Comparative Analysis of Electric Buses as a Sustainable Transport Mode Using Multicriteria Decision-Making Methods

Abstract

The transition to electric public transportation is crucial for reducing the carbon footprint and promoting environmental sustainability. However, successful implementation requires strong public policies, including tax incentives and educational programs, to encourage widespread adoption. This study identifies the optimal electric bus model for Cuenca, Ecuador, using the multicriteria decision-making methods PROMETHEE and TOPSIS. The evaluation considers four key dimensions: technical (autonomy, passenger capacity, charging time, engine power), economic (acquisition, operation, and maintenance costs), social (community acceptance and accessibility), and environmental (reduction of pollutant emissions). The results highlight passenger capacity as the most influential criterion, followed by autonomy and engine power. The selected electric bus model emerges as the most suitable option due to its energy efficiency, low maintenance costs, and long service life, making it a cost-effective long-term investment. Additionally, its adoption would enhance air quality and improve the overall user experience. Beyond its relevance to Cuenca, this study provides a replicable methodology for evaluating electric bus feasibility in other cities with different geographic and socioeconomic contexts.

1. Introduction

The growing global concern for environmental sustainability and the urgent need to reduce polluting emissions have made the transition to more ecological transport systems imperative and urgent. In this regard, the electrification of public transport, specifically buses, is a strategic move to mitigate the negative impacts of private car transport and the environmental damage caused by internal combustion engines. The selection of electric buses (EB) represents a fundamental step in this transition, as it involves the evaluation and choice of vehicles that not only meet operational requirements but also contribute appropriately to reducing the carbon footprint. However, this decision-making process faces increasing complexity due to the diversity of factors involved, ranging from technical and economic considerations to environmental and social criteria.

In the city of Cuenca, the current transportation system faces challenges such as congestion and pollution due to the growth of the vehicle fleet. Currently, 475 diesel buses emit 112 tons of CO₂ daily and consume 11,175 gallons of diesel [1]. The motorized transport sector is responsible for generating 94.5%, 71.2%, and 30.2% of carbon monoxide, nitrogen oxides, and sulfur dioxide emissions, respectively. Fossil fuels for transportation account for 59.9% of total energy consumption. Cuenca is located at 2500 meters above sea level and near the equator, and, thanks to its excellent climatic conditions, which reduce the need for air conditioning, the main energy demand is focused on transportation [2,3].

Ecuador’s energy efficiency policies aim to prioritize the adoption of electric vehicles in public transport and heavy cargo transport. Progressive standards for consumption and emissions have been established for new vehicles, as well as energy efficiency labels. A plan for scrapping is expected to be implemented to replace old vehicles with electric vehicles, and deadlines have been set for all public transport vehicles on the Ecuadorian mainland to be electric. In the Ecuadorian context, the Organic Law of Energy Efficiency (2019) establishes the obligation to transition the bus fleet to electric vehicles. However, there is still uncertainty in the commercial sector regarding the effectiveness of the implementation of this technology. The coming into force of the Organic Law of Energy Competition (2024) is expected to have a direct impact on the transition to EB in Ecuador [4,5].

Specifically, the Organic Law of Energy Efficiency provides the legal framework to promote the efficient use of energy in all its forms. Chapter III, Article 14, establishes that, from the year 2025 (currently extended to 2030), all urban and inter-parish transport vehicles incorporated on the Ecuadorian mainland must use electric propulsion. Similarly, the municipality of the city of Cuenca, through its mobility plan, promotes sustainable mobility [4,5].

In Cuenca, other alternatives such as fuel cell buses have been studied [1]. However, these require a hydrogen refueling infrastructure that is currently under development and limited. Additionally, implementation costs are high due to hydrogen production and the purchase of the buses, which are more expensive than electric ones. Hydrogen storage is bulky and may require costly solutions, and the availability of bus models is limited, as companies mainly develop larger versions due to hydrogen tanks. In contrast, EB have a more developed charging infrastructure and lower implementation costs, although they face challenges in range and charging times. In conclusion, hydrogen buses offer advantages in terms of range and quick refueling, but electric buses are a more accessible and scalable option in the short term.

Hydrogen and electric buses are considered effective options for decarbonizing the public transportation sector. However, their environmental benefits are closely tied to the energy sources used for hydrogen production or battery charging. From an economic standpoint, hydrogen technology is more expensive than electric buses [6]. The literature indicates that, so far, hydrogen systems and, to a lesser extent, electric systems are still more costly than diesel, natural gas, and biomethane powertrain technologies. This is due to the high investment costs associated with fuel cells, electrolysers, and batteries, which means these technologies are not yet economically competitive without incentive programs. From a technological perspective, hydrogen technologies are still in their early stages of development, which brings uncertainty regarding their safety and reliability. Consequently, from economic, environmental, and technical standpoints, electric buses are currently the preferred solution for urban public transportation in the short term [7].

In Ecuador, in the short term, the barrier to achieving the inclusion of electric vehicles is the cost of the infrastructure required by these units. Therefore, the main motivation for this research is to propose a methodology that allows the identification of the most suitable electric bus to introduce in the city of Cuenca (Ecuador) and determines a set of criteria to determine the most favorable technology. Decision-making tools are fundamental, as a successful decision in a complex area involving various criteria determines the success of a project. In this work, decision-making is modeled to ensure that it is easily understood and simple to implement. Cuenca, with its medium-sized city status, is an ideal scenario for which to carry out this comparative study. The research developed in this area will not only generate specific results for the selection of EB in the city of Cuenca itself but could also serve as an example for other cities with similar characteristics considering the transition to electric public transport.

