Source-linked AI summary
Planning electric bus systems with solar photovoltaic integration using open transit data: A case study of the Dakar BRT
Jérémy Dumoulin, Cheikh Mouhamed Fadel Kebe, Babacar M. Ndiaye, Noémie Jeannin, Christophe Ballif, Nicolas Wyrsch
TL;DR
Detailed transit data are scarce in many settings, complicating electric-bus planning across operational, economic, environmental, and grid objectives. GTFS4EV uses open transit data to simulate operations and assess electrification strategies; in Dakar, opportunity charging reduced battery and peak-demand requirements versus depot-only charging, while PV integration brought fuel-cost savings.
Problem
Electric-bus planning must balance operational feasibility, costs, environmental benefits, and grid impacts, but detailed vehicle-level operational data are scarce in many contexts.
Source: Context and motivation · p. 2Method
GTFS4EV uses GTFS data to reconstruct bus operations and assess charging scenarios, technical requirements, and multidimensional impacts.
Source: Fleet operation simulation · p. 7, §6 · p. 25Results
Versus Depot only, Terminal and depot reduces average battery capacity from 285 to 57 kWh and peak charging demand from 5.3 to 1.5 MW; PV fuel-cost savings reach 36% for Terminal and depot and 46% for Terminal only.
Source: §5 · p. 23, §6 · p. 25Takeaways & Limitations
The Dakar application demonstrates that widely available GTFS data can support strategic planning for electric-bus systems.
Source: §6 · p. 25The study applies one energy-consumption rate to all buses, so it does not capture possible line-specific differences such as stopping patterns.
Source: §5 · p. 25
Abstract
from arXiv · showhide
Electrification of urban bus systems is expected to accelerate due to its clear environmental benefits. However, effective electric bus deployment requires planning tools that can support decision-making on electrification strategies while balancing operational feasibility, costs, environmental benefits, and impacts on the local electricity grid. Such planning depends on the availability of detailed transit data, which remain scarce in many cities, particularly in developing countries. To address this gap, we present GTFS4EV, an open-source framework that uses publicly available General Transit Feed Specification (GTFS) data to simulate bus operations and evaluate electrification scenarios. The framework provides quantitative insights across multiple dimensions. It estimates the minimum onboard battery capacity required for each bus, alongside charging infrastructure needs, economic and environmental impacts, and the potential for solar photovoltaic integration. We apply the framework to the Dakar Bus Rapid Transit system, comparing three charging strategies ("Depot only", "Terminal and depot" and "Terminal only") across different levels of PV capacities. Results show that the "Terminal and depot" strategy reduces the minimum onboard battery capacity from 285 to 57 kWh per bus, and the maximum charging load from 5.3 to 1.5 MW relative to the "Depot only" strategy. The "Terminal only" also provides benefits compared to the "Depot only", although it requires larger onboard battery capacities (155 kWh). Moreover, combining opportunity charging with solar photovoltaics reduces the charging costs by up to 46% relative to grid charging only. The study demonstrates how GTFS data can be leveraged to support electric bus planning and photovoltaic integration in data-scarce contexts.
1. Introduction
Electric-bus planning must balance the needs of bus operators, power-system operators, and public authorities, but detailed operational data remain difficult to obtain in many contexts. Existing approaches trade operational realism against data accessibility, while comprehensive, open, multidimensional planning evidence—particularly for emerging economies—remains limited.
Context and motivation
- Bus electrification requires coordinated decisions on fleet batteries, charging infrastructure, reliable operations, grid stability, and environmental and service objectives.The relevant stakeholders are bus operators, power-system operators, and public authorities.Source: Context and motivation · p. 2
- Vehicle schedules, routes, and dwell times are needed to estimate energy use and charging needs, yet detailed vehicle-level data are often unavailable or not systematically collected.Studies may therefore rely on assumptions or proxy data, introducing uncertainty into planning outcomes.Source: Context and motivation · pp. 2–3
Literature review
- High-fidelity models use granular operational data that are often inaccessible, while generic screening tools can overlook operational characteristics that affect charging demand and investment decisions.The literature points to open approaches using available transit data while retaining sufficient operational realism for context-specific insights.Source: Literature review · p. 3
- Comprehensive GTFS-based planning, renewable integration, and evidence from Africa and other emerging economies remain limited in the literature.Existing applications are concentrated in Europe, North America, and selected parts of Asia, raising questions about transferability to differing mobility, grid, and data contexts.Source: Literature review · p. 4
Identified gaps and contribution
The paper addresses gaps in open, realistic, multidimensional bus-electrification planning and its application to emerging-economy contexts. It introduces GTFS4EV, applies it to Dakar BRT, and sets out a modular workflow for evaluating charging scenarios and their technical, economic, environmental, and PV dimensions.
