Clean and zero-emission urban buses: Compliance with EU regulations and fleet transition in Seville

Clean and zero-emission urban buses: Compliance with EU regulations and fleet transition in Seville M.A. Tagua Navarrete a,* , J. Serrano Reyes a , J.A. V´ elez Godi˜ no b , F.J. Jim´ enez-Espadafor Aguilar a a Department of Energy Engineering, School of Engineering, University of Seville, Camino de los Descubrimientos S/N, 41092, Seville, Spain b Department of Construction Engineering and Engineering Projects, School of Engineering, University of Seville, Camino de los Descubrimientos S/N, 41092, Seville, Spain ARTICLE INFO Handling editor: X Ou Keywords: EU regulations CO 2 emissions Battery electric buses CNG buses Energy consumption of urban buses Fleet transition ABSTRACT The EU public transport sector is dominated by diesel and natural gas buses, but fleet structures are rapidly changing. EU climate laws drive two strategies: one for bus manufacturers and another for public transport operators (PTO), with different timelines. PTOs must progressively replace fleets with clean and zero-emission buses. This paper develops bus replacement strategies aligned with EU standards, focusing on cost minimization and CO 2 emission reduction. Three propulsion technologies (diesel, compressed natural gas, and electric) and two bus types (regular and articulated) are analyzed. The strategies consider acquisition, operational, and service costs, along with CO 2 emissions. The proposed approach, applied over a 10-year period, uses real-word data from Seville (Spain) PTO to meet EU standards for 2025–2035. Operational data allow precise energy demand estimation, considering propulsion, climatization air compressor, and auxiliaries. CO 2 emissions calculations show that CNG buses emit 24 % more CO 2 per km than diesel buses. Optimization results indicate that a CO 2 minimization strategy reduces emissions by 16.3 %, with only a 6.7 % cost increase compared to the minimal-cost strategy. 1. Introduction The transition toward sustainable urban transport is a global priority driven by environmental concerns, emissions regulations, and clean energy advances. In 2022, transport accounted for 23 % of global CO 2 emissions [1], and in the EU, it contributed 29,7 %, with road transport responsible for about 60 % of that –over 25 % from heavy-duty vehicles and buses [2]. With 70 % of its population living in urban areas [3], the EU promotes sustainable, low-emission urban mobility through policies like the Urban Mobility Framework [4], emphasizing public transport, clean, safe and connected alternatives. Similarly, sectors such as Information and Communication (ICT) are increasingly engaging in efforts to reduce CO 2 emissions, in alignment with policy targets established in EU [5] and echoed in other regions, including Arab countries, where studies have highlighted the need for strategic action [6,7]. From regulatory standpoint, the EU promotes sustainable transport through the European Green Deal, a strategic plan to achieve climate neutrality by 2050. This initiative was formalized in Communication COM (2019) 640 [8], which outlines key policies to reduce emissions, foster clean energy adoption, and support sustainable economic growth. EU is promoting the use of alternative fuels with the Regulation EU 2023/1804 [9] was approved in September 2023, in line with the objectives of the European Green Deal. Together with this Regulation, the Directive 2019/1161 [10] date from June 20, 2019, which promotes clean and energy-efficient road transport vehicles, includes an update to the definition of alternative fuels. This Directive are transposed into Spanish Law. The Directive is motivated by the EU commitments to reduce greenhouse gas emissions by at least 40 % by 2030 compared to 1990 levels, to increase the consumption of renewable energy by at least 27 %, and to achieve energy savings of no less than 27 %. Within this Directive, several vehicle categories are defined in accordance with Regulation (EU) 2018/858 [11]. Among them, category M 3 corresponds to urban buses. This classification specifies vehicle types based on their fuel or energy source, including. * Corresponding author. E-mail address: [email protected] (M.A. Tagua Navarrete). Contents lists available at ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy https://doi.org/10.1016/j.energy.2025.137025 Received 19 March 2025; Received in revised form 5 June 2025; Accepted 6 June 2025 Energy 332 (2025) 137025 Available online 13 June 2025 0360-5442/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). •Clean vehicles: Those that use alternative fuels or energy sources that replace, at least in part, traditional fossil fuels and that can contribute to decarbonization, including: Electricity, hydrogen, ammonia, synthetic paraffinic fuels (HVO), biofuels, natural gas in gaseous form (CNG) and liquefied form (LNG), and liquefied petroleum gas (LPG). In the case of liquid, synthetic, or paraffinic biofuels, mixing with conventional fossil fuels is not allowed. Excluded from the above are fuels produced from raw materials with a high risk of indirect land-use change, for which there is a significant expansion of the production area on lands with high carbon stock. •Zero-emission vehicles: Clean vehicles without an internal combustion engine (ICE), or with an ICE that emits than 1 g of CO 2 /kWh (measured in accordance with Regulation (EC) No 595/2009) or emits less than 1 g of CO 2 /km (measured in accordance with Regulation (EC) No 715/2007). With current technology, zero-emission vehicles are typically defined as those powered by batteries or hydrogen fuel cells (HFC). However, several manufacturers — including Volvo, WestPort, MAN, and IVECO — have already developed internal combustion engines that operate exclusively on hydrogen and are expected to receive approval by 2025 [12,13]. As a result, the technological landscape for zero-emission vehicles is poised significantly in the near future. In addition to the previously mentioned definition of alternative fuels used in vehicles within the target category, Directive (EU) 2019/ 1161 sets minimum procurement targets that each EU Member State must meet regarding the public purchase of M 3 category vehicles for public transport. For Spain, the targets established by this Directive are presented in Table 1. It is worth highlighting that the previous standard applies exclusively to the procurement of new buses, not to the existing fleet. Furthermore, by December 31, 2027, the European Commission (EC) is required to review the implementation of Directive 2009/33/EC and propose new legislation applicable beyond 2030, from which additional regulatory constraints are anticipated. Consequently, fleet renewal strategies should be designed with sufficient flexibility to accommodate these potential legal modifications. Regulations targeting bus manufacturers must also be taken into account. A key example is the Regulation of the European Parliament and of the Council of May 24, 2024, which reinforces CO 2 emissions standards for new heavy-duty vehicles and establishes mandatory reporting obligations [14]. According to this regulation, bus manufacturers are required to comply with the implementation schedule presented in Table 2. Regulatory frameworks significantly influence urban transport operations, particularly through PTO which play a central role in the EU decarbonization efforts by aligning fleet renewal with legal, technological, and environmental targets. The transition to clean or zeroemission vehicles mandated by legislation poses intricate economic and operational challenges. Current regulations require PTOs to phase out non-compliant ICE vehicles during new bus procurement, driven by aging fleets and regulatory pressure to reduce greenhouse gas emissions. Starting in 2035, EU-based manufacturers must supply zero-emission buses to PTO, marking a pivotal shift as outlined in Table 2. Traditionally powered by diesel or compressed natural gas (CNG), PTO fleets increasingly consider electrification, citing battery electric buses (BEB) as cleaner and more energy-efficient alternatives [2,15]. This evolution necessitates strategic frameworks for decision-making that encompass economic viability, environmental impact assessments, and operational continuity. The shift from diesel and CNG to electric and other low-emission technologies, including HFC, demands careful optimization in timing, scale, and long-term financial planning. Addressing these complexities requires holistic approaches that balance economic feasibility, environmental sustainability, and service reliability. This challenge is formally referred to as the bus fleet replacement problem (BFRP), which involves the systematic and cost-effective substitution of conventional buses with cleaner alternatives over a defined planning horizon. The BFRP is shaped by various interdependent factors, including high upfront investment costs, operational savings potential, emissions reduction goals, and the infrastructure required to support new propulsion technologies —particularly electric and hydrogen-based systems. Moreover, solving the BFRP requires maintaining a delicate balance between service reliability and environmental objectives, all while minimizing disruptions to PTO. Several studies have investigated the BFRP, significantly broadening its scope beyond traditional cost minimization to encompass the challenges of decarbonization and technological transition. Optimization models – frequently employing Mixed-Integer Programming – and the vehicle characterization, fleet capacity and information and period of analysis are central to this research. These models typically aim to minimize total Life Cycle Cost (LCC), as in the work of Frieβ & Pferschy [16], who analyzed a zero-emission mixed fleet and demonstrated the advantages of a technology-diverse composition. This work presents similarities with [16] in the sense that it includes the exploitation cost of the actual bus fleet, which also exhibits technological