In this context, multicriteria methods offer an effective solution to address the inherent complexity of selecting EB. These approaches allow for the integration and weighting of multiple factors, providing an analytical framework that facilitates informed and balanced decision-making. By employing multicriteria evaluation techniques, the goal is to impulse clean public transport systems with EB, considering aspects such as range, energy efficiency, total cost of ownership, and other crucial elements simultaneously.

2. Literature Review

The transport sector around the world emits approximately 8 billion tons of greenhouse gases annually, accounting for a quarter of all emissions [8]. It is the primary consumer of fossil fuels, and leads in CO₂ emissions [9]. Electric buses are a sustainable public transport option compared to those that use diesel. In Bogotá, Colombia, fleets of EB have been implemented, achieving an 8 % reduction in carbon emissions and improving air quality [10]. The sale of electric vehicles in Latin America is relatively low, with the exception of in Colombia and Mexico, where sales are significantly higher [11]. China, France, and the United States have also adopted EB to replace diesel units [12]. Most electric bus fleets are in the demonstration phase and are modest, except in Chile. In Ecuador, the Organic Law on Energy Efficiency mandates the transition of public transport buses to electric ones starting in 2030, although there is still uncertainty about the implementation of this technology. To evaluate this project, a multicriteria analysis has been chosen, an effective tool for analyzing and comparing different electric bus options. Currently, environmental sustainability is a global concern and a key objective in the fight against climate change. Transitioning to cleaner modes of transport, such as electric buses, is essential for reducing greenhouse gas emissions and improving air quality in urban areas. Cuenca is not immune to these environmental challenges, and this project directly addresses the selection of EB for the city’s sustainability [13]. An efficient and sustainable public transport system directly impacts the quality of life of residents. Reducing traffic congestion and air pollution promotes healthier and more livable urban environments. The selection of electric buses can be fundamental in achieving this goal, giving significant relevance to this research in terms of the well-being of the Cuenca community [14]. Moreover, the implementation of EB has economic implications, as improvements in electric vehicle technology can offer long-term savings in fuel and maintenance compared to conventional buses [15].

MCDM Applied to Electric Transport Selection

To improve the decision-making process, multicriteria techniques are applied to identify the most suitable electric bus, thereby ensuring an efficient investment. Multicriteria decision-making methods (MCDM) are classified into three groups: (a) value-based methods, (b) preference-level methods, and c) ranking or possible overqualification methods [16]. The initial models are used to determine a preference order among various alternatives, with each assigned a numerical value. Weights are distributed to attributes based on the relevance of each criterion in the decision-making process. An example of a method within this group is the Analytic Hierarchy Process (AHP) [16]. The second type models enables the determination of an optimal solution according to the importance level of the established criteria and includes the TOPSIS and VIKOR methods [17]. The last type of model compares alternatives to determine which is preferred according to the established criteria and includes the ÉLECTRE and PROMETHEE methods [18].

In Guwahati, a method was proposed to efficiently design the kinematic chain of urban electric buses through a multicriteria decision-making process. This method classifies and selects traction motors based on the number of motors, energy sources, and type of motor drive, using the PROMETHEE method for the final decision [19]. In Beijing, the selection of EB is carried out using the PROMETHEE method to meet the population’s needs in transport services. Key criteria emphasized include battery performance, the passenger capacity of each bus, and the route they cover [20]. In Medellín, 64 standard electric buses and one articulated electric bus from the BYD brand were incorporated into the Metroplús system under the direction of Empresas Públicas de Medellín, who also developed the necessary charging infrastructure. For this decision, the PROMETHEE multicriteria method was used [21]. In a study in Beijing on alternative fuel vehicles, options such as electric, hydrogen, and methanol were evaluated. Using the Analytic Hierarchy Process (AHP), experts weighted criteria. Subsequently, multicriteria methods such as TOPSIS and VIKOR were used to compare and determine the best fuel mode [22]. In a case study in Malaysia, a hybrid approach was implemented for the selection and classification of EB in metropolitan environments, evaluating six alternatives. The process included data collection, criterion evaluation, weight assignment through AHP, and classification through TOPSIS [23]. In Taiwan, replacing diesel buses with electric vehicles is proposed, facing challenges such as total cost, batteries, dispatch schedules, and recharging. To address these issues, a simulation model has been suggested along with the AHP method and a multicriteria approach, aimed at resolving this complex situation in the appropriate selection of electric buses [24]. In Nanjing, China, the Nanjing Bus Company is at the forefront of implementing EB through dedicated policies and infrastructure. They have developed a robust on-road testing method using multicriteria techniques such as AHP to optimize the selection of electric vehicles [25]. In Turkey, electric buses were evaluated for their capacity to improve urban air quality. The diversity and technological advancement of EB complicate decision-making. The TOPSIS technique and the Moora method were used in a study, comparing five electric bus models based on six criteria [26]. The studies mentioned above utilize multicriteria techniques. For more details, refer to

Table 1. Studies with multicriteria methods.