Identified gaps and contribution
- Existing tools and studies leave gaps in open GTFS-based planning, grid and environmental assessment, distributed PV integration, and evidence from emerging economies.The literature also raises transferability questions for regions with different objectives and constraints.Source: Identified gaps and contribution · pp. 4–5
- GTFS4EV is an open-source framework that uses GTFS to simulate bus operations and assess charging demand, battery and infrastructure needs, economic and environmental impacts, and solar PV potential.It is intended to support planning where detailed vehicle-level data are unavailable.Source: Identified gaps and contribution · p. 5
- The Dakar BRT application evaluates charging strategies and PV integration to quantify trade-offs among battery sizing, infrastructure, PV use, costs, and environmental performance.The study describes Dakar BRT as one of the first fully electric BRT systems in Sub-Saharan Africa.Source: Identified gaps and contribution · p. 5
Overview of the GTFS4EV framework
- GTFS4EV preprocesses GTFS data, simulates fleet operations, models scenario-based charging, and conducts an ex-post multidimensional assessment.The assessment includes PV potential and economic and environmental metrics.Source: Overview of the GTFS4EV framework · pp. 5–6
- The modular framework determines battery capacity and charger numbers endogenously from simulated operations, while also allowing exogenous design variables for constrained evaluations.Additional charging strategies can be incorporated without changing the core structure.Source: Overview of the GTFS4EV framework · p. 6
GTFS data pre-processing
GTFS4EV validates and optionally enriches GTFS data, reconstructs daily vehicle operations, and simulates charging scenarios to estimate battery and charging-infrastructure requirements.
GTFS data pre-processing
- GTFS4EV checks required GTFS files iteratively for inconsistent relationships, then optionally filters routes or periods and adds dwell times absent from schedules.The model requires frequencies.txt despite its optional status in the GTFS specification.Source: GTFS data pre-processing · p. 7
Fleet operation simulation
- Trips are reconstructed as sequences of travel, stops, and terminal dwell, while headways determine vehicle dispatch and fleet requirements.Fleet size is fixed at the maximum required across service-frequency periods, and the model assumes vehicles progressively enter and leave service.Source: Fleet operation simulation · pp. 7–8
Scenario-based charging
- Charging events follow user-defined locations and decision logic; cumulative charged energy minus traction use determines each bus’s theoretical minimum battery capacity.The model assumes equal beginning- and end-of-day state of charge, and its battery-capacity measure excludes additional nominal-capacity allowances such as reserve capacity.Source: Scenario-based charging · pp. 8–9, Figure 3 · p. 9
- Required charging infrastructure is based on the peak number of vehicles charging simultaneously at each location.Charging schedules for trips sharing a location are aggregated before the peak is evaluated.Source: Scenario-based charging · p. 9
Local PV production model
The local PV model estimates hourly electricity production over a year from weather data, module specifications, and installation parameters.
Local PV production model
- PV production is simulated at hourly resolution over one year using local weather data, module specifications, and installation parameters.The weather data come from PVGIS-SARAH3, and the simulation uses pvlib.Source: Local PV production model · p. 9
- The model computes plane-of-array irradiance from irradiance data and PV module orientation, then estimates electrical output with the PVWatts model.PVWatts is implemented in pvlib.Source: Local PV production model · p. 9
- PV output estimation accounts for cell temperature, angular losses, and an overall system-loss factor.Cell temperature uses the built-in PVsyst thermal model, and angular losses follow Martin and Ruiz.Source: Local PV production model · p. 10
Multi-dimensional performance assessment
GTFS4EV evaluates electrification scenarios through economic, environmental, and PV-integration indicators, including how local PV aligns with bus charging demand.