diversity. However, it is oriented toward the end user—the PTO—and in this regard, it addresses both cost issues and compliance with mandatory EU regulations. To the authors’ knowledge, there is currently no research work that addresses both aspects simultaneously. Islam & Lownes [17], for instance, developed a dual-objective approach addressing both LCC and GHG emissions, identifying optimal solutions based on the integration of BEB and Hybrid Electric Buses (HEB), our work similarly addresses dual objectives (cost and CO 2 minimization) but as separate optimization goals, focusing on a different technological mix. Under uncertainty, Avenali et al. [18] applied real options to determine optimal replacement timing and technology adoption, reporting up to 10 % Total Cost of Ownership (TCO) savings when fleet renewal is anticipated; in our current work, although employing a distinct optimization framework, TCO considerations are expanded to include a more granular cost breakdown, such as specific operational expenses (maintenance, personnel) and the salvaged values for vehicles and BEB batteries. Furthermore, Stasko & Gao [19] explored the interplay between acquisition, retrofitting, and task assignment decisions, showing that CNG buses can be profitable under appropriate CO 2 pricing schemes; our approach differs by directly modelling EU regulatory targets for fleet composition over a 10-year horizon, analysing the replacement of diesel, CNG and BEB buses based on real-world operational data, rather than focusing on CO 2 pricing schemes, retrofitting, or task assignment. Other studies – such as Ribeiro & Mendes [20] – focused on policy-driven fleet renewal criteria, including vehicle age thresholds; while our study is also policy-driven, it applies specific EU procurement targets (percentages of Table 1 Percentages required of public companies for the purchase of heavy vehicles for public transport. Bus (Category M3) From August 2, 2021, to December 31, 2025 From January 1, 2026, to December 31, 2030 Clean vehicles 45 % 65 % Zero-emission vehicles At least 50 % of the clean vehicles’ percentage Table 2 Vehicle fleet by type in percentages for EU bus manufacturers. Bus (Category M3) Before 2030 From January 1, 2030, to December 31, 2034 From January 1, 2030, to December 31, 2039 As from 2040 Zeroemission vehicles 0 % 90 % 100 % 100 % M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 2 Table 3 BFRP literature review. Author (Reference) Vehicle type Objectives Policy Gov Conclusions Othman et al. [22] BEB Develop tools (replacement factors, prediction models) to estimate required BEB fleet size for equivalent service level. No Replacement factor depends on heating system & season (higher in winter). ML models predict fleet size accurately based on distance, temp, battery size. Frieβ & Pferschy [16] BEB, HFCB Minimize LCC of mixed-fleet ZEB system by optimizing technology mix, infrastructure, and schedules. Context Mixed fleets financially outperform singletech solutions. Optimal mix sensitive to operational constraints & planning assumptions. Enayati et al. [23] DB, BEB Find equilibrium strategies for PTS, manufacturer & passenger choice using game theory under different gov. policies Context Subsidies for electric buses and taxes on diesel ones are the most effective government measures to promote fleet electrification, while marketbased tools like Cap & Trade are less effective. Public support is driven by environmental concern. Stasko & Gao [19] DB, CNGB, BEB, HEB Minimize long-term operational costs + emission penalties by integrating purchase, retrofit & task assignment decisions. No Long-horizon planning superior to short-horizon. CNG can be costeffective. Carbon price influences tech choice. Ribeiro & Mendes [20] DB, DHB, CNGB Develop & evaluate a national (Portugal) bus replacement scenario based on age criteria for full decarbonization by 2034. Context Full fleet replacement with ZEVs by 2034 is feasible based on age criteria. Requires significant investment but greatly reduces average fleet age. Bakker et al. [21] Electric and Diesel Trucks Minimize total costs of fleet replacement (purchase, salvage, operation, infra.) over multi-year horizon. No ETs can reduce total costs but require high upfront CAPEX (trucks, chargers, grid). Depot charging key; public charging cost critical for long-haul ET viability. Tang et al. [24] BEB Select optimum EB type for single line replacement by analyzing trade-off between LCC and passenger waiting time. No Trade-off: Large EBs have lower LCC (operator view), small EBs have lower waiting time (passenger view). Type selection robust to Table 3 (continued) Author (Reference) Vehicle type Objectives Policy Gov Conclusions line length & low subsidies. Lu et al. [25] DB, HEB, BEB Evaluate lifecycle economic & environmental performance of alternative fuel buses in Europe, considering climate & energy mix. No Mixed fleets (Hybrid +Electric) can balance cost & environment better than full electric in some European contexts. Seasonal operation adjustments beneficial. Islam & Lownes [17] DB, HEB, BEB Minimize LCC & GHG emissions simultaneously via optimized parallel fleet replacement schedule & mix. No Optimized replacement reduces LCC & GHG. Optimal mix depends on constraints (e.g., 79 % BEB/21 % HEB found costoptimal under relaxed constraints). Avenali et al. [18] DB, CNGB, HEB, BEB Evaluate fleet replacement decisions under uncertainty using Real Options to find optimal timing & technology, maximizing cost savings. No RO shows anticipating replacement (vs. end-of-life) saves costs (up to 10 % TCO) by leveraging flexibility under uncertainty. BEBs become dominant over time. Short contracts hinder optimal investment without guarantees. This work DB, CNGB, BEB Minimize CO 2 emissions and overall cost by applying a BFRP strategy to each of the problems in a PTO in Seville, taking into account not only economic, technological, and logistical constraints, but also EU legislation applicable to PTO and urban buses Yes The methodology enables cost or CO 2 focused fleet optimization using real Seville data. Despite BEB high purchase cost, labor factors reduce TCO differences. Emissions drop by 16.3 % with only a 6.7 % cost gap. Accurate energy modeling and infrastructure planning are crucial for effective, regulationcompliant fleet renewal in urban settings. Vehicle type DB: Diesel bus; CNGB; CNG bus; HEB; Hybrid electric bus; DHB: Diesel hybrid bus; HFCB: Hydrogen Fuel Cell bus; BEB: Battery Electric Bus Policy of government Yes: Government policies are explicitly integrated into the optimization model as formal constraints, directly influencing the feasible solution space Context: Policy considerations are acknowledged and inform the overall problem setting, but are not explicitly as constraints within the optimization framework No: Government policies are neither considered in the formulation referenced in the context of the optimization problem M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 3 clean and zero-emission vehicles) directly to a PTO in Seville, rather than focusing on age-based criteria for a broader national decarbonization scenario. Bakker et al. [21] incorporated a detailed analysis of infrastructure-related decisions, such as the availability of depot and public charging stations, into cost minimizations frameworks that include purchase, salvage, and operational costs. In this study adopts a similar cost structure; however, it is uniquely grounded in real operational data from a Seville PTO. This provides actual local data, distinguishing our work from studies that might rely on referenced information from other sources to inform their models. A comprehensive comparison between prior research and the current study is delineated in Table 3, where key literature is synthesized by comparing objectives, vehicle types, policy integration (as formal constraints, contextual elements, or absent), and findings from each study. While these contributions have greatly advanced the field, the transition to low-or zero-emission fleets still presents major challenges for PTO [20], especially under scenarios requiring full fleet electrification. These challenges stem from dynamic market conditions, technology uncertainties, and increasingly stringent EU directive promoting sustainable urban mobility. Moreover, although various cost dimensions have been studied in isolation, few models simultaneously account for regulatory compliance, real-world operations, and the joint evaluation of environmental and economic performance. While informed by studies such as Ribeiro & Mendes [20] (focusing on CO 2 and cost minimization, including capital, maintenance, and energy expenditure) and Bakker et al. [21] (incorporating purchase, salvage, and operational costs), the present research offers a distinct contribution. We uniquely ground our analysis in a local context using real operational data from a PTO in Seville, Spain. This study introduces an integrated optimization framework of the BFRP for the PTO, designed to fill identified research gaps. A key advancement of our model is its explicit incorporation of multiple fleet renewal strategies that are directly constrained by EU regulations, a level of regulatory integration not consistently found in existing approaches. The following sections outline the structure of the paper: Section 2 introduces the BFRP problem description and mathematical formulation. Section 3defines the costs associated with the bus fleet to integrate them into the optimization model. Section 4provides data on the specific case study obtained from real-world operational data. Section 5 presents the results of the optimization model, discussing them and providing a sensitivity analysis of various parameters to validate the study. The conclusions are presented in Section 6. 