No.	Objective	Year	MCDM	Ref.
1	The PROMETHEE method is employed for the design of the kinematic chain of EB through a multicriteria decision-making process.	2018	PROMETHEE	[19]
2	A fleet of EB is implemented in Beijing using the PROMETHEE method to meet the population’s needs.	2016	PROMETHEE	[20]
3	Systems for the implementation of EB in Medellín are evaluated and compared using the PROMETHEE method, seeking the best solution for public transport in the city.	2018	PROMETHEE	[21]
4	In Beijing, the TOPSIS and VIKOR methods were applied to compare and determine the best alternative fuel mode, including electric.	2016	TOPSIS–VIKOR	[22]
5	AHP–TOPSIS was applied to evaluate solutions for EB and promote electric urban transport.	2018	AHP–TOPSIS	[23]
6	The AHP method was used for the replacement of diesel buses with EB.	2022	AHP	[24]
7	The TOPSIS technique and the Moora method were used to classify alternatives for the introduction of EB.	2021	TOPSIS–Moora	[26]
8	A multicriteria methodology is employed to prioritize bus routes for the transition to EB in the city of Cuenca, Ecuador.	2021	Proposed method	[27]
9	The PROMETHEE method is used to prioritize the bus routes that should migrate first to electric units in the city of Cuenca, Ecuador.	2023	PROMETHEE	[5]

, which provides a comprehensive summary of the discussed information.

3. Methodology

3.1. Identification of Existing Buses

Given the continuous advancements in the availability of EB, certain specific models are emphasized, as detailed in the following table.

These buses are certified for public transport in Ecuador [28]. According to an analysis conducted in [29], there are several notable candidates, which are detailed in

Table 2. Homologated candidate electric bus models.

Model	Origin
Golden Dragon/XML 6125	Xiamen, China
Zhongtong/LCK6122EVG5	Shandong, China
BYD/K9G-l	Shenzhen, China
BYD/K11A	Shenzhen, China
Skywell/NJL6129BEV	Shahid Lashgari Blvd, Irán

.

3.2. Selection of the Multicriteria Method

The techniques for multicriteria decision-making allow for decisions to be made among several options or multiple alternatives (a1, a2, …, an). The alternatives are evaluated based on a set of attributes g1(·), g2(·), …, gk(·) that can be qualitative or quantitative. The best alternatives are chosen after making comparisons involving the selected attributes. Multicriteria techniques have the following steps in common: (i) problem definition, (ii) definition of the appropriate method, (iii) identification of alternatives, (iv) selection of criteria, (v) development of the decision matrix, (vi) assignment of weights, (vii) prioritization of alternatives, and (viii) decision-making.

Based on Table 1, it was verified that the most commonly used multicriteria methods are PROMETHEE and TOPSIS. The PROMETHEE methodology stands out over the other two methodologies in its application to transportation-related issues. The analysis suggests that, among the classification methods, PROMETHEE excels, being more suitable for problems involving multiple alternatives and various types of criteria. This advantage is particularly evident in situations where the alternatives exhibit significant differences in their criteria. TOPSIS, as a multi-attribute decision analysis method, is associated with value-based methods, producing a final ranking of alternatives based on their proximity to ideal and non-ideal solutions without explicitly validating the appropriateness of these rankings. Its strengths include the ability to adjust to different scales and handle both qualitative and quantitative data. However, it may present challenges such as a lack of consistency verification between the evaluated criteria and the judgment applied during the evaluation.

3.3. Description of the PROMETHEE Method

In [30], it is explained that the PROMETHEE method generates preference functions using the numerical discrepancy between pairs of alternatives, reflecting the perceived priority differences of the decision-maker. These functions assign values on a scale from 0 to 1, where a higher value indicates a greater preference discrepancy. A value of 0 indicates that there is no discernible preference difference between the alternatives, while a value of 1 indicates a clear preference of one alternative over the other. Six variants of preference functions are derived from different criteria: the true criterion, quasi-criterion, linear preference criterion, level criterion, linear preference criterion with indifference area, and Gaussian criterion. According to ref. [31], different generalized criteria show various levels of importance and strength in preferences. It is highlighted that the most prevalent criterion among users is the linear preference with an indifference area, followed by the Gaussian criterion in practical applications. In both criteria, the intensity of preference varies gradually from 0 to 1, while, in the third criterion mentioned, the changes in the intensity of preference are more abrupt. In this study, the criterion that incorporates linear preference and an indifference area will be employed, and is defined as follows:

$$P_j\left(d_j\right)=\left\{\begin{array}{cc}0& d_j\le s_j\\ {\displaystyle \frac{\left|d_j\right|-s_j}{r_j-s_j}}& s_j<d_j<r_j\\ 1& d_j>r_j\end{array}\right\}$$

(1)

where $d_j=y_{ij}-y_{kj}$ denotes the preference difference between a pair of alternatives in criterion $g_s,r_{j}y{s}_j$ for the preference and indifference thresholds, respectively.

The hypothesis to be investigated is based on understanding the preference and indifference thresholds, allowing the decision-maker to assign importance to certain alternatives. To clarify this concept, ref. [30] establishes an initial set of alternatives $A^T=a_1,a_2,\dots ak\subset A,(k<m)$. For each pair of alternatives $a_i,a_k$ belonging to $A^T$, one of the following two relationships holds: $a_i,S{a}_k,a_{k}S{a}_k$. Here, $a_{i}S{a}_k$ represents the preferred alternative a, $a_i$, as determined by the decision-maker. To infer the weights of the criteria, the following linear programming was constructed:

$$max\sum _{i=1}^k\sum _{j=i+1}^k{e}_{ij}$$

(2)

subject to $s.t$ $S(A_i,A_j-e_{ij})\ge S(A_i,A_j)$ for all pairs of alternatives in $A^T$, satisfying the relation of $A_{i}S{A}_j$.

