure development, as key drivers of adoption. Other studies have taken a more specific focus on analysing the adoption factors of particular AFVs. Li et al. (2017) conducted a comprehensive review of electric vehicle adoption and categorized these factors into three main groups: demographic, situational, and psychological. Demographic factors pertain to individual or family characteristics, including age and education (Carley et al., 2013;Prakash et al., 2014). Situational factors are associated with vehicle characteristics, such as cost (Dumortier et al., 2015), driving range (Egbue and Long, 2012), or GHG emissions (Noppers et al., 2014). Finally, psychological factors are directly linked to customers’attitudes and perceptions and affect the decision process. For example, range anxiety, a significant psychological barrier, has been thoroughly investigated. Pevec et al. (2020) demonstrated that strategically expanding charging infrastructure could alleviate this concern, particularly if designed to mimic the convenience of traditional gas stations, even if it remains as a challenge in rural areas (Steadman and Higgins, 2022). Similarly, Ullah et al. (2022,2023) and Straka et al. (2020) emphasized how insufficient charging infrastructure creates a feedback loop that discourages adoption, underscoring the urgency of robust deployment strategies. Egbue D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 2 and Long (2012) revealed that customers’concerns about AFV battery range, price, and performance outweighed sustainable characteristics in significance, a categorization which was subsequently employed by Jia (2019) to examine AFV adoption in the United States of America through a novel prediction model aimed at determining AFV penetration in each state. In this context, Domarchi and Cherchi (2024) used a cross-nested logit model to prove that fuel type choice is highly correlated with car segment choice, with stronger correlations within the same segment and weaker correlations between segments sharing the same fuel type. Their findings suggest that promoting cleaner fuel alternatives may be more effective when considering substitutional patterns, while also highlighting the critical role of operating costs in influencing consumer decisions. Meanwhile, Qian and Soopramanien (2011) utilized an online survey to capture Chinese customer preferences, noting that most studies rely on empirical factors without directly engaging drivers. The critical factors identified in their study included cost, running expenses, the availability of charging facilities, autonomy, and incentives. Psychological factors have also been further explored by Pamidimukkala et al. (2023), who highlighted how first-hand experiences with electric vehicles can positively influence perceptions by altering stereotypes and demonstrating benefits such as low noise emissions and acceleration performance, finally deriving 63 factors influencing adoption, grouping them into technological, psychological, and contextual categories (Pamidimukkala et al., 2024). Among those, limited driving range, lengthy charging times, and high purchase prices were the most significant barriers, with environmental benefits acting as key motivators for adoption. In this vein (Coffman et al., 2017), noted that increasing the visibility of electric vehicles on the road could enhance awareness and interest. These customer preferences and adoption factors have also found application in various MCDM methods. For instance, the Preference Ranking Organization Method for Enrichment Evaluation (PROMETHEE) has been employed to assess different fuel-based vehicles in road transportation (Mohamadabadi et al., 2009). The Interval Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) has been utilized to compare AFVs in road transport alternatives, considering different life cycle emissions and cost scenarios within each interval (Streimikiene et al., 2013). This concept was further refined by incorporating fuzzy TOPSIS to address uncertainty (Liang et al., 2019) in ranking AFVs with respect to sustainable transportation. With the same objective, other methods like hierarchical hesitant fuzzy linguistic sets (Yavuz et al., 2015) and the Decision-Making Trial and Evaluation Laboratory (DEMATEL) method in combination with an Analytical Network Process (ANP) derived from expert opinions (D. S. Chang et al., 2015) have been utilized. More recently, the fuzzy Full Consistency Method (FUCOM-F) and the Neutrosophic Fuzzy Measurement Alternatives and Ranking according to the COmpromise Solution (MARCOS) have been jointly employed to prioritize six different AFVs in the United States, thoroughly investigating and selecting the drivers behind the decision (Pamucar et al., 2021). In summary, the studies and proposed methods mentioned above have addressed various issues and made significant contributions to understanding the factors influencing AFV adoption by customers. However, there are certain limitations in existing methods. First, many existing studies focus on static analyses that do not adequately account for the dynamic nature of evolving regulatory and technological landscapes. Second, psychological barriers, such as range anxiety or concerns over charging infrastructure availability, are often examined in isolation rather than being integrated holistically with other factors. Third, the implications of new European regulations, such as the 2030 reduction targets and the 2035 ban on combustion-engine vehicles, remain underexplored. These regulations are likely to significantly alter market dynamics, including technology availability, infrastructure development, and consumer priorities. Finally, many methodologies rely heavily on expert-driven evaluations, which may overlook the diverse experiences, preferences, and psychological considerations of actual consumers. This paper aims to address these gaps by proposing a comprehensive framework that integrates multiple customer factors, while explicitly considering the impacts of evolving regulations through multiple timespans. 