2. Problem description In this section, the BFRP is introduced and mathematically formulated. Fig. 1 shows the main structure of the optimization problem where six main cost elements are considered: purchase cost, salvage revenue (once the bus has reached salvage age, it is sold, making this term a source of income), energy costs (in terms of diesel, CNG, or electricity), maintenance costs, insurance costs, and driver costs. The problem has been formulated with two distinct objectives to be analyzed. One is to minimize the discounted sum of all operating costs, and the other is to reduce the total amount of carbon dioxide emissions as much as possible. Both objectives are addressed independently over the planning horizon and are subject to technical, company, and standard constraints. In this regard, the most relevant aspects are compliance with EU standards and budget limitations, assuming that all input parameters are known with certainty. The methodology presented in this work lies in the use of real, operational data rather than relying on generic assumptions or literature-based estimates. Unlike many studies that base their cost analysis on theoretical models or external sources, this study derives the costs of all agents involved in the current operation of the PTO from actual operational data collected over a full calendar year. In addition, critical cost-related information was directly provided by the PTO, ensuring that the analysis reflects the specific context and realities of the fleet’s daily operations. A general description of the problem is provided, followed by its formulation as an integer linear program. Fig. 1. Main structure of the optimization BFRP. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 4 Within the scope of the technologies available, three different elements can be purchased: buses, batteries, and battery chargers. For buses, a range of different bus types has been considered K ={1, …, m}, where K E , K G , and K D are three different subsets of bus types that are, respectively, powered by electricity (batteries), CNG, and diesel fuel. Additionally, for each propulsion technology, there are two types of buses: regular and articulated, so m =6. Hydrogen buses powered by FC have not been considered as an alternative propulsion technology for buses. This is because EU legislation [10]can be fulfilled with battery electric buses, the cost of hydrogen buses is almost double that of electric ones [26,27], and the bus parking site lacks infrastructure for a hydrogen fueling station, which would incur additional costs. For batteries, two different types are considered, directly related to their size: the low-capacity battery is used for the regular electric bus, while the high-capacity battery is used for the articulated bus; in this regard let B ={1, …, rr}, where B L and B H are the two subsets of batteries of low and high capacity, respectively, so rr =2. For the battery chargers we consider a range of charger types G ={1, …, n}, where G L and G H are the two subsets of charger of medium and high capacity, so n =2. A planning horizon T ={1, …, t F } consisting of t F periods is considered, with two target periods, t G1 and t G2 , included in T at which at least P G1 and P G2 proportions should be fulfilled, according to the schedule in Table 1. Clearly, t G1 <t F and t G2 <t F , but all periods t >t G1 and t >t G2 in the planning horizon should meet the targets P G1 and P G2 , respectively. Regarding bus task assignment, the routes do not have any special characteristics, which allows adopting the same modeling strategy as in Ref. [28]. This is due to two reasons: the first one is a consequence of the PTO schedule, in which any bus type — whether regular or articulated and powered by batteries, diesel fuel, or CNG — can follow any route. Both types of fuels used for bus propulsion, diesel and CNG, produce CO 2 locally and since there are no restrictions in Seville regarding this issue, any bus type can operate on any route. The second reason is connected to the city’s orography; Seville is a completely flat city, and therefore, there are no specific routes that present different energy demands (from the perspective of the energy required for bus propulsion or climatization) to be considered. If Seville had areas with steeper routes compared to flatter ones, each specific route should be considered, or at least grouped, to account for the impact on energy consumption, as clearly explained in section 4. From the point of view of available energy, the PTO organizes the buses so that every bus type starts its route every day with full capacity. This means that gas tanks are completely filled with CNG at 200 bar, the diesel tank is full, and the batteries of regular or articulated buses are fully charged. Due to the nature of the city of Seville, the bottle capacity of the CNG buses, the tank capacity of the diesel buses, and the charging infrastructure for the battery electric buses allow every bus, irrespective of the subset of bus types within K, see Table 4, to complete its full day’s run (around 14 h per day) without the need to recharge batteries or refill the tanks or bottles. Therefore, it is not necessary to express periodic demands in a more granular manner as in Refs. [29,30], because there are no compatibility considerations to be taken into account between the bus type and the operation they will perform. The age of each bus and battery must be considered in the model because every technology has its own end-of-life. This is not the case for battery chargers, which are assumed to have an infinite lifespan. Let the set J k ={0, 1, …, r k +1} be defined to contain all possible ages for a bus of type k, with j ∈J k representing an age of j periods. The age j = 0 corresponds to a new bus and the age j =r k +1 is the point at which the bus must be salvaged. At the start of the planning horizon, the fleet consists of a j k buses of type k and age j. For batteries, let the set Y B ={0, 1, …, z B +1} be defined to contain all possible ages for a battery of type B, with y ∈Y B representing an age of y periods. The age y =0 corresponds to a new battery and the age y = z B +1 is the point at which the battery must be salvaged. At the start of Table 4 Parameter and variable definitions of the BFRP model. Sets BSet of battery types B H Set of batteries of high size B L Set of batteries of low size GSet of battery chargers G H Set of battery chargers of high capacity G L Set of battery chargers of low capacity J K Set of possible ages for a bus of type K KSet of bus types K DA Set of diesel articulated buses K DS Set of diesel regular buses K EA Set of electric articulated buses K ES Set of electric regular buses K GA Set of CNG articulated buses K GS Set of CNG regular buses TSet of t F period in the planning horizon Y S Set of possible ages for a battery of type B Parameters a j k Number of buses of type k and age j in the fleet at the start of the planning horizon CMinimum age required for any bus purchased to be salvaged c y b Number of batteries of type b and age y at the start of the planning horizon d yt b Salvage value of a battery of type b and age y retired at the start of period t E DS Emission factor Kg CO 2 per km of regular diesel bus E DA Emission factor Kg CO 2 per km of articulated diesel bus E GS Emission factor Kg CO 2 per km of regular CNG bus E GA Emission factor Kg CO 2 per km of articulated CNG bus ΔtMaximum allowable time to achieve a complete charge of the batteries f t k Purchase cost of bus type k at the start of period t h g Number of chargers of type g owned at the start of the planning horizon H t Budget for purchases of buses, batteries and battery chargers at the start of period t ir Annual inflation rate KMM Annual distance [km] traveled by each bus per year L H Electric load capacity of high-capacity chargers L L Electric load capacity of low-capacity chargers M H Capacity of large-sized batteries M L Capacity of small-sized batteries N A Capacity of articulated bus (maximum number of passengers) N S Capacity of regular bus (maximum number of passengers) NN t Capacity of the whole fleet at the start of period t o jt k Total periodic operation cost when a bus of type k is of age j at the start of period t P G1 Minimal proportion of the fleet to be purchased of clean and zero emission vehicles from target period G1 P G2 Minimal proportion of the fleet to be purchased of clean and zero emission vehicles from target period G2 pr Annual discount rate q t g Purchase cost of battery charger type g at the start of period t r k +1Age at which buses of type k must be salvaged R 1t Minimum ratio between regular and articulated buses at the start of period t R 2t Maximum ratio between regular and articulated buses at the start of period t s jt k Salvage value of a bus of type k and age j retired at the start of period t t G1 Target period G1 for the application of current EU standard [10] t G2 Target period G2 for the application of current EU standard [10] w t b Purchase cost of battery type b at the start of period t z b +1Age at which battery of type b must be salvaged Decision variables u t g Number of chargers of type g purchased at the start of period t ua t g Number of available chargers of type g at the start of period t x t k Number of buses of type k purchased at the beginning of period t xs jt k Number of buses of type k and age j salvaged at the beginning of period t xa jt k Number of available buses of type k and age j at the beginning of period t v t b Number of batteries of type b purchased at the beginning of period t vs yt b Number of batteries of type b and age y salvaged at the beginning of period t va yt b Number of available batteries of type b and age y at the beginning of period t M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 5 the planning horizon, there are c y B batteries of type B and age y. Although age is not considered for battery chargers, at the start of the planning horizon there are h G chargers of type G. Considering cost issues, o j,t K represents the total periodic operating cost when a bus of type K is at age j at the start of period t. These operating costs include expenses for energy, maintenance, personnel and insurance and are specific to each propulsion technology. Let f t K be