Then,

$$e_{ij}\ge 0$$

(3)

In the linear programming above, there are k (k − 1)/2 non-negative decision variables. The objective function is a linear function over the unknown weights. This process can be carried out using the Visual PROMETHEE software. For the calculation of the aggregated preference index, which represents the degree of preference that one alternative will surpass another, Equation (4) is used:

$$\forall a,b\in A,\pi (a,b)\sum _{j=1}^k{P}_j(a,b)w_j$$

(4)

where $wj$ is the weight of each criterion, and $\pi (a,b)$ is associated with the degree of preference that alternative a will surpass alternative b.

• When $\pi (a,b)=0$, the global preference of a over b is weak.

• When $\pi (a,b)=1$, the global preference of a over b is strong.

To calculate the positive and negative outranking flows, there are two types. Equation (5) determines the positive outranking flow, which represents the alternative that will surpass the others. In other words, the alternative with the highest value will be the optimal one.

$${\varphi }^{+}\left(a\right)={\displaystyle \frac{1}{n-1}}\sum _{x\in a}\pi (a,x)$$

(5)

Then, using Equation (6), the negative outranking flow is calculated, which represents an alternative that is surpassed by the others. In other words, the best alternative will be the one with the lowest negative flow.

$${\varphi }^{-}\left(a\right)={\displaystyle \frac{1}{n-1}}\sum _{x\in a}\pi (a,x)$$

(6)

After determining the positive and negative outranking flows, the net preference flow is calculated. The alternative with the highest net flow will be identified as the best option, allowing for the establishment of a descending ranking of alternatives. Equation (7) expresses the discrepancy between the positive and negative flows, resulting in the net preference flow:

$$\varphi \left(a\right)={\varphi }^{+}\left(a\right)-{\varphi }^{-}\left(a\right)$$

(7)

3.4. Description of the TOPSIS Method

The TOPSIS method, developed by Hwang and Yoon in 1981, introduces the concept of an “ideal alternative”, defined as the solution with the shortest distance to the positive ideal solution (PIS) and the longest distance to the negative ideal solution (NIS) [32].

A fundamental aspect for the application of this technique is the assignment of weights to the criteria considered in the selection of a particular product. These weights are fundamental as they allow the relative importance of each criterion to be defined in the final evaluation.

Before starting,

Table 3. Decision matrix.

	$W_1$	$W_2$	⋯	$W_{n-1}$	$W_n$
	$C_1$	$C_2$	⋯	$C_{n-1}$	$C_n$
$A_1$	$X_{11}$	$X_{12}$	⋯	$X_{1,n-1}$	$X_{1,n}$
$A_2$	$X_{21}$	$X_{22}$	⋯	$X_{2,n-1}$	$X_{2,n}$
⋮	⋮	⋮	⋱	⋮	⋮
$A_{m-1}$	$X_{m-1,1}$	$X_{m-1,2}$	⋯	$X_{m-1,n-1}$	$X_{m-1,n}$
$A_m$	$X_{m,1}$	$X_{m,2}$	⋯	$X_{m,n-1}$	$X_{m,n}$

was constructed with the data obtained from each criterion and their respective weights.

According to ref. [32], the TOPSIS method addresses a problem in five stages: First, the performances of the alternatives in terms of various criteria are collected. Then, in the second step, these performances are normalized. The third stage involves weighting the normalized scores, followed by the calculation of the distances to the ideal and anti-ideal solutions in the fourth step. Finally, in the fifth step, the closeness of each alternative is determined based on these calculated distances. These five steps will be explained in more detail below.

Step 1: Normalization of the performances of the different criteria is started to facilitate comparison between different units. There are several methods available for this normalization, such as the following:

• Distributive normalization: Requires that the performances be divided by the square root of the sum of each element squared in a column [33].

  $$r_{ia}={\displaystyle \frac{x_{ia}}{\sqrt{{\sum }_{a=1}^n{x}_{ia}^2}}}$$

  (8)
• Ideal normalization: Requires dividing each performance by the maximum value in each column for criteria to be maximized, and by the minimum value in each column for criteria to be minimized [33].
  - Maximization:

  $$r_{ia}={\displaystyle \frac{x_{ia}}{u_a^{+}}}\mathrm{For}a=1, \dots , n \mathrm{and} i=1, \dots , m$$

  (9)

  where $u_a^{+}=max\left(x_{ia}\right)$ for all a = 1, …, n.
  - Minimization:

  $$r_{ia}={\displaystyle \frac{x_{ia}}{u_a^{-}}}\mathrm{For}a=1, \dots , n \mathrm{and} i=1, \dots , m$$

  (10)

  where $u_a^{-}=min\left(x_{ia}\right)$ for all a = 1, …, n.

Step 2: Now the weights are considered in the normalized decision. The weighted normalized values $v_{ai}$ of the matrix V are calculated by multiplying the normalized values $r_{ai}$ by their corresponding weight ${\omega }_n$. Here, ${\omega }_n$ is the weight of the n-th criterion, and the sum of all weights $\sum {\omega }_n$ can be equal to 1 in the general case [33].

$$v_{ai}={\omega }_n\ast r_{ai}$$

(11)

Step 3: The weighted scores will be used to compare each action with a virtual ideal action and with a virtual anti-ideal or negative ideal action. There are several methods for performing this calculation:

(1) The optimal performance and the most deficient performance of each criterion present in the normalized decision matrix are collected. To determine the ideal action, the following is considered:

$$A^{+}=(v_1^{+},\dots ,v_m^{+})$$

(12)

And, for the negative ideal,

$$A^{-}=(v_1^{-},\dots ,v_m^{-})$$

(13)

is considered, where $v_i^{+}=max\left(v_{ai}\right)$ if criterion i is desirable and is to be maximized or $v_i^{+}=min\left(v_{ai}\right)$ if, on the contrary, the criterion is not desirable and is to be minimized, as in the case of cost. And vice versa with the negative ideal [34].