3. Problem definition As outlined in the introduction, numerous variants of conventional fuels, specifically oil and gas, have been developed. Concurrently, the adoption of AFVs has increased as research into their utilization has advanced. In this study, five alternatives concerning different fuel vehicle options were defined: 1. Conventional fuel vehicles (Diesel/Gasoline, CVFs): These vehicles represent most contemporary automobiles equipped with internal combustion engines (ICE). These fuels give rise to substantial GHG emissions, serving as the primary impetus for the European prohibition. 2. Biofuel vehicles (BFVs): Biofuels are derived from biomass, and are employed in an ICE. The most prevalent biofuels include bioethanol, an alcohol synthesized from biomass via fermentation, and biodiesel, which is produced from oils and fats through a transesterification process. These fuels can be incorporated into conventional fuels in various concentrations, denoted as EX or BX. Here, E and B denote bioethanol and biodiesel, respectively, while X signifies the biofuel percentage within the blend. 3. Hybrid electric vehicles (HEVs): These vehicles employ an electric motor (EM) in tandem with an ICE to mitigate GHG emissions, primarily for shorter to moderate distances. They utilize a regenerative braking system to harness electricity from the kinetic energy generated by the vehicle during ICE operation, subsequently storing it in a battery. Additionally, there are plug-in hybrid electric vehicles (PHEVs), which can connect to the power grid to recharge their EM. 4. Electric vehicles (EVs): EVs are propelled by electric motors, deriving their power from energy stored in onboard batteries. While solar panels can be utilized to charge these batteries, the most prevalent charging points are typically found at private residences or service stations. 5. Hydrogen vehicles (HyVs): HyVs harness the chemical energy stored in hydrogen fuel for power generation. These vehicles can feature an EM that derives electricity from the chemical reaction between hydrogen and oxygen in a fuel cell, or they may employ an ICE fuelled by hydrogen. Having outlined the considered alternatives, it is essential to present the assumptions, delimitations, and limitations of the problem. The assumptions made in this study pertain to the potential changes that may occur in governmental policies, market dynamics, and infrastructure facilities: 1. Dynamic results: The newly accepted EU regulation states a reduction of 55% in CO 2 car emissions in 2030 and 100% in 2035 (European Commission, 2021). Hence, it is assumed that these years serve as the demarcation points for distinguishing among three distinct periods: pre-2030, spanning from 2030 to 2035, and post-2035. 2. Second-hand market: The authors assume that, in light of the EU regulations, certain customers might opt for the secondary vehicle market should newly available vehicles fail to align with their requirements. Starting in 2035 and beyond, it is postulated that all non-zero emission vehicles (CFVs, BFVs, PHEVs) can solely be acquired in the second-hand market. 3. Infrastructure differences and customer preferences: Given the presumed disparities in AFV infrastructure across various geographic regions, the proposed methodology should accommodate these D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 3 variations based on the city of origin for each customer. Consequently, the proposed method should be applied individually for each city. Considering that certain proposed alternatives involve multiple fuels and that an MCDM method might involve a substantial set of criteria, this study has outlined the following delimitations: 1. Fossil fuel variants: It is important to note that CNG, LPG, and LNG are not included as alternatives in this study. The authors consider that these technologies are unlikely to undergo further development due to their fossil fuel origins and high GHG emissions. Instead, they anticipate that these technologies, if used at all, will be gradually replaced by low-GHG emission biofuels. Therefore, these variants are not included in the study. 2. Biofuel variants: Only pure biofuels, E100 and B100, are considered. However, it acknowledges that there could be situations where a significant disparity exists in bioethanol or biodiesel infrastructure. In such cases, the analysis will employ the predominant biofuel available in the respective country or region. 3. Hybrid electric variants: This study specifically focuses on PHEVs as the chosen hybrid alternative. This choice is motivated by the similarity of PHEVs to conventional vehicles and the potential for enhanced electric charging infrastructure for PHEVs. 4. Car characteristics: The study’s primary focus is on analysing customer behaviour regarding fuel selection, rather than delving into specific vehicle types or classes. Consequently, for indicators requiring detailed vehicle information, data will be sourced either from existing databases or from the best-selling cars in the region to accurately represent the chosen area’s real circumstances. Given the limited availability of information on AFV characteristics, multiple data collection strategies may be necessary. For the same rationale, specific vehicle criteria like performance, speed, comfort, or aesthetics are considered negligible for the criteria selection. Finally, the only considered limitation encompass information constraints and the potential disparities that may arise between real-world developments and the estimates made for the specified time periods. Therefore, the proposed model must consider criteria and indicators suitable to account for uncertainties. 