the purchase cost of bus type K at the start of period t, considering that only new buses are purchased, although considering used buses or leasing could be a viable alternative. Only K E and K G buses are here considered in order to reduce both local and global pollution [31]. The salvage value of a bus of type K and age j retired at the start of period t is s j,t K . Let w t B be the purchase cost of battery type B at the start of period t, considering that only new batteries are purchased. In the case of battery electric buses, the cost of the bus does not include the cost of the battery, which must be considered separately. This modeling decision has been made because the lifespan of the battery is significantly shorter than that of the electric bus. The salvage value of a battery of type B and age y retired at the start of period t is d yt B ; since batteries can only be salvaged at the end of their useful life, y =z B +1. For battery charger, let q t G represent the purchase cost of battery charger type G at the start of period t. Since battery chargers are considered to have an infinite lifespan, no salvage value is considered. It is assumed that the parking area for battery electric buses has sufficient space for the installation of battery chargers, and therefore there are no restrictions in this regard. Additionally, because the electric demand is relatively low compared to the total installed electric power at the bus parking site, there are no extra costs for electricity consumption beyond the actual energy used. A budget H t is available at the start of period t for purchase of buses, batteries and battery chargers, which can be increased by salvage revenues earned at the start of t. However, the salvage value of buses and batteries is very low compared to the acquisition cost, 15 % of purchase price [32]. Finally, let pr be the periodic discount rate used to consider the time-value of money, being 1/(1 +pr) the corresponding one-period discount factor. All cash flows are assumed to occur at the beginning of each period; therefore, an amount incurred in period t will be discounted by 1/(1 +pr) t−1 . Finally, a constant annual inflation rate ir has been incorporated into the model for the entire analysis period. 2.1. Integer linear programming model The optimization model is formulated as a deterministic heterogeneous fleet replacement model, where all input parameters are known with certainty. The decision variables are defined in Table 4; all are integers, non-negative, and represent the number of each component purchased (batteries, chargers, buses) of different types, the number of salvaged components, and the number of available components. Two independent objective functions are considered, with the model’s constraints remaining the same for both. One of the objective functions minimizes the total discounted cost over the planning period, including purchase costs, salvage revenues, and operating costs. The purchase cost varies for each period for buses, batteries, and chargers over the analysis period. The objective function is defined by equation (1), while the complete set of equations that comprise this model includes equations (1), (3) and (4), …, (29). Similarly, the objective function for minimizing the total CO 2 emissions over the analysis period is defined by equation (2), with the corresponding model consisting of equations (2)–(4), …, (29), where it is assumed that all buses cover the same distance (KMM) each year. It is worth mentioning that the electricity consumed by the PTO is certified green, meaning that 100 % of the energy used comes from renewable sources. As a result, the electricity used to charge the batteries is entirely carbon-free, meaning there are no CO 2 emissions from the battery electric buses. Objective function for minimizing the total exploitation cost minimize ∑ t∈T(1+ir 1+pr)t−1 ∑ k∈K(fk txk t−∑ rk+1 j=1 sk jt xsk jt +∑ rk j=0 ok jtxak jt)+ +∑ t∈T(1+ir 1+pr)t−1 ∑ b∈B(wb tvb t−∑ rk+1 j=1 db yt vsb yt)+∑ t∈T(1+ir 1+pr)t−1 ∑ g∈G qg tug t(1) Objective function for minimizing the total CO 2 emissions minimize (∑ k∈KDS xak jt EDS +∑ k∈KDA xak jt EDA ∑ k∈KGS xak jt EGS +∑ k∈KGA xak jt EGA )KMM t∈T,j∈Jk\{rk+1}(2) Constrains ∑ k∈KG,KE xk t≥PG1∑ k∈K xk tt∈T,∀t≥tG1(3) ∑ k∈KG,KE xk t≥PG2∑ k∈K xk tt∈T,∀t≥tG2(4) ∑ k∈KE xk t≥PG1∑ k∈K xk tt∈T,∀t≥tG1(5) ∑ k∈KE xk t≥PG2∑ k∈K xk tt∈T,∀t≥tG2(6) xk t=xak jt j=0,k∈K,t∈T\{1}(7) vb t=vab yt y=0,b∈B,t∈T\{1}(8) xk t+ak j=xak jt,j=0,k∈K,t=1 (9) vb t+cb y=vab yt,y=0,b∈B,t=1 (10) ug t+hg=uag tt=1,g∈G(11) uag t=uag t−1+ug tt∈T\{1},g∈G(12) xak jt =xak j−1,t−1−xsk jt k∈K,t∈T\{1},j∈Jk\{0,rk+1}(13) vab yt =vab y−1,t−1−vsb yt b∈B,t∈T\{1},y∈Yb\{0,zb+1}(14) xsk rk+1,t=xak rk,t−1k∈K,t∈T\{1},(15) vsb zb+1,t=vab zb,t−1b∈B,t∈T\{1},(16) xak jt =ak j−xsk jt k∈K,t=1,j∈Jk\{0,rk+1}(17) vab yt =cb y−vsb yt b∈B,t=1,y∈Yb\{0,zb+1}(18) ∑ k∈K(fk txk t−∑ rk+1 j=1 sk jt xsk jt)+∑ b∈B(wb tvb t−∑ rk+1 j=1 db yt vsb yt)+∑ g∈G qg tug t≤Htt∈T (19) vab yt ≥xak jt t∈T,b∈BL,k∈KES,y∈Yb\{zb+1},j∈Jk\{rk+1}(20) vab yt ≥xak jt t∈T,b∈BH,k∈KEA,y∈Yb\{zb+1},j∈Jk\{rk+1}(21) ∑ k∈KES,KDS,KGS xak jt NS+∑ k∈KEA,KDA,KGA xak jt NA≥NNtt∈T,j∈Jk\{rk+1}(22) M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 6 ∑ k∈KES,KDS,KGS xak jt ≥R1t⋅∑ k∈KEA,KDA,KGA xak jt t∈T,j∈Jk\{rk+1}(23) ∑ k∈KES,KDS,KGS xak jt ≤R2t⋅∑ k∈KEA,KDA,KGA xak jt t∈T,j∈Jk\{rk+1}(24) xsk jt=0k∈K,t∈T,j∈Jk\{C+1,C+2,…,rk+1}(25) vsb yt=0b∈B,t∈T,y∈Yb\{zb+1}(26) Δt(∑ g∈GL uag tLL+∑ g∈GH uag tLH)≥∑ b∈BL vab yt ML +∑ b∈BH vab yt MHt∈T,b∈B,g∈G,y∈Yb\{zb+1}(27) uag t≥xak jt t∈T,g∈GL,k∈KES,j∈Jk\{rk+1}(28) uag t≥xak jt t∈T,g∈GH,k∈KEA,j∈Jk\{rk+1}(29) Constraints (3) to (6) express the restrictions imposed by the current EU directive [10], as given in Table 1. Although this standard permits the continuous purchase of diesel buses, a major goal of public companies is to reduce local contamination. Since replacing diesel buses does not achieve this goal, the purchase of new diesel buses is not allowed. Constraint (7) states that the number of available buses of each type with age 0 is equal to the number of buses of the same type purchased at the start of period t. Restriction (8) states that the number of available batteries of each type with age 0 is equal to the number of batteries of the same type purchased at the start of period t. Constraint (9) expresses that the number of available buses of age 0 in the first period of analysis is equal to the number of purchased buses plus the number of new buses of the same type acquired before the first analysis period. Restriction (10) expresses that the number of available batteries of age 0 in the first period of analysis is equal to the number of purchased batteries plus the number of new batteries of the same type acquired before the first analysis period. Restriction (11) states that the number of available charges in the first period of analysis is equal to the number of purchased chargers plus the number of new battery chargers of the same type acquired before the first analysis period. Restriction (12) states that the number of available chargers of any type for any year t >1 is equal to the number of chargers of the same type available in the previous year, plus the chargers of the same type purchased at the beginning of year t. Equation (13) states that the number of available buses of any type and age j at the beginning of year t is equal to the available buses of the same type from the previous year, one year younger, minus the buses of the same type and age salvaged at the beginning of year t. Equation (14) expresses that the number of available batteries of any type and age y at the beginning of year t is equal to the available batteries of the same type from the previous year, one year younger, minus the batteries of the same type and age salvaged at the beginning of year t. Constraint (15) states that the number of buses of type k that have reached the maximum available age at the beginning of year t is equal to the number of available buses that, in the previous year, were one year away from being salvaged. Constraint (16) states that the number of batteries of type b that have reached the maximum available age at the beginning of year t is equal to the number of available batteries that, in the previous year, were one year away from being salvaged. Equation (17) states that the number of available buses of any type and age j in the first year of analysis is equal to the available buses of the same type and age previously at disposal, minus the number of salvaged buses of the same type and age at the beginning of year t. Equation (18) expresses that the number of available batteries of any type and age y in the first year of analysis is equal to the available batteries of the same type and age previously at disposal, minus the number of salvaged batteries of the same type and age at the beginning of year t. Constraints (19) state that the amount of funds available for the purchase of buses, batteries, and chargers at the start of a period is the sum of the budget and salvage revenues generated at the start of that period. The model could accommodate different annual budget amounts if those figures were known with certainty; this is of great importance because, in the context of urban public companies, the budget can change due to political reasons. Restriction (20) establishes that the number of available batteries of any age y and type B L must be equal to or greater than the number of available regular battery electric buses of any age j at the beginning of the analysis year t. Restriction (21) establishes that the number of available batteries of any age y and type B H must be equal to or greater than the number of available articulated battery electric buses of any age j at the beginning of the analysis year t. This relationship between battery and bus type is due to the fact that it is not possible to interchange different battery capacities; that is, the batteries are not modular. Constraint (22) states that the sum of the capacity of regular and articulated buses of any type and age must be at least equal to