This is the method most commonly used.

(2) Assuming the existence of an absolute ideal point and an absolute anti-ideal point, which are defined independently of the actions considered in the decision problem, $A^{+}=(1, \dots , 1)$ y $A^{-}=(0,\dots ,0)$.

(3) The ideal and anti-ideal points are defined by the decision-maker. These points must be located between the ideal and anti-ideal points obtained by the other two methods previously explained. This method is not frequently used, as it requires user input, which is usually difficult to obtain.

Step 4: In this case, the respective distances to the positive ideal point are calculated.

$$S_i^{+}=\sqrt{\sum _i{\left(v_i^{+}-v_{ai}\right)}^2},a=1, \dots , m.$$

(14)

And, in parallel, for the negative ideal,

$$S_i^{-}=\sqrt{\sum i{\left(v{i}^{-}-v_{ai}\right)}^2},a=1, \dots , m.$$

(15)

Step 5: To finish the process, the relative distance, or closeness coefficient, of each criterion is calculated:

$$P={\displaystyle \frac{S_i^{-}}{S_i^{-}+S_i^{+}}}$$

(16)

The closeness coefficient, which varies between 0 and 1, indicates the proximity of an alternative to the ideal. A value close to 1 suggests that the alternative is almost ideal, while a value close to 0 indicates proximity to the anti-ideal. Therefore, an alternative with a coefficient closer to 1 has a higher priority [33]. Based on this coefficient, the alternatives are ordered in descending order to identify the best option [34].

3.5. Definition of Criteria

To determine the appropriate type of electric bus for implementation in the city of Cuenca, several aspects need to be defined. The literature review indicates the need to analyze economic factors, such as acquisition and maintenance costs, and compare them with the technical characteristics of the units. Additionally, the importance of social factors as an essential component of the infrastructure is emphasized, making them significant criteria. Similarly, it is crucial to consider the environmental factors that could be impacted by the project. Therefore, the criteria to be considered for the selection of the electric bus are (1) economic factors, (2) technical factors, (3) social factors, and (4) environmental factors.

After reviewing the studies conducted using multicriteria methods, a summary table should be developed indicating the most frequently used subcriteria in these studies.

Table 4. Bibliographic review of subcriteria.

Criteria	Subcriteria	Citation
Economic	Investment cost (C11)	[24,35,36,37,38,39]
	Operation and maintenance cost (12)	[20,36,37,40]
	Amortization period of the investment (13)	[36]
Technical	Autonomy (21)	[24,36,37,39,41]
	Motor power (22)	[20,41]
	Passenger capacity (23)	[20,24,37,38,39,41,42]
	Maximum speed (24)	[37,38,41,42,43]
Social	Access for people with disabilities (31)	[24,38,44]
	Acceptance level (32)	[20,37]
	Passenger comfort (33)	[20,37]
Environmental	Visual impact (41)	[39,45]
	Auditory impact (42)	[37,46]
	Gas emissions (43)	[37,38,39,47]

summarizes the subcriteria most commonly used in these.

By refining Table 4, the most frequently mentioned subcriteria in the selected articles were identified. The subcriteria analyzed are not necessarily the only ones; however, they are assumed to contribute to the discussion in this field. In fact, the definition of these subcriteria is a contribution to knowledge, as they were identified after a literature review. Moreover, their importance was assessed through consultation with local experts. One of the limitations of the study is that the relationships between the subcriteria were not evaluated. Identifying interdependencies among them could strengthen the results.

Table 5. Subcriteria for the electric bus selection problem.

Criteria	Subcriteria		Unit
Economic	C11	Investment cost	USD
	C12	Operation and maintenance cost	USD/year
Technical	C21	Autonomy	km
	C22	Motor power	kW
	C23	Passenger capacity	n
	C24	Maximum speed	km/h
Social	C31	Access for people with disabilities	Yes/No
	C32	Acceptance level	n
Environmental	C41	Visual impact	n
	C42	Gas emissions	kg CO2 eq/kWh

presents the relevant subcriteria for the electric bus selection problem. The relevant subcriteria for the selection of electric buses include the investment cost (C11), which represents the initial expenditure required to acquire the vehicles. Additionally, the operation and maintenance cost (C12) is considered, which encompasses the daily costs related to operation and maintenance, directly impacting the profitability of the project. Autonomy (C21) refers to the maximum distance that the buses can travel on a full charge, crucial for ensuring operational efficiency and compliance with scheduled routes. The power (C22) of the vehicle determines its propulsion capacity, influencing acceleration and the ability to climb slopes. Passenger capacity (C23) indicates the maximum number of people that can be transported safely and comfortably. Maximum speed (C24), adjusted for urban and suburban environments, is influenced by factors such as engine power and vehicle design. In the social realm, access for people with disabilities (C31) is evaluated, which involves the installation of appropriate infrastructure to ensure mobility and comfort for people with disabilities. The acceptance level (C32) is also considered, which analyzes the willingness of users and stakeholders to adopt this technology compared to conventional buses. In environmental terms, the visual impact (C41) of EB in the urban environment is evaluated, considering the vehicle design and its landscape integration. Finally, gas emissions (C42) are evaluated to determine the contribution of electric buses to reducing atmospheric pollution and the carbon footprint, using indicators such as the carbon footprint to quantify total greenhouse gas emissions during battery charging.