Multi-dimensional performance assessment
- The assessment focuses on fuel-cost savings, CO2-emission reductions, and the potential for local PV to meet bus-charging demand.GTFS4EV also supports local air-pollutant indicators, but those are outside this study’s scope.Source: Multi-dimensional performance assessment · p. 10
- PV integration is characterized by energy coverage, self-sufficiency, and self-consumption, with the latter two accounting for coincidence between PV generation and charging demand.Energy coverage measures theoretical PV contribution irrespective of temporal matching; the metrics can be evaluated over different time horizons.Source: Multi-dimensional performance assessment · p. 10
- Annual CO2 emissions and fuel costs are estimated for diesel and electric bus operation using annual distance and energy or fuel consumption.Electric-bus emissions and costs use electricity carbon intensity and price, respectively; diesel calculations use diesel intensity and price.Source: Multi-dimensional performance assessment · pp. 10–11
- When PV is included, effective electricity carbon intensity and price are adjusted using annual self-sufficiency and self-consumption.The weighted-average approach combines grid and PV carbon intensities and uses grid tariffs and the levelized cost of PV electricity.Source: Multi-dimensional performance assessment · p. 11
3. Case study
Dakar’s dense, uneven urban structure contributes to chronic congestion, and its transport plan prioritizes an integrated multimodal network. The fully electric Dakar BRT is an 18 km corridor whose current overnight-depot charging leaves daytime opportunity charging and PV integration unexplored.
3.1 Urban mobility context in Dakar
- Dakar concentrates nearly 4 million people in 547 km2, and its western employment concentration and eastward residential expansion have contributed to chronic congestion and other urban externalities.The cited externalities include air pollution, noise exposure, and high commuting costs.Source: §3.1 · p. 11
- Dakar’s Urban Mobility Plan 2035 envisions a multimodal system with mass transit, feeder buses, and last-mile services, alongside a gradual phase-out of informal transport.Earlier modernization initiatives include a toll highway, the Regional Express Train, and the BRT.Source: §3.1 · p. 12
3.2 The Dakar BRT
- The fully electric Dakar BRT is an 18 km corridor designed for up to 300,000 passengers daily, with a planned 121-bus fleet; two lines were operational at the study’s time.The current strategy uses overnight depot charging, leaving daytime opportunity charging and PV integration untapped.Source: §3.2 · p. 12
GTFS feed and pre-processing
The study uses weekday GTFS schedules to represent high-demand Dakar BRT operations, with minor preprocessing. It also estimates local PV production and identifies about 10 MWp of potential deployment capacity at key BRT facilities.
GTFS feed and pre-processing
- Weekday GTFS data represent a high-demand case, and preprocessing identifies regular B1 and B2 services plus B1 reinforcement buses during peak hours.The feed adequately captures weekday operating patterns after minor preprocessing.Source: GTFS feed and pre-processing · pp. 12–13
- The corridor map shows B1 serving all displayed stops, while B2 serves only the green stops between Petersen and Guédiawaye.Source: Figure 4 · p. 13
PV system
- PV simulations use Dakar 2020 weather and predict an annual yield of 1877.9 kWh/kWp, with lowest production during the June–September rainy season.The model assumes 22% module efficiency, 14% system losses, and a 17° optimal tilt.Source: PV system · p. 13
- About 10 MWp of PV capacity is estimated across the Gadaye depot, planned stations, and Grand Médine parking area.The estimate is conservative because it excludes terminals, other operator facilities, and future BRT infrastructure.Source: PV system · pp. 13–14
Fuel costs and CO2 emissions
The assessment sets Dakar-specific fuel, electricity, and emissions parameters and estimates PV carbon intensity and generation cost. It compares three charging strategies under assumptions of unchanged schedules and uncontrolled charging.