the established capacity of the bus fleet in each analysis period. Inequality (23) states that the ratio of regular to articulated buses of any propulsion category and age must be at least greater than R 1t at the beginning of the analysis period. Inequality (24) states that the ratio of regular to articulated buses of any propulsion category and age must be less than or equal to R 2t at the beginning of the analysis period. Equality (25) states that any purchased bus of any type can only be salvaged after reaching a minimum age C, in accordance with Spanish requirements. Equality (26) states that any purchased battery of any type can only be salvaged after it has reached its allowable operational life. Constraint (27) establishes that the charging capacity of the battery chargers during a time interval Δt, typically less than 6 h, is sufficient to fully charge all the batteries, regardless of whether they are of low or high capacity. Restriction (28) states that at least is necessary one low capacity charger for each regular electric bus and (29) states that at least is necessary one high capacity charger for each articulated electric bus. 3. Useful life and costs 3.1. Propulsion battery cost and salvaged value According to Bloomberg NEF’s annual battery price survey [33], prices have fallen in 2023. The price evolution is the result of three opposing forces: the evolution in raw material and component prices, the growth in production capacity, and expected demand, all of which are constantly changing. Therefore, battery prices are expected to evolve in the near future with a negative slope of approximately 8 € /kWh per year starting with a constant price of 245 € /kWh in January 2024. Propulsion batteries have a maximum operational life of 10 years, as assured by the manufacturer [34]. Once this limit is reached, the batteries must be salvaged from use as propulsion energy storage. However, EU regulations concerning batteries require to be designed in a way that allows for easy recovery, reuse, and recycling [35]. In this regard, it is expected that, in the short term, there will be a secondary market for the use of electric bus batteries in other applications, and therefore a salvage value must be considered. The National Renewable Energy Laboratory (NREL) from the U.S. Department of Energy has studied the cost of second-life batteries in detail. According to their findings, a price for second-life batteries of around 15 % of the original purchasing cost can be expected, which has been included in the analysis [32]. 3.2. Bus cost and salvaged value Since diesel buses will not be purchased, only the prices for CNG and battery electric buses are considered. There are two types of CNG buses in the fleet: regular and articulated. The average capital cost of a new M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 7 unit is 375,000 € and 460,000 € , respectively. The current EU standard [14], which mandates that EU manufacturers reduce the supply of this type of propulsion technology to no more than 10 % by 2034 (see Table 2), will affect the prices of these buses. However, there are currently no projections regarding their price evolution, so a constant price is assumed over the planning horizon. For battery electric buses, future price changes will primarily come from battery costs, as there is little room to further reduce the price of other vehicle components. The capital cost of a regular electric bus is 515,000 € , and for an articulated bus, it is 700,000 € , both of which include the cost of the batteries. The prices come from Ref. [36] and have been updated form average values taken from a local survey of manufacturers. The batteries of the regular battery electric buses have a capacity of 470 kWh, while those of the articulated battery electric buses have a capacity of 720 kWh. Buses powered by hydrogen through FC or internal combustion engines have not been considered for three reasons. •The cost of these buses is roughly 35 % higher than the price of an electric bus [20]. •The cost of green hydrogen is approximately twice that of electricity [37]. •With battery electric buses, the EU standard [10] is fully met. The maximum useful life of diesel and CNG buses is fixed: 22 years for diesel buses and 19 years for CNG buses [38,39]. In the case of CNG buses, this limit is imposed by the useful life of the compressed natural gas tanks. The salvage value of both diesel and CNG buses is negligible because, after such a long operational period, the value is almost null. In this regard, Hensher’s residual price estimate of 15 % of the purchase cost for diesel/CNG buses [40] is not considered. Instead, a fixed salvage return value of € 2000 per bus has been assigned. For battery electric buses, the situation is different. Although the expected useful life is 20 years, the degradation of components (excluding the battery) is expected to be lower [31], resulting in a salvage value of € 50,000 for regular buses and € 60,000 for articulated ones. 3.3. Maintenance costs Maintenance costs refer to various bus components such as tires, tools, external workshops, spare parts, containers, batteries (non-propulsion), valves and brake wear. The maintenance costs for battery electric buses are expected to be lower than those for diesel or CNG buses, excluding battery replacement costs. For instance, battery electric buses have a regenerative braking system, which reduces brake wear and lowers energy consumption [41]. Additionally, the cooling system of diesel or CNG buses requires significantly more maintenance and spare parts compared to battery electric buses, which experience much lower thermal loads. The average maintenance cost is known and is based on a PTO survey of the entire current fleet, as shown in Table 5. However, maintenance data presented in technical publications differ significantly from current Table 5 Yearly average maintenance cost for the entire fleet of 425 buses of all types. Maintenance costs € /year for the whole fleet Tires 268,418 Batteries (not for electric propulsion) 62,367 Accessories/spare parts 2,817,248 Tools 111,029 Cleaning, Containers, recycle bins 142,880 External bus workshop 165,340 ∑3,567,282/425 = € 8394 per bus Table 6 Complete bus fleet composition as of January 1, 2025; the age of each bus and battery type are included in the first column. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 8 proven data. For example [41], reports a maintenance cost of 0.35 € /km for diesel buses and 0.18 € /km for battery electric buses (excluding the propulsion battery). For the PTO of Seville, this translates to an annual maintenance cost of € 15,750 per diesel bus (similarly for CNG buses) and € 8100 per electric bus, assuming an average of 45,000 km/year per bus [42]. With battery electric buses accounting for 7.8 % of the entire fleet, as shown in Table 6, and the average maintenance cost of the entire fleet being € 8394 per bus per year, it is clear that the estimated maintenance costs for diesel and CNG buses are excessive. A similar overestimation of maintenance costs for diesel and CNG buses is also found in Ref. [17]. Nevertheless, the ratio of maintenance costs between electric and diesel buses in Refs. [17,41] is in the range of 0.35–0.55. In this regard, this work considers the same global maintenance costs presented in Table 5 but assumes a ratio of 0.45 between electric and diesel or CNG buses, which results in a final cost of € 0.19/km for each diesel or CNG bus and € 0.087/km for each electric bus. 3.4. Insurance and driver costs Insurance costs are the same for every bus, regardless of the propulsion system or size (regular or articulated), and are fixed at € 6000 per bus per year. The cost per driver is € 61,000 per year, which includes the total salary plus social security contributions paid by the company for each driver. With 1264 drivers for 425 buses, this implies an average ratio of 2.97 drivers per bus. This ratio will also apply to every new bus purchased. 3.5. Fuel and energy costs Diesel and CNG are the fuels used to power diesel and CNG buses, respectively. Although fuel prices can fluctuate, the PTO benefits from a flat rate that falls within the scope of the subsidies it receives. Diesel is priced at € 1.20 per liter, while CNG is priced at € 57 per MWh. The cost of electricity for charging bus batteries is also a flat rate, set at € 0.14 per kWh. In any case, reference [43] provides an EU benchmark for electricity prices applicable to vehicle charging. 3.6. Battery charger cost and charging schedule Two types of battery chargers are used: fast chargers with a power of 100 kW and rapid chargers with a power of 150 kW, with corresponding costs of € 60,000 and € 80,000, respectively. A fast charger can fully charge a 300 kWh battery (for regular battery electric buses) in less than 4 h, while a rapid charger takes around 4 h to fully charge a 450 kWh battery (for articulated battery electric buses). Therefore, with the appropriate number of chargers, all of the battery electric buses can be charged in no more than 4 h, which is within the time that the buses are parked. Moreover, it has always been considered that a regular electric bus requires at least a fast charger, while an articulated electric bus requires at least a rapid charger. 3.7. External costs The EU has provided a comprehensive handbook for evaluating external costs, which quantifies the cost per kilometer ( € /km) for each item [44], applicable to EU countries as well as a group of non-EU countries. This topic has also been discussed in Ref. [45], but the impact of including these costs in the analysis is negligible, as the focus is on minimizing the total fleet operating cost or total CO 2 emissions [45]. For this reason, external costs have not been included in this analysis. 3.8. Fleet composition Table 6 presents the complete bus fleet composition as of January 1, 2025, with the age of each bus and battery type included in the first column. These correspond to the set of ages J k and Y s from Table 4. The number of battery chargers available includes 10 low-power units and 23 high-power units. 