3.6. Definition of Weights for Each Subcriterion

In the reviewed literature, it is highlighted that applying equal weights to all subcriteria, known as the equal weights (EW) approach, represents the most basic and straightforward method. However, since criteria and subcriteria can influence candidate alternatives in different ways, it is essential to assign different weights to reflect their relative importance adequately [48]. Three different methods are used to perform this weight assignment.

In the first scenario, the most basic method is employed where identical weights are assigned to all subcriteria. Equation (17) defines the respective weight value, where k is the number of subcriteria:

$${\omega }_j={\displaystyle \frac{1}{k}}$$

(17)

In the second scenario, the direct rating method (DRM) is used. Ratings are obtained through a questionnaire using a Likert scale administered to a group of experts. Subsequently, these ratings are normalized by dividing each value by the total sum of all values. In this method, the importance of each criterion is indicated on a scale of 1 to m. The weight of each of the k subcriteria ${\omega }_j$ is established through Equation (18):

$$\omega j={\displaystyle \frac{\sum {i=1}^m(P_{j,i}\ast R_{j,i})}{{\sum }_{j=1}^k{\sum }_{i=1}^m(P_{j,i}\ast R_{j,i})}}$$

(18)

where

• ${\omega }_j$ is the weight of subcriterion j;
• k is the number of subcriteria;
• $P_{j,i}$ is the number of points on the Likert scale i assigned by participants for each subcriterion j;
• $R_{j,i}$ is the fraction of the sum of each score (Pi) of the sum of all scores for each criterion.

In the third scenario, the weights are derived using the ordinal ranking (OR) method.

The ordinal rating method requires decision-makers to rank the criteria according to their importance. Then, the expected value method is applied [3]. With k criteria ordered in ascending order, the expected values, which represent the weights, are determined through Equation (19) [49]:

$${\omega }_1={\displaystyle \frac{1}{k^2}}$$

(19)

$${\omega }_2={\displaystyle \frac{1}{k^2}}+{\displaystyle \frac{1}{k·(k-1)}}$$

$${\omega }_{k-1}={\displaystyle \frac{1}{k^2}}+{\displaystyle \frac{1}{k·(k-1)}}+\cdots +{\displaystyle \frac{1}{k·2}}$$

$${\omega }_k={\displaystyle \frac{1}{k^2}}+{\displaystyle \frac{1}{k·(k-1)}}+\cdots +{\displaystyle \frac{1}{k·2}}+{\displaystyle \frac{1}{k·1}}$$

4. Results

4.1. Determination of Subcriterion Values

To obtain the results, various data collection techniques were employed, covering multiple subcriteria [37,39]. The prices of EB homologated in Ecuador were provided by companies in the sector, which allowed for an accurate assessment of investment costs [40,50]. For the analysis of operation and service costs, several urban public transport companies in Cuenca, Ecuador, were consulted [51]. Although there are currently no electric buses operating in the city, differences in security and registration costs compared to existing diesel buses were considered [36,52]. Technical data on autonomy per charge (km) were obtained from the data sheets and technical specifications provided by the electric bus manufacturers [53]. Similarly, information on bus load capacity was collected, emphasizing the importance of seated passenger capacity and the potential to increase it by approximately 89%.

The maximum speed of EB, another significant subcriterion, was evaluated considering that, although high speed is not required for urban service, a moderate level is necessary to ensure efficient transportation within the metropolitan area [53].

The level of social acceptance was assessed through a consultation with professionals specialized in electric mobility regarding the proposed electric buses. It was considered inappropriate to extend the survey to the general public due to their potential lack of knowledge about the technology. The survey was organized using a measurement scale ranging from 1 to 5 (where 1 symbolizes the lowest level of acceptance and 5 the highest level of acceptance). A total of 13 professionals, from both private and public entities, participated in the survey.

Since the candidate buses will be located in public spaces, a visual impact analysis was conducted to determine the degree of impact on the city. To measure this, surveys were carried out among eight professionals who work in the Technical Planning Department of the Municipality of Cuenca and are familiar with the local urban regulations. In the surveys conducted, a scale of 1 to 5 was established to determine the level of impact. Regarding environmental aspects, emissions of gases from each brand of electric bus were analyzed.

The data were obtained from reliable sources, including companies specializing in EB and professionals with in-depth knowledge of the subject. These data are fundamental for the evaluation and selection of appropriate electric buses for public transport in Cuenca.

In

Table 6. Description of criteria and subcriteria.

Models	Economic		Technical				Social		Environmental
	C11	C12	C21	C22	C23	C24	C31	C32	C41	C42
	(USD)	(USD/Year)	km	kW	n	km/h		n	n	(kg CO2 eq/kWh)
Golden XML 6125	350,000	34,514.00	300	150	40	85	1	0.24	0.13	127.51
Zhongtong	475,500	41,259.00	250	196	42	69	1	0.21	0.235	164.00
BYD/K9G-l	419,500	39,485.00	250	299.77	32	80	1	0.19	0.30	132.84
BYD/ K11A	505,500	39,501.54	250	359.41	37	60	1	0.17	0.21	179.58
Skywell	360,000	38,841.00	400	423.936	45	125	0	0.18	0.15	123.33

, the values of each alternative for each subcriterion are detailed using the specific units of measure for each one (see Table 5). For the application of the method, the necessary parameters were entered subsequently in Visual PROMETHEE and, in the case of TOPSIS, in Excel.