4. Methodology In this section, the proposed methodology is introduced. The first subsection outlines the MCDM method used for aggregating and processing indicator values. In the second subsection, the methods implemented to address uncertainty are described. Finally, the comprehensive proposed methodology is presented. 4.1. MIVES MIVES is a well-established multi-attribute utility theory (MAUT) MCDM method used to assess both homogeneous and heterogeneous alternatives aimed at fulfilling an overall objective (Boix-Cots et al., 2022), by typically incorporating the three pillars of sustainability across the evaluation. Among the main characteristics of MIVES, several features make it particularly well-suited for addressing the needs outlined in the previous section. First, MIVES exhibits high flexibility, allowing the inclusion of indicators with diverse units and scales, which can be assessed either qualitatively or quantitatively and are subsequently converted into a non-dimensional scale. Second, MIVES facilitates the creation of a static analytical framework while accommodating dynamic alternatives. This means that the inclusion of new alternatives or modifications to existing ones does not alter the evaluation of previously assessed alternatives. Third, MIVES employs value functions that enable precise representation of the effects produced by each indicator. Unlike linear models, these value functions allow for the modulation of preference curves, capturing non-linear relationships and providing a more accurate reflection of how varying indicator values impact the overall outcome. Finally, the method is designed to incorporate stakeholder preferences through the assignment of weights to criteria and indicators, ensuring that the final decision aligns with the priorities and values of all involved parties. For these reasons, MIVES has been selected as the methodological framework in this study, as it effectively addresses the complexity of evaluating alternatives under dynamic, multi-criteria conditions. For the application of MIVES, three steps are required: defining the hierarchical tree, specifying the value functions for each indicator, and assigning a weight to each element. The first step is the definition of the hierarchical tree, which is based on organizing the decision framework into requirements, criteria, and indicators (refer to Fig. 1A). These requirements represent high-level objectives that are divided into criteria, capturing specific aspects of the decision problem, and further subdivided into indicators, which measure relevant attributes. This hierarchical structure ensures a systematic and transparent evaluation, as each element’s contribution to the overall objective is explicitly defined. The second step is the definition of the value functions for each indicator. These functions are used to translate the raw values of indicators, which may have different units and scales, into a standardized, non-dimensional preference scale. In this study, the comprehensive value function (CVF) has been employed (Boix-Cots et al., 2024), a function recently introduced to MIVES to address the need for considering negative impact on sustainability performance of certain indicators (such as land degradation or water quality indexes). In this context, the CVF enables indicators to impact the overall index both positively and negatively, with a preference scale ranging from 1 to −1. This improvement allows for a more accurate and nuanced reflection of the alternative’s overall performance, particularly in contexts where indicators raise significant concerns, such as some of the identified in the previous section. The CVF offers flexibility by adopting multiple shapes to capture the diverse behaviours of preference outcomes (refer to Fig. 1B), a choice that significantly influences the results, as it determines how preference evolves across the range of indicator values. A concave curve is used when incremental improvements yield diminishing returns, while a convex curve is better suited for cases where small changes initially have less impact but become more significant as values increase. S-shaped functions combine these behaviours, representing scenarios where the impact of changes varies across the indicators’value range (Monserrat-L´ opez et al., 2024). Linear functions are applied for indicators with steady, proportional effects. Furthermore, all these functions can be inverted, enabling their application to negative impacts by reversing X max and X min , or a piecewise function can be defined. The CVF value function is defined by 6 parameters: n, K i , C i , X max ., X min . and P i as shown in Eq. (1). These are used to endow the function with positive or negative preference, maximum and minimum values, inflexion points and a certain shape. As is common in MAUT functions (Zionts, 1979), the values of these parameters are decided by the experts and decision-makers, who have studied the general problem and the effect of the indicator in detail. Vind =n*B*⎡ ⎢ ⎣1−e −K⋅(X−Smin C)P⎤ ⎥ ⎦(1) where. V ind is the preference index of the evaluated indicator, n is the positivity or negativity parameter. B is a parameter that allows the function to remain within 0 and 1, where 1 is assumed to be the highest preference value. This parameter is determined by Eq. (2), D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 4 S min is the minimum preference value point on x-axis, S max is the x maximum preference value point on x-axis, X is the indicator value that generates the value equal to Vind, P defines the shape of the curve. P =1 the curve is linear, P <1 the curve is concave for positive CVFs and convex for negative CVFs; P >1 the curve is S-shaped or convex for positive CVFs and concave for negative CVFs, C is a parameter that defines the inflexion x-value point for P >1 curves, K is a parameter that defines the y-value C point. B=1 ⎡ ⎢ ⎣1−e −K⋅(Smax−Smin C)P⎤ ⎥ ⎦ (2) The third step is the assignment of weights to each element within the hierarchical structure, ensuring that the decision reflects the relative importance of the criteria