4. Energy demand and CO 2 emissions Considering all energy consumers of the electric bus, the following relationship applies: ˙ We TOTAL =˙ We prop +˙ We clim +˙ We air comp +˙ We aux +˙ We reg (30) Where. • ˙ We TOTAL is the total electric power provided by the bus batteries. • ˙ We prop is the electric power demanded for propulsion. • ˙ We clim is the electric power demanded by the climatization system. • ˙ We air comp is electric power demanded by the bus air compressor. • ˙ We aux is the electric power demanded by auxiliary equipment. • ˙ We reg is the electric power due to the regenerative braking, always negative. The most significant energy consumption comes from propulsion, and the power required by a vehicle for this purpose [46] is defined by the following equation: ˙ Wprop η mech =v[(mbus +mpas)(g sin( α )+g frcos( α )+a)+0.5CdA ρ v2+Θw r2 w] (31) Where. •˙ Wprop propulsive power provided by the engine (electric or thermal). •m bus is the mass of the vehicle. •m pas is the mass of the passengers. •g is the gravitational acceleration. • α is the terrain slope. •C d is the vehicle’s aerodynamic drag coefficient. •A is the frontal area of the vehicle. • ρ is the air density (at the ambient temperature and pressure of the location). •f r is the rolling resistance coefficient. •v is the vehicle speed. •a is the vehicle’s acceleration. • η mech is the mechanical efficiency between the engine (diesel/CNG/ electric) and the propulsion wheels that it is estimated as a constant value of 0.95. •Θw is the moment of inertia of a wheel. •r2 w is the wheel radius. From the analysis of the above expression, it can be seen that the power demand is a function of the cube of the vehicle’s speed, and therefore, at medium and high speeds, the shape and frontal area of the vehicle strongly influence this power. However, for buses operating strictly in urban areas, the average speeds are very low, with a commercial speed [42] below 14 km/h. Consequently, the term of (31) related to the aerodynamic drag carries little weight compared to the other terms. The term in equation (31) that depends on the slope of the terrain is influenced by the city’s topography [47]. While there is no possible action on this term, it can affect the others—primarily the term (m bus +m pas )—since if the slope is steep, the power reserve available for acceleration decreases. The term f r mainly depends on the characteristics of the road surface and the tires and remains constant for a vehicle whose tires are kept in good condition and properly inflated. Additionally, the moment of inertia term is very small in comparison with the other parameters and can be neglected, as it represent only a minimal M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 9 would be required, which would reduce the validity of the model’s results. Moreover, the model is designed to guide immediate decisions based on currently available information and to ensure compliance with the EU standard. Additionally, the model can be re-run if there is a change in the EU standards during the planning period. Furthermore, emerging technologies, such as FC buses or hydrogen-powered internal combustion engines (H 2 ICE), can be integrated into the model as soon as they become commercially available and demonstrate clear cost advantages. Given the data, considerations, and parameters from the previous sections, the problem now becomes determining the decision variables listed in Table 4 according to the constraints provided by equation (3) through (29) for each of the 10 years of the analysis, in such a way that the overall cost is minimized given by equation (1). Fig. 5 shows the total operational cost and annual CO 2 emissions, including all the factors discussed in Sections 3 and 4, for each year of the optimal solution, assuming a constant maximum allowable budget H t of € 16.5 million. It is worth noting that no feasible solution exists for H t below € 16.5 million. This indicates that the EU places significant economic pressure on the PTO; if it had not purchased the battery electric buses listed in Table 6 prior to January 1, 2024, it may not be able to comply with the EU standard. It is worth noting that the optimization procedure defined under this strategy does not explicitly target CO 2 emissions reduction as an objective. Consequently, the decrease in CO 2 emissions observed in Fig. 5 is solely a result of compliance with the EU standard, rather than any deliberate effort to minimize environmental impact. By examining the evolution of global costs over the planning period, as shown in Fig. 5, it becomes evident that costs experience a drastic reduction starting in 2030, followed by an almost flat trend up to 2033, when regular electric and CNG buses are acquired as it can be appreciated in Fig. 6a. Looking at Fig. 7, which shows the purchases of batteries and chargers, due to the short useful life of batteries, a large number of them need to be purchased starting in 2032, although these costs are low compared to operational costs. Therefore, from 2031 onward, the primary driver of total cost evolution is operational expenditure. However, a notable cost increase occurs in 2033, due to the procurement of 15 CNG buses and 8 battery electric buses, all of which are regular units, as can be seen in Fig. 6a. With this strategy, by 2029, the percentage of clean vehicles of the fleet reaches 95 %, and the share of zero-emission vehicles exceeds 17 %. As a result of the increase in battery electric buses, the required electric power rises from 6.2 MW in 2025 to 12 MW by 2029, reaching a peak of 12.9 MW in 2033. This increase in power demand is relatively small and can be accommodated by the installed capacity at the PTO facilities. Since diesel buses are no longer permitted for new purchases, the number of both standard and articulated diesel units gradually decreases as they reach the end of their service life, as illustrated in Fig. 6a. The ratio of articulated to standard buses is maintained at 40 % each year, reflecting the operational limit currently set by the PTO. This constraint is incorporated into the model as inequalities (23) and (24) (see Section 2.1). Despite the higher unit cost of articulated buses in each category, this ratio remains the most cost-effective configuration due to the associated reductions in driver-related operational costs. The main reason for this result is the high driver-related costs, with a driver-to-bus ratio of 2.97, as discussed in Section 3. In this context, articulated salvaged diesel buses are replaced by articulated CNG buses to maintain cost-efficiency. Additionally, the number of battery electric buses purchased is kept at the minimum required to satisfy the constraints set by equations (3)–(6). Although battery electric buses offer lower maintenance and energy costs compared to CNG alternatives, increasing the share of zero-emission vehicles beyond the minimum mandated by the EU standard [10] is not a cost-effective strategy, primarily due to the higher capital investment required for electric vehicles, batteries, and charging infrastructure. Moreover, the purchase of new battery electric buses imposes an additional electric power load on the PTO, which evolves annually as shown in Fig. 8. In this figure, it is assumed that all the battery electric buses are available to charge their batteries simultaneously. If a different charging schedule were implemented to reduce charging simultaneity, the required electric power could be reduced. 5.2. Sensitivity analysis The purpose of the sensitivity analysis is to account for future variations in the model outputs due to unexpected changes in the model parameters. Given the current global economic and geopolitical situation, a reduction in the costs of batteries, battery electric buses, electricity, diesel, and CNG is expected in the medium term, though in an unpredictable manner and therefore it has not been included. Additionally, inflation is expected to decrease smoothly in the short term, but it cannot be predicted from the middle of the analysis period onwards. In this context, independent changes in the allowed ratio between articulated and regular buses are considered. Increasing the ratio range Fig. 9. Annual global cost and CO 2 emissions for the minimum CO 2 emissions strategy over the planning period. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 16 between articulated and regular buses to 0.3–0.5 reduces the global cost from 905.7 M € to 881.2 M € , a 2.7 % reduction. The number of articulated CNG buses rises to 158 units by 2029, compared to 128 units in the previous solution (see Fig. 5b). The total number of articulated buses increases to 201 due to the purchase of CNG articulated buses (there is no purchase of articulated battery electric buses), reducing the fleet at the end of the period from 415 to 401. Further increasing the ratio range to 0.3–0.6 (see R 1t -R 2t in Table 4) reduces the global cost to 863.1 M € , a 4.7 % reduction, with the number of articulated buses increasing to 233 units by 2030. In this scenario, the number of articulated battery electric buses reaches 60 units by 2029, while in the base solution there was no change in the number of units. This increase in articulated battery electric buses results in a reduction in CO 2 emissions, from 299.2 Mkg in the base case to 287.9 Mkg in the new ratio, a reduction of 3.8 %. However, the required electrical power for charging battery electric buses has increased from 13 to 14 MW, a 7.7 % increase. 