Fuel costs and CO2 emissions
- The assessment assumes diesel consumption of 0.41 L.km−1, diesel at 1.184 USD.L−1, and grid electricity at 0.20 off-peak and 0.32 USD.kWh−1 on-peak.Grid price is weighted by charging energy in each tariff period; Senegal grid carbon intensity is 500 gCO2.kWh−1 and diesel emissions are 2.7 kgCO2.L−1.Source: Fuel costs and CO2 emissions · p. 14
- PV generation is estimated at 0.0695 USD.kWh−1, while its carbon intensity is 22.1 gCO2.kWh−1.The carbon intensity is lower than the cited European average of 36 gCO2.kWh−1; the PV cost estimate assumes a 25-year lifetime and 9% discount rate.Source: Fuel costs and CO2 emissions · pp. 14–15
3.4 Charging strategies
- The scenarios are depot-only overnight charging at 120 kW, terminal-and-depot charging at 120 kW, and terminal-only fast charging at 360 kW.Terminal-only charging uses a 0.7 charging probability calibrated to distribute demand while retaining 100% operational feasibility.Source: §3.4 · p. 15
- Charging is restricted to existing layovers without bus rescheduling, and buses charge immediately at maximum available power without smart charging.Bus energy consumption is assumed to be 1.3 kWh.km−1.Source: §3.4 · pp. 15–16
4. Results
Terminal-based opportunity charging shifts demand into daytime, reduces battery and charger requirements, and creates charging opportunities from PV. The benefits vary by strategy and PV capacity, with emissions- and cost-minimizing capacities not coinciding.
4.1 Bus operation and the potential for alternative charging strategies
- 20–28% of buses idle at stops or terminals, with most idle time at terminals, while 26% remain at the depot during off-peak hours.Terminal idle periods span 1.6–2.6 hours across service patterns; daily distances range from 74 km for B1 reinforcement buses to 269 km for B2 buses.Source: Figure 6 · p. 16, §4.1 · pp. 16–17
4.2 Charging load curve
- 5.3±0.3 MW at 22:30 under Depot only falls to 1.5±0.2 MW under Terminal and depot; Terminal only remains near 1.8 MW from 8:00 to 13:00.Terminal-based charging shifts demand into daytime; Terminal and depot has a relatively flat daytime profile around 1 MW.Source: §4.2 · p. 17, Figure 7 · p. 18
4.3 Battery and charging infrastructure requirements
- Average required battery capacity falls from 285 kWh under Depot only to 57 kWh under Terminal and depot, while Terminal only averages 155 kWh.Terminal and depot requires 13 depot and 12 terminal chargers; Terminal only requires 10, with simulations ranging from 9 to 11.Source: §4.3 · pp. 18–20, Figure 8 · p. 19
4.4 Potential for PV-based charging
- At 10 MWp, annual self-sufficiency reaches 65% for Terminal and depot and 82% for Terminal only; at 2.8 MWp, both strategies exceed 50% in annual self-sufficiency and self-consumption.At 2.8 MWp, daily self-sufficiency ranges from 23–58% for Terminal and depot and 27–74% for Terminal only; Terminal and depot can perform better on less sunny days.Source: §4.4 · pp. 20–21, Figure 9 · p. 21
4.5 Implications on carbon emissions and fuel costs
- At 5 MWp, annual per-bus CO2 emissions reach 23.5 and 14.6 tCO2.yr−1 for Terminal and depot and Terminal only, respectively; fuel-cost minima occur at 2 and 3 MWp.At those cost minima, annual fuel costs are US$15.9k and US$13.5k, respectively; emissions rise slightly beyond 5 MWp as self-sufficiency plateaus.Source: Figure 10 · p. 22, §4.5 · p. 22
5. Discussion
The discussion shows that charging strategy shapes technical requirements, PV use, and economic outcomes, making integrated planning relevant to both operators and power-system stakeholders. Opportunity charging offers benefits, but its infrastructure and PV trade-offs depend on modeling assumptions.