and indicators. These weights can be obtained with other decision-making techniques, such as the analytic hierarchy process, namely AHP (Saaty, 1980). When multiple stakeholders are involved in the decision, their opinions and preferences on the indicator weights must be aggregated. In this case, the Hierarchical Integration of Values and Evaluations under Social constraints (HIVES) method is used to integrate the preferences of multiple decision-makers (Boix-Cots et al., 2023). This method allows for mathematically combining diverse opinions while considering the main social choice axioms, ensuring that the final decision reflects a balanced and comprehensive evaluation that respects the input of all stakeholders involved. Furthermore, this method is specifically designed for multiple criteria problems, using combinatory analyses that evaluate criteria individually, maximizing the satisfaction of the solution. Once these steps are completed, the alternatives can be evaluated by transforming the values of their indicators into preference indexes using the respective value functions and aggregating them through the Weight Sum Model (WSM). 4.2. Uncertainty As described in Section 3, this study offers a holistic analysis of customer behaviour across multiple future periods. Within these periods, the current indicator data may undergo various fluctuations, requiring the introduction of uncertainty. To embed this uncertainty, the adoption of Monte-Carlo simulation (MCS) with a triangular distribution (MCS-TD) is proposed. MCS-TD stands out as an appropriate tool for handling quantitative data where only approximate knowledge is available for three key parameter references: the minimum (L), the most probable (M), and the maximum (H) values that the variable can encompass (Hihn and Lum, 2004). For example, if an indicator measures fuel costs in a future scenario, experts might estimate a minimum value of 1.20 € per litre, a most probable value of 1.50 € , and a maximum of 1.80 € . MCS-TD generates numerous random samples from these inputs to model –under the hypothesis of TD probability density function –potential fluctuations in fuel costs and assesses their impact on the overall decision. This probabilistic approach provides a realistic representation of uncertainty and allows the evaluation of the value distribution across different scenarios (refer to Fig. 2) when there is limited information on the magnitudes that the involved variables can present. These parameter references are established with the insights of experts, aligning seamlessly with the proposed model’s requirement for expert assumptions. MCS has already proven its effectiveness when integrated with classic MIVES (De La Cruz et al., 2014;Jato-Espino et al., 2014). However, some indicators do not rely on quantitative data; instead, they are based on whether specific requirements have been met. To Fig. 1. A) General MIVES framework tree. B) CVF function. Fig. 2. Triangular probability density function. D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 5 address these indicators, the use of discrete profile distributions (DPDs) is proposed. This approach is employed to manage uncertainty by considering the probability of achieving specific profiles, as indicated by the decision maker (Kunsch and Ishizaka, 2018). For instance, consider an indicator assessing whether a charging infrastructure is available in rural areas. The decision maker may specify a set of profiles (P), such as “No infrastructure,” “Limited infrastructure,”or “Adequate infrastructure,”with probabilities of occurrence (V) assigned to each profile (e.g., 20%, 50%, and 30%, respectively). This approach models the uncertainty in meeting specific qualitative requirements and transforms these probabilities into a weighted average that reflects the overall likelihood of achieving the desired profile. DPD is particularly useful when there is hesitance in the indicator values, as they resemble a binary profile, similar to a probability function expressed as percentages. These hesitant values are used to compute the grade point average (GPA). Considering a profile set P ={P 1, P 2, P 3 , …, P m };r∈P and its corresponding membership value set V ={V 1, V 2, V 3 ,…, V m }; r ∈V, for each indicator the GPA is defined as follows: GPA =∑ m r=1 Pr⋅Vr(3) where: m is the maximum number of profiles, P r is the “r”profile, V r is the indicator’s profile “r”membership percentage degree. To simplify, the GPA aggregates the probabilities of each profile and calculates an overall preference score for the indicator. This allows qualitative indicators, such as the likelihood of meeting a specific condition, to be integrated seamlessly into the analysis. Therefore, indicators can be expressed as exact data values, ranges that include minimum, most likely, and maximum values, or as DPD profile values. These values are aggregated using the MIVES method to derive an index, which encapsulates not just a precise probability value for selecting an alternative, but the entire distribution of selection probabilities. 