5.3. Shifting the optimization focus: CO 2 minimization Based on the data, considerations, and parameters outlined in the previous sections, the task now is to determine the decision variables listed in Table 4, subject to the constraints defined by equation (3) through (29) for each of the 10 years of analysis, in a way that minimizes the overall CO 2 emissions as described by equation (2). Fig. 9 presents the results in terms of global cost and CO 2 emissions. Compared to Fig. 5, this solution shows a slightly higher global cost of € 967.5 million (an increase of 6.6 %) and CO 2 emissions of 254.0 Mkg, representing a significant reduction of 16.3 %. In Fig. 9, the cost decreases more progressively than in Fig. 5, while emissions experience a sharp reduction starting in 2029, driven by the purchase of regular battery electric buses beginning in 2030, as illustrated in Fig. 10. The cost of the required electric facilities is excluded from the analysis, as there are numerous feasible solutions for the charging schedule that could significantly reduce the power demand. Fig. 10. Optimum solution minimizing CO 2 emissions per each year of the planning period; a) number of purchased buses, b) number of available buses. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 17 It is surprising that the increase in cost is not proportional to the decrease in CO 2 emissions, although the cost of the batteries must also include the cost of the chargers. This is due to the fact that the overall cost of the buses is dominated by driver costs, as each bus requires nearly three drivers to provide full service. The increase in battery electric buses necessitates the installation of more battery chargers, and consequently, the maximum installed electrical power at the PTO must be increased from 13 MW (the optimum solution for minimum global cost) to 30 MW, more than twice, as indicated by the results in Fig. 11. In this figure, it is assumed that all battery electric buses are available to charge their batteries simultaneously. If a different charging schedule were implemented to reduce charging simultaneity, the required electric power could be reduced. 6. Conclusions The proposed methodology determines the number of buses of each type that need to be purchased to either minimize total operating costs or reduce CO 2 emissions while consistently meeting EU standards and PTO requirements, taking into account the existing fleet composition at the start of the analysis. It also calculates the required electrical installed power, providing valuable information for the PTO to adjust the power capacity in the parking area. The main strength of this work lies in the implementation of a methodology based on the use of detailed, realworld data from the current fleet, particularly through the measurement of all relevant electric energy consumption parameters in terms of power. In this regard, all operational data were obtained from the CANBUS, which provides the instantaneous electric power of the various energy-consuming components of the buses. The comparison between the two optimization strategies reveals a small cost increase—below 6.7 %—when CO 2 emissions are minimized, but a significant reduction in total CO 2 emissions of 16.3 %. This solution, therefore, makes a greater contribution to the decarbonization of the city of Seville. Nevertheless, to comply with EU standards, a minimum annual budget of € 16.5 million for purchasing buses, batteries and chargers is required, which places significant pressure on the PTO. Despite the significant cost difference between CNG and battery electric buses—approximately 40 % higher for standard buses and over 50 % for articulated buses—the overall cost difference between the two optimization strategies remains relatively small. This can be attributed to two key factors. •The high driver-to-bus ratio, approximately 2.9 •The substantial share of driver salaries in the total operating costs These two factors help mitigate the financial impact of acquiring more expensive battery electric buses. Furthermore, increasing the proportion of articulated buses in the fleet leads to lower overall costs, as fewer drivers are required to transport the same number of passengers—thus reducing the total labor cost. The methodology used to estimate specific energy consumption [kJ/ km] is particularly appropriate, as it allows the model to be customized to the specific conditions of the city of Seville, rather than relying on generalized average values from existing literature. Using standard figures from other studies could result in optimization outcomes that significantly deviate from reality. In this study, notable differences—approximately 20 %—were observed between the energy consumption measured directly from Seville’s fleet (for both standard and articulated buses) and the values commonly reported in the literature. These findings underscore the importance of using localized, real-world data to ensure the accuracy and reliability of the optimization process. Regarding specific CO 2 emissions [kgCO 2 /km], it is worth noting that this value is higher for CNG buses compared to diesel buses, despite the fact that the kgCO 2 /MJ of natural gas is lower than that of diesel fuel. This result is explained by the higher energy consumption of CNG buses, which is due to the technical design of CNG engines for power control. Factors such as a lower compression ratio compared to diesel engines, a fuel-air stoichiometric ratio close to one, and a high pumping loop at part load—where these buses typically operate—contribute to this increased consumption. As the demanded power decreases, power losses increase, leading to lower efficiency. It should also be noted that the average demanded power generally does not exceed 50 % of the engine’s maximum power. Future research should focus on incorporating the required installed electric power for battery charging into the optimization model. This aspect introduces two significant limitations. First, there is the increased installation cost associated with upgrading infrastructure to support higher power demands. Second, there is the physical space requirement needed to accommodate the necessary charging facilities. Ultimately, both factors contribute to a higher total cost in electric bus-based solutions and will therefore influence and constrain the outcomes of future optimization strategies. CRediT authorship contribution statement M.A. Tagua Navarrete: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Investigation, Formal analysis, Data curation, Conceptualization. J. Serrano Reyes: Writing – review & editing, Validation, Supervision. J.A. V´ elez Godi˜ no: Writing – review & editing, Validation, Formal analysis. F.J. Jim´ enez-Espadafor Aguilar: Writing – review & editing, Visualization, Validation, Supervision, Methodology, Formal analysis, Data curation, Conceptualization. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Miguel A. Tagua Navarrete reports administrative support was provided by University of Seville. Miguel A. Tagua Navarrete reports a relationship with University of Seville that includes: employment. Collaboration agreement between the University of Seville and Transportes Urbanos de Sevilla S.A.M. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We would like to express our sincere gratitude to TUSSAM for Fig. 11. Annual maximum electric power demand for the PTO in the minimum CO 2 emission scenario. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 18 granting us access to the data generated by their buses. This valuable information has been essential for our analysis and has significantly contributed to the development of this study. We truly appreciate their collaboration and support, which have been instrumental in advancing our research. Data availability The data that has been used is confidential. References [1] IEA, “Global CO2 emissions by sector, 2019-2022, IEA, Paris.” Accessed: April. 27, 2025. [Online]. Available: https://www.iea.org/data-and-statistics/charts/g lobal-co2-emissions-by-sector-2019-2022. [2] Doulgeris S, et al. Evaluation of energy consumption and electric range of battery electric busses for application to public transportation. Transport Eng Mar. 2024; 15:100223. https://doi.org/10.1016/j.treng.2023.100223. [3] European Commission. Urban and rural living in the EU [Online]. Available: http s://ec.europa.eu/eurostat/web/products-eurostat-news/-/edn-20200207-1. [Accessed 27 April 2025]. [4] European Union, “Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the regions the new EU urban mobility framework.” Accessed: April. 27, 2025. [Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/? uri=celex:52021DC0811. [5] European Commission. Communication from the Commission to the European Parliament, the Council, the European economic and social Committee and the Committee of the regions [Online]. Available: https://eur-lex.europa.eu/legal -content/EN/TXT/?uri=CELEX:52020DC0562. [Accessed 10 May 2025]. [6] Islam MS, Rahaman SH. The asymmetric effect of ICT on CO2 emissions in the context of an EKC framework in GCC countries: the role of energy consumption, energy intensity, trade, and financial development. Environ Sci Pollut Control Ser Jul. 2023;30(31):77729–41. https://doi.org/10.1007/s11356-023-27590-1. [7] Islam MS, Rahaman SH, ur Rehman A, Khan I. ICT’s impact on CO2 emissions in GCC region: the relevance of energy use and financial development. Energy Strategy Rev 2023;49(Sep). https://doi.org/10.1016/j.esr.2023.101147. [8] European Commission, “COM (2019) 640 Communication from the Commission to the European Parliament, the European Council, the Council, the European Economic and Social Committee and the Committee of the Regions the European Green Deal.” Accessed: April. 27, 2025. [Online]. Available: COM (2019) 640 Communication from the Commission to the European Parliament, the European Council, The CounciL, The European Economic and Social Committee and the Committee of the regions the European Green Deal.. [9] European Commission, “Regulation (EU) 2023/1804 of the European Parliament and of the Council of 13 September on the deployment of alternative fuels infrastructure, and repealing Directive 2014/94/EU.” Accessed: April. 