5. Discussion
- Charging strategy is a key design variable because it shapes investment requirements, so strategic and tactical bus planning should be considered jointly rather than sequentially.GTFS4EV offers a rapid, flexible way to explore scenarios using widely available GTFS data and provides bus-specific battery requirements.Source: §5 · p. 23
- Terminal only favors PV integration but requires greater battery capacity and higher peak charging power without PV; the comparison with Terminal and depot reveals trade-offs across indicators.Table 1 groups the qualitative strategy comparison by stakeholder and distinguishes indicators with and without PV.Source: §5 · p. 23, Table 1 · p. 24
- 36% and 46% fuel-cost reductions versus Depot only, for Terminal and depot and Terminal only respectively, correspond to 12-year savings of about US$108k and US$137k per bus.The study reports that both opportunity-charging strategies outperform Depot only on CO2 emissions across the PV-capacity range.Source: §5 · p. 23
- PV benefits are conservative because smart charging, stationary storage, and feed-in tariffs are excluded; system-wide aggregation also leaves site-level PV and grid effects unresolved.The study also notes that daytime depot idling is not used to improve PV integration.Source: §5 · pp. 24–25
- Battery-capacity reductions correspond to estimated first-order savings of US$79,800 per bus for Terminal and depot and US$45,500 for Terminal only.These estimates exclude operating margins, degradation, and auxiliary loads, so they are not final battery-sizing costs.Source: §5 · p. 24
- Opportunity charging reduces required charging stations, but infrastructure-cost effects are uncertain: 360 kW chargers cost about twice as much as 120 kW chargers, and terminal charging assumes no queues.The model’s no-queue assumption could affect both the number of chargers needed and achievable charging schedules.Source: §5 · pp. 24–25
6. Conclusion
GTFS4EV uses widely available GTFS data to support electric-bus planning, and the Dakar BRT comparison shows that charging strategies entail different battery, charging-demand, and infrastructure requirements. The conclusion is that these investments should be planned jointly, with the framework offering a transferable decision-support tool where vehicle-level operational data are unavailable.
6. Conclusion
- 285 to 57 kWh average battery capacity and 5.3 to 1.5 MW peak charging demand are reported for Terminal and depot versus Depot only; combining opportunity charging with PV cuts annual charging costs by up to 46%.Terminal and depot also reduces required chargers by more than half; Terminal only requires 155 kWh and fast terminal chargers.Source: §6 · p. 25
- Battery capacity, charging infrastructure, and PV sizing are closely coupled to charging strategy, so strategic and tactical planning should be considered jointly.The conclusion calls for a shift away from conventional sequential bus planning.Source: §6 · pp. 25–26
- GTFS4EV is presented as a transferable decision-support tool for regions where detailed vehicle-level operational data remain unavailable.The framework uses widely available GTFS data to support strategic electric-bus planning.Source: §6 · pp. 25–26
Appendix A. Derivation of the effective electricity price
The appendix derives an effective electricity-price formulation from annual grid purchases and PV generation costs, using self-sufficiency and self-consumption. It treats uncompensated PV exports as a conservative case.
Appendix A. Derivation of the effective electricity price
- Annual electricity cost is framed as grid-purchase costs plus PV-generation costs, then divided by annual demand to obtain an effective electricity price.The derivation introduces self-sufficiency and self-consumption into that price expression.Source: Appendix A · p. 26
- Self-sufficiency, self-consumption, and PV levelized cost are not independent; more PV typically raises self-sufficiency while lowering self-consumption.The cost effect depends on the balance between these changes.Source: Appendix A · p. 26
- The formulation assigns no value to excess PV electricity exported to the grid, making the no-compensation case conservative.Feed-in remuneration could be included as revenue or treated as lowering net electricity cost.Source: Appendix A · p. 26
Appendix B. Charging probability calibration
The appendix calibrates a stochastic terminal-charging probability to balance demand dispersion and operational feasibility, then documents GTFS data preprocessing and author contributions.
Appendix B. Charging probability calibration
- pcharge = 0.7 is selected for Terminal only, with 100% feasibility; lower values spread charging events but risk insufficient charging opportunities.Feasibility was tested across pcharge values from 0.5 to 1.0 over multiple stochastic realizations.Source: Appendix B · p. 27, Figure B.1 · p. 27
Appendix C. GTFS data pre-processing
- The GTFS feed’s trip schedules were used to reconstruct service frequencies; preprocessing also simplified stops and shapes and added 10 minutes of terminal idle time.The inspection identified B1 Omnibus, B2 Semi-Express, and peak-period B1 reinforcement service patterns.Source: Appendix C · pp. 27–28
CRediT authorship contribution statement
- The contribution statement assigns conceptualization, software, and data curation to Jérémy Dumoulin, and supervision and funding acquisition to Christophe Ballif and Nicolas Wyrsch.Source: CRediT authorship contribution statement · p. 28