4.3. Proposed methodology This study proposes a structured, multi-stage methodology to evaluate customer behaviour under evolving conditions, integrating both quantitative and qualitative data while addressing the specific assumptions outlined in Section 3. The framework consists of four main stages (refer to Fig. 3), each contributing to the comprehensive evaluation of the proposed alternatives. The preparation stage establishes the foundation for the entire analysis by identifying the alternatives, assumptions, limitations, and constraints through a literature review of the problem and its specifications. In this study, the preparation stage is closely tied to the third assumption, which emphasizes applicability on a city-by-city basis and the importance of involving citizens in assigning weights to the indicators. Additionally, key considerations, such as managing uncertainty and outlining potential scenarios, are defined at this stage. The first stage involves constructing the hierarchical MIVES tree. Considering the three sustainability requirements (economic, ambiental and societal), citizens’perceptions in the selected location are collected to identify relevant indicators, ensuring that the framework reflects local priorities and contextual realities. Structured surveys are recommended to capture comparable responses, which are then translated into measurable indicators, which are grouped under criteria. The second stage focuses on assigning values to the indicators and defining their corresponding value functions. Expert input is critical in this stage, as assumptions specific to each indicator must be established to account for uncertainties, allowing to quantify indicator values for each future period. The third stage involves determining the relative importance of each indicator by assigning weights. Citizen input plays a key role here, ensuring that the weights reflect the community’s priorities and values. The AHP and HIVES methods are recommended for deriving these weights, as they provide a consistent and rational method for assessing and aggregating preferences. Finally, the fourth stage is based on the application of the developed value functions and the MIVES tree to evaluate the alternatives. By aggregating the criteria weights and transforming the indicator values, sustainability index distributions are obtained, providing actionable recommendations for decision-making. 5. Case study A real case study is provided to illustrate the proposed approach, which is implemented in Barcelona, Spain. Barcelona, a Mediterranean city, has a population of 1.6 million and covers an area of 101.9 km 2 . It stands as Spain’s second-largest city and serves as the economic hub of Catalonia, its autonomous community. Consequently, this study involves the city’s residents in the data collection processes. Considering the preparation stage done in section 3, the predominant BF alternative has to be selected. Barcelona primarily relies on Bioethanol as its most prevalent Biofuel option, as opposed to Biodiesels like HVO100, which are more common in Northern Europe (glpautogas.info, 2022). Hence, the alternative set encompasses CFVs, BFVs (E100), PHEVs, EVs and HyVs. With this information, the following framework steps can be applied. 5.1. Stage 1 For the present study, the authors have used the sustainability requirements to generate a semi-structured interview for the first data collection process carried out on Barcelona citizens, a qualitative research method that combines predefined questions with the flexibility Fig. 3. Scheme and stages of the proposed methodology. D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 6 to explore additional insights provided by the participants, including broad, open-ended questions that ensure that respondents can freely express their perspectives without being constrained by overly rigid categories. This approach was chosen because it allows for both the collection of standardized data (to ensure comparability) and the exploration of nuanced, context-specific perceptions. The questionnaire (see Table 1) was constructed to align with the three sustainability requirements (economic, environmental, and social) and to capture citizens’primary concerns and preferences regarding AFVs. Due to the primary objective of this article being to show a new consistent and robust method, and considering the academic nature of this study, the sample size established was sufficient to ensure heterogeneity. In total, 82 citizens answered the questions shown in Table 1 (59 men and 23 women, with averages of 40 and 27 years). The responses provided insights into specific characteristics that are common concerns among those interviewed. It was evident that the cost of both the vehicle and fuel were critical factors in their decision-making process. A significant majority, 63 respondents (76.83%), expressed concerns about the expenses associated with repairing and maintaining an AFV, emphasising the need for more knowledge in this regard. Meanwhile, 33 respondents (40.24%) considered AFV incentives related to vehicle taxes as a determining factor. Regarding environmental interests, there was unanimous agreement among respondents on considering fuel emissions as a decisive factor. However, only six respondents (7.32%) highlighted the importance of factoring in the environmental impact of fuel production. There were also concerns about technical factors that could impact the social aspects of the respondents’lives, which exhibited a degree of uniformity. All respondents considered factors such as charging time, availability of charging points, and vehicle autonomy as current areas of concern. However, it is important to emphasize that their concerns were not evenly distributed. A significant majority of the respondents expressed that issues related to charging time and the accessibility of charging points were not only major concerns but also critical barriers that could deter them from adopting AFVs, which were perceived as significant enough to outweigh other considerations. To process these responses, a thematic analysis was conducted. This approach involves coding the responses to identify patterns or categories that emerge from the data. Each response was reviewed, and recurring factors were categorized into themes corresponding to the sustainability requirements. Once these themes were established, the methodological transition to the MIVES tree was performed. Each theme was translated into measurable indicators and categorized under the appropriate criteria (see Fig. 4), ensuring alignment with the hierarchical structure of the MIVES framework. 5.2. Stage 2 At this stage, the values and value functions associated with the indicators depicted in Fig. 4 must be established. In the following, each indicator is defined, along with its data collection processes and its value functions. 