27, 2025. [Online]. Available: Regulation (EU) 2023/1804 of the European Parliament and of the Council of 13 September on the deployment of alternative fuels infrastructure, and repealing Directive 2014/94/EU.. [10] European Commission. Directive (EU) 2019/1161 of the European Parliament and of the council of 20 June 2019 amending Directive 2009/33/EC on the promotion of clean and energy-efficient road transport vehicle [Online]. Available: http s://eur-lex.europa.eu/eli/dir/2019/1161/oj/eng. [Accessed 27 April 2025]. [11] European Commission. Regulation (EU) 2018/858 of the European Parliament and of the Council of 30 May 2018 on the approval and market surveillance of motor vehicles and their trailers, and of systems, components and separate technical units intended for such vehicles, amending Regulations (EC) no 715/2007 and (EC) no 595/2009 and repealing Directive 2007/46/EC [Online]. Available: https://eu r-lex.europa.eu/eli/reg/2018/858/oj/eng. [Accessed 27 April 2025]. [12] Cespira. Westport and Volvo Cespira [Online]. Available: https://cespira.com/ourtechnology/hpdi-h2/. [Accessed 18 December 2024]. [13] Iveco. IVECO S-Way H2-ICE [Online]. Available: https://www.iveco.com/global /Press/PressReleases/2024/IVECO-strides-ahead-on-the-road-of-change-withzero-emission-offering-unveiled-at-IAA-Transportation-2024. [Accessed 18 December 2024]. [14] European Commission. REGULATION (EU) 2024/1610 of the European Parliament and of Council of 14 May 2024 amending Regulation (EU) 2019/1242 as regards strengthening the CO2 emission performance standards for new heavy-duty vehicles and integrating reporting obligations, amending Regulation (EU) 2018/ 858 and repealing Regulation (EU) 2018/956 [Online]. Available: https://eur-lex. europa.eu/eli/reg/2024/1610/oj/eng. [Accessed 18 December 2024]. [15] Siddiqui O, Ishaq H, Khan DA, Fazel H. Social cost-benefit analysis of different types of buses for sustainable public transportation. J Clean Prod Jan. 2024;438: 140656. https://doi.org/10.1016/j.jclepro.2024.140656. [16] Frieß NM, Pferschy U. Planning a zero-emission mixed-fleet public bus system with minimal life cycle cost. Public Transp Mar. 2024;16(1):39–79. https://doi.org/ 10.1007/s12469-023-00345-4. [17] Islam A, Lownes N. When to go electric? A parallel bus fleet replacement study. Transp Res D Transp Environ Jul. 2019;72:299–311. https://doi.org/10.1016/j. trd.2019.05.007. [18] Avenali A, De Santis D, Giagnorio M, Matteucci G. Bus fleet decarbonization under macroeconomic and technological uncertainties: a real options approach to support decision-making. Transp Res E Logist Transp Rev 2024;190(Oct). https:// doi.org/10.1016/j.tre.2024.103690. [19] Stasko TH, Oliver Gao H. Reducing transit fleet emissions through vehicle retrofits, replacements, and usage changes over multiple time periods. Transp Res D Transp Environ 2010;15(5):254–62. https://doi.org/10.1016/j.trd.2010.03.004. [20] Ribeiro PJG, Mendes JFG. Public transport decarbonization via urban bus fleet replacement in Portugal. Energies (Basel) Jun. 2022;15(12). https://doi.org/ 10.3390/en15124286. [21] Bakker J, Lopez Alvarez JA, Veldman J, Buijs P. Strategic fleet replacement for the electrification of heavy-duty road freight transportation. Appl Energy 2025;391 (Aug). https://doi.org/10.1016/j.apenergy.2025.125935. [22] Othman K, Hamed S, Da Silva D, Shalaby A, Abdulhai B. Decision support tools for effective bus fleet electrification: replacement factors and fleet size prediction. Transp Res Interdiscip Perspect 2024;28(Nov). https://doi.org/10.1016/j. trip.2024.101267. [23] Enayati E, Rasti-Barzoki M, Jafari H, Altmann J. A game-theoretic approach for the sustainable development of public bus fleet under government policies considering passenger behavior: a case study of China. Energy May 2025;323. https://doi.org/ 10.1016/j.energy.2025.135345. [24] Tang C, Zhao A, Liu T, Zhang J. Electric vehicle type selection for single-line transit fleet replacement. J Public Trans 2023;25(Jan). https://doi.org/10.1016/j. jpubtr.2023.100077. [25] Lu C, Xie DF, Zhao XM, Qu X. The role of alternative fuel buses in the transition period of public transport electrification in Europe: a lifecycle perspective. Int J Sustain Transp 2023;17(6):626–38. https://doi.org/10.1080/ 15568318.2022.2079445. [26] Lajunen A, Lipman T. Lifecycle cost assessment and carbon dioxide emissions of diesel, natural gas, hybrid electric, fuel cell hybrid and electric transit buses. Energy Jul. 2016;106:329–42. https://doi.org/10.1016/j.energy.2016.03.075. [27] Ally J, Pryor T. Life cycle costing of diesel, natural gas, hybrid and hydrogen fuel cell bus systems: an Australian case study. Energy Policy Jul. 2016;94:285–94. https://doi.org/10.1016/j.enpol.2016.03.039. [28] Feng W, Figliozzi M. Vehicle technologies and bus fleet replacement optimization: problem properties and sensitivity analysis utilizing real-world data. Public Transp Apr. 2014;6(1–2):137–57. https://doi.org/10.1007/s12469-014-0086-z. [29] Pelletier S, Jabali O, Mendoza JE, Laporte G. The electric bus fleet transition problem. Transport Res C Emerg Technol Dec. 2019;109:174–93. https://doi.org/ 10.1016/J.TRC.2019.10.012. [30] Yıldırım S ¸, Yıldız B. Electric bus fleet composition and scheduling. Transport Res C Emerg Technol Aug. 2021;129:103197. https://doi.org/10.1016/j. trc.2021.103197. [31] Mu˜ noz P, Franceschini EA, Levitan D, Rodriguez CR, Humana T, Correa Perelmuter G. Comparative analysis of cost, emissions and fuel consumption of diesel, natural gas, electric and hydrogen urban buses. Energy Convers Manag Apr. 2022;257:115412. https://doi.org/10.1016/j.enconman.2022.115412. [32] Johnson C, Nobler E, Eudy L, Jeffers M. “Financial Analysis of Battery Electric Transit Buses,” Golden, CO (United States). Jun. 2020. https://doi.org/10.2172/ 1659784. [33] BloombergNEF. Lithium-ion battery pack prices hit record [Online]. Available: https://about.bnef.com/blog/lithium-ion-battery-pack-prices-hit-record-low-of -139-kwh/. [Accessed 2 January 2025]. [34] Kumar Parveen, Mulukutla Pawan, Doshi Priyansh. “Real-world Electric Bus Operation: trend in Technology, Performance, Degradation, and Lifespan of Batteries. ,”. WRI INDIA; 2024. [35] European Commission. Regulation (EU) 2023/1542 of the European Parliament and of the Council of 12 July 2023 concerning batteries and waste batteries, amending Directive 2008/98/EC and Regulation (EU) 2019/1020 and repealing Directive 2006/66/EC [Online]. Available: https://eur-lex.europa.eu/eli/r eg/2023/1542/oj/eng. [Accessed 27 April 2025]. [36] Bloomberg, “Electric buses in cities. Driving towards cleaner aie and lower CO2.”. [37] Kotowicz J, Baszcze´ nska O, Niesporek K. Cost of green hydrogen. Energies (Basel) Sep. 2024;17(18):4651. https://doi.org/10.3390/en17184651. [38] Dyr T, Misiurski P, Zi´ ołkowska K. Costs and benefits of using buses fuelled by natural gas in public transport. J Clean Prod Jul. 2019;225:1134–46. https://doi. org/10.1016/j.jclepro.2019.03.317. [39] Razy-Yanuv E, Meron N. Early transition to cleaner bus fleets: benefits, costs, and policy evaluation of alternatives from a life cycle perspective. Clean Environ Sys Mar. 2024;12:100172. https://doi.org/10.1016/j.cesys.2024.100172. [40] Hensher DA. Bus transport: economics, policy and planning, vol. 18. Elsevier; 2007. [41] Quarles N, Kockelman KM, Mohamed M. Costs and benefits of electrifying and automating bus transit fleets. Sustainability May 2020;12(10):3977. https://doi. org/10.3390/su12103977. [42] TUSSAM.” Accessed: September. 18, 2024. [Online]. Available: https://www.tuss am.es/es/node/1276. [43] Alternative Fuels Observatory EC EU. Electric vehicle recharging prices [Online]. Available: https://alternative-fuels-observatory.ec.europa.eu/consumer-portal /electric-vehicle-recharging-prices. [Accessed 30 April 2025]. [44] Commision European. Directorate-general for mobility and transport. In: Handbook on the External costs OD Transport; 2019. [45] Zhou Y, Ong GP, Meng Q. The road to electrification: bus fleet replacement strategies. Appl Energy May 2023;337:120903. https://doi.org/10.1016/j. apenergy.2023.120903. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 19 [46] Guzzella L, Sciarretta A. Vehicle propulsion systems. Berlin, Heidelberg: Springer Berlin Heidelberg; 2013. https://doi.org/10.1007/978-3-642-35913-2. [47] García-Afonso ´ O. Impact of powertrain electrification on the overall CO2 emissions of intercity public bus transport: tenerife Island test case. J Clean Prod Aug. 2023; 412:137365. https://doi.org/10.1016/j.jclepro.2023.137365. [48] Vagg C, Brace CJ, Hari D, Akehurst S, Poxon J, Ash L. Development and field trial of a driver assistance system to encourage eco-driving in light commercial vehicle fleets. IEEE Trans Intell Transport Syst Jun. 2013;14(2):796–805. https://doi.org/ 10.1109/TITS.2013.2239642. [49] Saboohi Y, Farzaneh H. Model for developing an eco-driving strategy of a passenger vehicle based on the least fuel consumption. Appl Energy Oct. 2009;86 (10):1925–32. https://doi.org/10.1016/j.apenergy.2008.12.017. [50] C. A. E. E. M. I. and T. Federal Ministry Republic of Austria, Guidelines for National eco-driving initiatives.. [51] Rymarz J, Niewczas A, Krzy˙ zak A. Comparison of operational availability of public city buses by analysis of variance. Maintenan Reliab Jun. 2016;18(3):373–8. https://doi.org/10.17531/ein.2016.3.8. [52] Manzolli JA, Trov˜ ao JPF, Henggeler Antunes C. Electric bus coordinated charging strategy considering V2G and battery degradation. Energy Sep. 2022;254:124252. https://doi.org/10.1016/j.energy.2022.124252. [53] Zhang S, et al. Real-world fuel consumption and CO2 emissions of urban public buses in Beijing. Appl Energy Jan. 2014;113:1645–55. https://doi.org/10.1016/j. apenergy.2013.09.017. [54] Deliali A, Chhan D, Oliver J, Sayess R, Godri Pollitt KJ, Christofa E. Transitioning to zero-emission bus fleets: state of practice of implementations in the United States. Transp Rev Mar. 2021;41(2):164–91. https://doi.org/10.1080/ 01441647.2020.1800132. [55] Rosero F, Fonseca N, L´ opez J-M, Casanova J. Effects of passenger load, road grade, and congestion level on real-world fuel consumption and emissions from compressed natural gas and diesel urban buses. Appl Energy Jan. 2021;282: 116195. https://doi.org/10.1016/j.apenergy.2020.116195. M.A. Tagua Navarrete et al. Energy 332 (2025) 137025 20