4.2. Weights

To apply the direct rating method, a survey was conducted among 13 participants with expertise in the subject matter. Several references indicate that adequate results are obtained with 10 to 20 participants [4,5,27]. Professionals in the city of Cuenca were identified whose academic and professional profiles align with the requirements of the research. The profiles included managers and educators in transportation and mobility, as well as managers, university professors, researchers, and maintenance technicians. Leaders of public transportation companies and mobility management in government institutions were also included. In these surveys, participants were asked to rate criteria on a scale of 1 to 4 (if the criterion was divided into four subcriteria) or 1 to 2 (if the criterion was divided into three subcriteria), with 1 being less important and 4 or 2, respectively, being more important. After tabulating the results (which were normalized and are presented in

Table 7. Weights for each scenario according to the calculation method.

Subcriterion	Weight for Cenario 1	Weight for Scenario 2	Weight for Scenario 3
C11	0.1	0.0802	0.0645
C12	0.1	0.0802	0.0479
C21	0.1	0.1374	0.1929
C22	0.1	0.1145	0.1429
C23	0.1	0.1641	0.2929
C24	0.1	0.1145	0.1095
C31	0.1	0.0840	0.0845
C32	0.1	0.0725	0.0100
C41	0.1	0.0802	0.0336
C42	0.1	0.0725	0.0211

), Table 7 shows the weights derived according to Equation (17) for the values of the second column, associated with the first scenario. The values of the third column were determined using Equation (18) for the second scenario, while those of the fourth column were calculated by applying Equation (19) in the third scenario.

In Scenario 1, all subcriteria have an equal weight of 0.1. This implies that all subcriteria contribute equally to the decision-making process or the analysis in question. There is no particular emphasis on any specific subcriterion. In Scenario 2, the respondents assign the highest weights to subcriteria C23 (passenger capacity) and C21 (range) (0.1641 and 0.1374, respectively). Meanwhile, subcriteria C32 (acceptance level) and C42 (gas emissions) have the lowest weights (0.0725). In Scenario 3, subcriterion C23 has the highest weight (0.2929). Subcriteria C21 and C24 also receive significant weights (0.1929 and 0.1095, respectively). Subcriteria C32, C41, and C42 have the lowest weights, indicating that they are less relevant in this scenario. The decrease in the weight of C41 is particularly notable, as it goes from being the most important in Scenario 2 to being one of the least important in Scenario 3. That is, technical criteria are the main concern when considering the choice of buses, while criteria related to transportation and social acceptability are less significant. The emphasis on technical criteria may be related to the need to improve the efficiency and capacity of public transportation rather than necessarily being driven by environmental or social factors.

4.3. Analysis of the PROMETHEE Method

The analysis of positive, negative, and net flows was carried out using Visual PROMETHEE, evaluating the positive flows (Phi+), negative flows (Phi−), and the net flow (Phi) for each alternative. These values result from pairwise comparisons through the PROMETHEE method and allow for a preliminary classification of the alternatives in descending order.

Visual PROMETHEE allowed comparison of the flows of each candidate bus in different scenarios. In all scenarios, the Skywell/NJL6129BEV stood out as the most favorable alternative. In the first scenario, the Skywell and Golden Dragon models presented flow values close to each other. In the second and third scenarios, the Skywell/NJL6129BEV significantly surpassed the other candidates. The Skywell/NJL6129BEV model is considered the optimal electric bus for implementation in the city of Cuenca based on the consistent results across all evaluated scenarios.

Table 8. Comparison of the final ranking for the three proposed scenarios: PROMETHEE.

Model	Rank Scenario 1	Rank Scenario 2	Rank Scenario 3
Golden Dragon/XML 6125	2	2	2
Zhongtong/LCK6122EVG5	4	3	3
BYD/K9G-l	3	4	5
BYD/K11A	5	5	4
Skywell/NJL6129BEV	1	1	1

compares the rankings obtained in the three scenarios using the PROMETHEE methodology. The first column shows the models of the electric bus alternatives, while the other three columns identify the position of each alternative in the ranking corresponding to each scenario.

4.4. Analysis of the TOPSIS Method

The analysis using the TOPSIS method was carried out by calculating the relative distance or closeness coefficient of each criterion, which allowed the different alternatives to be ordered from best to worst. The results are presented in

Table 9. Comparison of the final ranking for the three proposed scenarios: TOPSIS.

Model	Rank Scenario 1	Rank Scenario 2	Rank Scenario 3
Golden Dragon/XML 6125	1	2	2
Zhongtong/LCK6122EVG5	5	5	5
BYD/K9G-l	2	3	3
BYD/ K11A	4	4	4
Skywell/NJL6129BEV	3	1	1

, and each ranking obtained in the three proposed scenarios in this project was compared using the TOPSIS methodology.

Is clear that the results obtained for the same decision-making problem can vary from one method to another. This does not imply that there are incorrect solutions, but rather that each method operates differently.