5.2.1. Vehicle cost (I 1 ) This indicator specifies the average vehicle cost for each alternative. Following the 4th delimitation, the vehicle values are defined as follows. In the case of CFVs, the known value is 22,755 € for new passenger vehicles, a figure provided by the Spanish tax agency in April 2022 (Agencia Tributaria, n.d.). However, obtaining average data for other alternatives is more challenging since there are no readily available databases in Spain for comparison. This requires employing various strategies to gather information. For BFVs, an estimated cost of 1200 € is added due to the adaptation kit required to modify the car’s characteristics. This adaptation is necessary because blended ethanol can be used in self-ignition engines, while E100 vehicles require spark ignition engines (Chłopek, 2007). As for PHEVs, the average cost is calculated based on the three best-selling cars in Spain for this category: Peugeot 2008, Kia Xceed, and Mercedes A-Class, with an average cost of 36,455 € . For EVs, the average cost is determined using the Tesla Model 3, FIAT 500 electric, and KIA e-Niro, with an average price of 40,400 € . Finally, HyVs, represented by the Toyota Mirai and Hyundai Nexo, have an average cost of 69,225 € . Additionally, financial incentives from the Spanish government, such as the Moves III Plan, can impact the vehicle’s cost. Under this plan, buyers receive 7000 € (for EVs and HyVs) or 5000 € (for PHEVs) if they scrap their old car. These incentives are reduced to 4500 € (for EVs and HyVs) or 2500 € (for PHEVs) otherwise. The average of these values has been used to adjust the current cost of PHEVs, EVs, and HyVs in the first period, considering that these incentives are likely to disappear after 2030. To analyse future vehicle price values, the authors have considered the following variables: •An annual price increase of 7.3% due to the Consumer Price Index (CPI) sourced from the National Statistical Institute of Spain. This increase is applied differently depending on scenarios: it’s fully applied to worst-case scenarios, halved for the most likely scenarios, and quartered for the best-case scenarios. •A scarcity factor in the production of CFVs, BFVs, and PHEVs is considered until 2035. This factor results in an annual 1% price increase. After 2035, this factor transitions to a societal factor: It is hypothesized that some vehicle owners will opt to retain their existing fuel types, leading to a 5% price increase. •Conversely, a factor promoting increased production of EVs and HyVs is considered, leading to an annual 1% reduction in their prices. After 2035, this factor is doubled. Therefore, the authors have considered the vehicle’s possible cost values shown in Table 2. 5.2.2. Incentives (I 2 ) This indicator assesses the potential for receiving incentives on Table 1 Questions used to gather the citizens’perceptions on AFVs. Requirement Question Economic What economic factors would you consider when purchasing an AFV? Ambiental What environmental characteristics would you consider when purchasing an AFV? Social What socio-technical factors would be of utmost concern to you when buying an AFV? Fig. 4. Proposed MIVES tree. D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 7 recurring taxes. In Barcelona, the sole recurring tax is the motor vehicle tax, which is an annual payment based on a vehicle’s effective power. Currently, the tax discounts in Barcelona are as follows: 0% for CFVs, 75% for BFVs and PHEVs, and 100% for EVs and HyVs. It must be noted that these values are subject to change, transitioning from being a sustainability commitment to a mandatory requirement over time. Furthermore, it also must be considered that this tax contributes significantly to the city council budget, amounting to 5%, making it challenging to sustain multiple tax discounts. A DPD system is employed to analyse multiple future scenarios, which are categorized into four cases based on quartiles (0–25%, 25–50%, 50–75%, and 100% tax discount). The DPDs considered for analysis are presented in Table 3, following the subsequent criteria: •CFVs are assumed to remain unchanged, as they represent the alternative that the government aims to phase out. •BFVs, a decline in the discount class is considered during the initial period due to their ICE usage. This decline becomes more significant in the second period, ultimately leading BFVs to be classified in the 1st quartile from 2035 onwards. A similar approach is applied to PHEVs, with a relatively smaller decrease due to their electric motor. •Bonuses for EVs and HyVs will be gradually phased out as these alternatives become more commonplace. However, it was found representative to consider that HyVs may retain some incentives due to their limited adoption in Spain. Following Eq. (3), the discrete values of each alternative in each period are CFV [1.0, 1.0, 1.0], BFV [1.7, 1.2, 1.0], PHEV [1.5, 1.3, 1.0], EV [3.8, 3.0, 1.75], HyV [3.8, 3.0, 1.75]. 5.2.3. Fuel cost (I 3 ) This indicator specifies the fuel cost for each alternative in the form of fuel cost per 100 km due to their differences in units of measurement (litres and kW). According to data from the Spanish Ministry for Ecological Transition and Demographic Challenge as of September 2022, the fuel cost per 100 km is as follows: 10.86 € for CFVs, 8.46 € PHEVs, and 6.06 € for EVs. For BFVs, despite their potential cost savings of up to 25% when compared to conventional fuel, they can consume up to 30% more fuel than CFVs. Therefore, the CFV’s value is used as a reference for BFVs in this analysis. Finally, HyVs have an estimated consumption cost of approximately 9 € per 100 km. To analyse future fuel cost values presented in Table 4, the authors have considered the following variables: •Statista, a specialized data collection website (www.statista.com), provided data to calculate the percentage variance for each type of fuel. Different periods are considered, accounting for events such as the Russia-Ukraine conflict’s impact on Spain’s fuel prices. The first period assumes the Russian oil and gas sub-supply cut-off is in effect, while the second one considers the end of the war and price stabilization. The third period includes a tax increase for CFVs and BFVs while reducing prices for EVs and HyVs. This change is driven by the European Union’s directives to reduce dependence on fossil fuels and promote renewable energy investment. •For CFVs and BFVs, the price variance is based on estimated retail prices per litre. PHEVs use an average between CFVs and EVs. HyVs, as hydrogen is primarily obtained through electrolysis, share variables with EVs. 5.2.4. Reparation (I 4 ) This indicator uses maintenance cost data from Spencer (2021). CFVs and EVs have maintenance costs of 28,538.06 € and 17,235.85 € , respectively, for a lifespan of 482,803 km. It is assumed that BFVs share the same maintenance cost as CFVs, and PHEVs have an average of CFVs and EVs. HyVs incur 18% higher costs for some commonly replaceable components compared to electric components (Offer et al., 2010), which leads to a maintenance cost of 20,338 € at 482,803 km. According to Eurostat, the average annual distance travelled in Europe is 10,000 km (Focas and Christidis, 2017). Considering the car characteristic delimitation mentioned in section 3, an average lifespan of 150,000 km is assumed for each vehicle type. Therefore, the estimated lifecycle maintenance costs for the different alternatives are: 8866 € for CFVs and BFVs, 7111 € for PHEVs, 5355 € for EVs, and 6318.82 € for HyVs. To obtain data presented in Table 5, the authors have considered the following variables: •The same CPI values are applied across all vehicles as they have a similar impact on spare parts and mechanic prices. •Factors affecting CFVs, BFVs, and PHEVs are assumed to be similar, considering factors like scarcity and societal demand. •EVs and HyVs maintain their reduction factor in maintenance costs. This is attributed to an expected increase in the number of mechanics capable of performing repairs (Wr´ oblewski et al., 2021). 5.2.5. Emissions (I 5 ) This indicator assesses the GHG emissions produced by each fuel Table 2 Alternatives likely cost regarding each period. Vehicle t <2030 2030 <t<2035 t >2035 CFVs [22755, 30162, 35976] [30469, 35452, 45419] [24765, 29038, 37586] a BFVs (E100) [23955, 31922, 38112] [32076, 37322, 47814] [26287, 30824, 39897] a PHEVs [32705, 43351, 51707] [48813, 56797, 72764] [39675, 46521, 60215] a EVs [34690, 41125, 49988] [44444, 53300, 71013] [38782, 47638, 65351] HyVs [63475, 75250, 91467] [76078, 91239, 121559] [66387, 81547, 111868] a Second hand market. Table 3 Incentives discrete profile distributions for each alternative and period. t<2030 2030 <t<2035 t >2035 Case 1 2 3 4 1 2 3 4 1 2 3 4 CFVs 100.00% 100.00% 100.00% BFVs 30.00% 70.00%   80.00% 20.00%   100.00%  PHEVs 50.00% 50.00%   70.00% 30.00%   100.00%  EVs 20.00% 80.00% 25.00% 50.00% 25.00% 50.00% 25.00% 25.00% HyVs 20.00% 80.00% 25.00% 50.00% 25.00% 25.00% 50.00% 25.00% Table 4 Fuel cost per 100 km in € for each alternative and period. Vehicle t <2030 2030 <t<2035 t >2035 CFVs [8.99, 10.86, 13.02] [7.29, 8.39, 9.20] [8.02, 9.23, 10.11] BFVs (E100) [8.99, 10.86, 13.02] [7.29, 8.39, 9.20] [8.02, 9.23, 10.11] PHEVs [6.71, 8.46, 10.86] [4.19, 5.94, 8.33] [4.50, 6.18, 8.41] EVs [4.43, 6.07, 8.71] [1.09, 3.49, 7.46] [0.98, 3.14, 6.71] HyVs [6.58, 9.00, 12.92] [1.62, 5.18, 11.07] [1.29, 4.14, 8.86] D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 8 type. To determine its potential values, a DPD is employed to analyse various future scenarios in alignment with European regulations that support the principles of this manuscript. These scenarios adhere to the maximum emissions permitted by these regulations: (i) 2021 CO 2 limit (95 gr/km), (ii) 2030 target (55% CO 2 reduction compared to 2021), and (iii) the 2035 goal: 100% CO 2 reduction compared to 2021. In accordance with the DPDs presented in Table 6, the authors have considered the following: •CFVs conform to European regulations directly. An intermediate value is assumed for BFVs, as some studies have questioned the 90% CO 2 reduction (Ward and Singh, 2002), analysing the values in non-experimental driving (Liaquat et al., 2010). PHEVs also assume an intermediate value, recognising their emissions are highly dependent on driving patterns (ˇ Ceˇ rovský and Mindl, 2008b;Wolfram and Lutsey, 2016). EVs and HyVs consistently achieve 100% CO 2 reduction in all periods. •In the second period, it is assumed that manufacturers of CFVs, BFVs, and PHEVs will meet European regulations, and no new vehicles with lower emissions will be developed due to regulatory restrictions. •In the third period, a second-hand market for CFVs, BFVs, and PHEVs is supposed to emerge, with values set at 2/3 of the first period and 1/3 of the second period. Following Eq. (3), the discrete values of each alternative in each period are CFV [1.0, 2.0, 1.67], BFV [1.7, 2.0, 1.9], PHEV [1.6, 2.0, 1.87], EV [3.0, 3.0, 3.0], HyV [3.0, 3.0, 3.0]. 5.2.6. Charging time (I 6 ) This indicator assesses the time required for full vehicle charging. It compares the charging times of various alternatives to conventional CFVs, which is about 3–5 min, according to one of Spain’s leading electric charging point and gas station providers (Repsol, n.d.). Additionally, electric charging points are categorized into four types: slow charging (4–8 h), semi-fast charging (1–3 h), fast charging (approximately 40 min), and ultra-fast charging (5–10 min). To determine the values in Table 7 related to charging times, the authors have considered the following: •BFVs and HyVs have the same charging time as CFVs, given their similar charging processes. Additionally, since PHEVs have both fuel systems, they adopt the charging times of CFVs, which is considered the most common out-of-home procedure. •Slow charging is not considered, as it is typically utilized in domestic installations. Semi-fast charging, fast charging, and ultra-fast charging are assigned values of 2 h, 40 min, and 10 min, respectively. •While the ability of vehicles to use certain charging infrastructures depends on their specifications, it is assumed that these will gradually transition to accept fast and ultra-fast charging. 