5. Comparison of the Methods

A comparative analysis between the PROMETHEE and TOPIS methods is essential in multicriteria decision-making. Both methods evaluate and rank alternatives considering multiple criteria, but they differ in their principles and algorithmic approaches. PROMETHEE, based on the theory of preferences, compares alternatives in a pairwise manner, evaluating all possible differences in criteria. This method generates positive and negative preference flows for each alternative, thereby facilitating the creation of a ranking. On the other hand, TOPSIS is based on the idea that the best alternative should have the smallest distance to the positive ideal solution and the largest distance to the negative ideal solution. This method involves calculating positive and negative ideal solutions, determining the distance of each alternative to these solutions, and ranking them based on their relative proximity. When comparing these approaches, PROMETHEE stands out for its accessibility and ease of implementation, available thanks to the availability of specialized software. In contrast, TOPSIS requires the use of tools such as Excel for its development. PROMETHEE offers a more detailed interpretation due to its analysis of preference flows and shows greater adaptability for managing various types of criteria and accommodating the decision-maker’s preferences. Additionally, it is usually more computationally efficient, which is particularly useful in situations with a large number of alternatives and criteria. After applying both methods in various scenarios, the results showed that, in Scenario 1, PROMETHEE favored the Skywell/NJL6129BEV model, while TOPSIS highlighted the Golden Dragon/XML6125 model, evidencing discrepancies in the results; in Scenario 2, PROMETHEE and TOPSIS agreed on selecting the Skywell/NJL6129BEV model as the best; in Scenario 3, both methods concluded that the optimal model is the Skywell/NJL6129BEV. In most of the analyzed scenarios, the Golden Dragon/XML6125 model occupied the second position according to both methods, suggesting it as an alternative to the Skywell/NJL6129BEV model.

6. Discussion

The research conducted to select the optimal electric bus for the city of Cuenca employed multicriteria methods, specifically PROMETHEE and TOPSIS, to evaluate several alternatives [32]. Each method offered different perspectives and allowed for a comparison of the available electric bus models [28].

In the analysis using the PROMETHEE method, the scenarios evaluated showed variability in the results depending on the weights assigned to each criterion. In the first scenario, the proximity in the flow values between the Skywell and Golden Dragon models indicates close competitiveness. However, in the second and third scenarios, the Skywell/NJL6129BEV model consistently positioned itself as the best alternative, surpassing the other candidates in terms of net flow.

The TOPSIS method showed similar results with certain variations in the ranking of the alternatives [21]. In the first scenario, the Golden Dragon/XML 6125 bus led the ranking, but, in the second and third scenarios, the Skywell/NJL6129BEV model once again emerged as the superior option. This repeated behavior across different methods and scenarios underscores the robustness of the Skywell model in meeting the evaluated criteria.

The criteria considered encompassed economic, technical, social, and environmental factors. The inclusion of these criteria ensured a comprehensive evaluation where not only costs and technical efficiency were considered, but also the social and environmental impact [36]. This approach is crucial for the implementation of sustainable technologies in public transportation.

The selected model was chosen as the optimal choice primarily due to its low investment cost, which is approximately 15% lower compared to the average of the other models. This factor is particularly crucial for intermediate-sized cities seeking a sustainable energy transition that takes into account investment costs [5]. Additionally, the selected model’s high autonomy, which exceeds the average of the other brands by 38%, and the power of the motor, which is 200% relative to the average of the other brands, are ideal for a city like Cuenca, characterized by irregular terrain with variable slopes and long bus route distances [5,27]. Furthermore, the passenger capacity of the selected model is 14% higher compared to the average of the other brands, which would promote comfort in the use of public transportation and prevent overcrowding [4]. However, specifications could be subject to change due to improvements in energy efficiency, advancements in battery technology, or changes in operating conditions.

In refs. [5,27], the most suitable routes for the initial implementation of EB in the city of Cuenca are analyzed. Meanwhile, in [4], the optimal location of charging stations is studied. The results of this study complement the findings of this previous research and, in the future, could be integrated into a broader investigation that evaluates the practical application of these results.

7. Conclusions

The research identifies the optimal electric bus model for the public transportation system in Cuenca. The Skywell/NJL6129BEV, Golden Dragon/XML6125, BYD/K9G-l, BYD/K11A, and Zhongtong/LCK6122EVG5 models are recommended as viable alternatives, all of which are approved for use in Ecuador. The technical criteria evaluated ensure optimal performance, considering range, power, passenger capacity, maximum speed, and charging time, which are essential for efficient operation in urban environments. From an economic perspective, the costs of acquisition, operation, and maintenance were evaluated, highlighting the efficiency of the recommended models. Socially, community acceptance and accessibility for people with reduced mobility were considered. After defining criteria and weights through objective and subjective methods, the PROMETHEE and TOPSIS analyses consistently highlighted the Skywell/NJL6129BEV model as the best option for Cuenca. This model demonstrates superiority in range, energy efficiency, and maintenance costs, adapting to the needs of urban public transportation. Economically, it represents an efficient long-term investment with low operating costs and a long service life. Socially, it is well received by the community, improving air quality and the user experience. Environmentally, it significantly reduces the carbon footprint and promotes clean technologies. Successful implementation requires continuous monitoring and supportive policies, such as tax incentives and educational programs. In summary, the adoption of the Skywell/NJL6129BEV model leads the transition towards more sustainable public transportation in Cuenca, with potential to serve as a regional model in Latin America.

The main challenge in transitioning to EB in Cuenca is selecting the most suitable model that meets operational requirements and sustainability goals while considering economic, technical, social, and environmental criteria. Ecuadorian regulations commit to transitioning to electric mobility in public transport. However, the lack of knowledge about the technology means that choices might be made based solely on economic factors. This study integrates a multicriteria analysis adapted to Cuenca’s specific context. The validation of results can be achieved by incorporating different models into the city’s routes, although this is costly as it requires technological availability and objective evaluation. The proposal considers Cuenca-specific data and offers an adaptable and updated solution for selecting the most suitable model for the city’s public transport.

Although the scope of the study is limited to Cuenca, the results can serve as a basis for future research that expands the study to other cities and conducts comparative analyses in different geographical and socioeconomic contexts. This expansion would help validate the methodology and increase the overall impact of the findings, allowing for the transfer of knowledge and experiences to other localities with similar characteristics.