5.2.7. Charging points (I 7 ) This indicator examines the availability of charging points for each alternative, using data sourced from the Ministry for Ecological Transition and Demographic Challenge in (Ministerio para la transici´ on ecol´ ogica y el reto demogr´ afico, n.d.), which provides comprehensive information on petrol stations and their distribution across the country. In the metropolitan area of Barcelona, the following stations are currently available: 81 for CFVs, 1 for BFVs, and 1 for HyVs. Additionally, the city council’s network of charging points (Plug-in Barcelona) is planned to obtain the values of 43 semi-fast charge and 41 fast charge points for EVs and PHEVs. The values presented in Table 8 are determined based on the following considerations: •The number of CFV stations remains constant as no additional permits for new stations are issued in Barcelona. PHEVs and BFVs are assumed to utilize CFV charging infrastructure, considering that the number of stations offering BFVs will increase during the initial period but will level off in the second period, aligned with the 2035 ban. Table 5 Reparation indicator data, considering total lifecycle vehicle maintenance cost. Vehicle t <2030 2030 <t<2035 t >2035 CFVs [8866, 12165, 14754] [12123, 14226, 18443] [12566, 14669, 18876] BFVs (E100) [8866, 12165, 14754] [12123, 14226, 18443] [12566, 14669, 18876] PHEVs [7111, 9756, 11832] [9722, 11409, 14783] [10078, 11765, 15139] EVs [5355, 6490, 8054] [5929, 7200, 9741] [5233, 6504, 9044] HyVs [6319, 7658, 9504] [6997, 8496, 11494] [6175, 7674, 10672] Table 6 Discrete profile distributions for each alternative and period. T<2030 2030 <t<2035 t >2035 Case 1 2 3 1 2 3 1 2 3 CFVs 100.00%    100.00% 66.67% 33.33%  BFVs 30.00% 70.00%   100.00% 10.00% 90.00%  PHEVs 40.00% 60.00%   100.00% 13.33% 86.67%  EVs   100.00%   100.00% 100.00% HyVs   100.00%   100.00% 100.00% Table 7 Charging time values of each alternative compared to CFVs. Vehicle t <2030 2030 <t<2035 t >2035 CFVs [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] BFVs (E100) [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] PHEVs [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] EVs [8.00, 24.00, 24.00] [2.00, 8.00, 24.00] [2.00, 8.00, 8.00] HyVs [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] [1.0, 1.0, 1.0] Table 8 Charging points indicator values. Vehicle t <2030 2030 <t<2035 t >2035 CFVs [81, 81, 81] [81, 81, 81] [81, 81, 81] BFVs (E100) [1, 20, 40] [20, 40, 60] [40, 50, 60] PHEVs [81, 81, 81] [81, 81, 81] [81, 81, 81] EVs [41, 50, 60] [50, 80, 90] [80, 100, 120] HyVs [1, 10, 20] [10, 30, 50] [30, 40, 60] D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 9 Straka, M., De Falco, P., Ferruzzi, G., Proto, D., Van Der Poel, G., Khormali, S., Buzna, L., 2020. Predicting popularity of electric vehicle charging infrastructure in urban context. IEEE Access 8. https://doi.org/10.1109/ACCESS.2020.2965621. Streimikiene, D., Baleˇ zentis, T., Baleˇ zentiene, L., 2013. Comparative assessment of road transport technologies. Renew. Sustain. Energy Rev. 20, 611–618. https://doi.org/ 10.1016/J.RSER.2012.12.021. Swanson, C., Levin, A., Stevenson, A., Mall, A., Spencer, T., 2020. Sailing to nowhere: liquefied natural gas is not an effective climate strategy. www.suerossi.com. Ullah, I., Liu, K., Layeb, S.B., Severino, A., Jamal, A., 2023. Optimal deployment of electric vehicles’fast-charging stations. J. Adv. Transport. 2023. https://doi.org/ 10.1155/2023/6103796. Ullah, I., Liu, K., Yamamoto, T., Zahid, M., Jamal, A., 2022. Prediction of electric vehicle charging duration time using ensemble machine learning algorithm and Shapley additive explanations. Int. J. Energy Res. 46 (11), 15211–15230. https://doi.org/ 10.1002/ER.8219. US EPA, 2022. Fast facts on transportation greenhouse gas emissions. https://www.epa. gov/greenvehicles/fast-facts-transportation-greenhouse-gas-emissions. Ward, O.P., Singh, A., 2002. Bioethanol technology: developments and perspectives. Adv. Appl. Microbiol. 51. https://doi.org/10.1016/S0065-2164(02)51001-7. Winebrake, J., He, D., Wang, M., 2000. Fuel-Cycle Emissions for Conventional and Alternative Fuel Vehicles: An Assessment of Air Toxics. http://www.ipd.anl.gov/. Winebrake, J.J., Wang, M.Q., He, D., 2001. Toxic emissions from mobile sources: a total fuel-cycle analysis for conventional and alternative fuel vehicles. J. Air Waste Manag. Assoc. 51 (7), 1073–1086. https://doi.org/10.1080/ 10473289.2001.10464325. Wolfram, P., Lutsey, N., 2016. Electric Vehicles: Literature Review of Technology Costs and Carbon Emissions. Wr´ oblewski, P., Dro˙ zd˙ z, W., Lewicki, W., Dowejko, J., 2021. Total cost of ownership and its potential consequences for the development of the hydrogen fuel cell powered vehicle market in Poland. Energies 14 (8). https://doi.org/10.3390/en14082131. Yavuz, M., Oztaysi, B., Cevik Onar, S., Kahraman, C., 2015. Multi-criteria evaluation of alternative-fuel vehicles via a hierarchical hesitant fuzzy linguistic model. Expert Syst. Appl. 42 (5), 2835–2848. https://doi.org/10.1016/j.eswa.2014.11.010. Zeng, Y., Tan, X., Gu, B., Guo, J., Jiang, J., Wang, D., Tang, J., 2021. Cost-effectiveness analysis on improving fuel economy and promoting alternative fuel vehicles: a case study of Chongqing, China. J. Clean. Prod. 323, 129075. https://doi.org/10.1016/J. JCLEPRO.2021.129075. Zionts, S., 1979. MCDM—if not a roman numeral, then what? Interfaces 9 (4), 94–101. https://doi.org/10.1287/inte.9.4.94. D. Boix-Cots et al. Journal of Cleaner Production 486 (2025) 144564 16