Managing grid impacts from increased electric vehicle adoption in African cities | Scientific Reports

Oct 18, 2024

Scientific Reports volume 14, Article number: 24320 (2024) Cite this article

Metrics details

Electric vehicles are pivotal for global climate solutions, particularly in emerging markets like Africa. Despite the continent’s clean energy potential, electric vehicle adoption faces unique challenges due to inefficiencies and reliability issues of distribution power grids. Here, we analyze the impacts of expanding electric vehicle fleets—private, commercial, and paratransit—on Nairobi’s power grid. We simulate traffic patterns, charging behaviors, and transformer utilization using local mobility data. Our results show that while electric commercial and paratransit fleets may improve power system efficiency, widespread private EV adoption could significantly strain the grid, increasing peak loads and transformer aging. Smart charging strategies could mitigate these issues, reducing potential transformer replacement costs by up to 40%. Our study highlights the importance of tailored demand management and infrastructure planning to support EV growth in African cities, providing critical insights for policymakers, utilities, and transport planners to facilitate sustainable electric mobility transitions.

Electric Vehicles (EVs) are emerging as a critical global climate solution. The latest State of Climate Action report highlights that while progress on global climate action in each of the 42 sectoral indicators assessed lags behind, electric mobility is an exception1. Over the past five years, the share of EVs in global passenger car sales has surged from 1.6% in 2018 to 10% in 2022, with the International Energy Agency projecting 240 million passenger EVs by 20302.

African cities are rapidly motorizing, impacting emissions, urban air pollution, and transit affordability. African countries, endowed with clean energy resources—making up 30% of electricity generation in 2019, and projected to rise to nearly half by 2040 3,4 — are incentivized to electrify transportation. EVs offer a promising avenue to reducing reliance on imported fuel and carbon-intensive technologies. African nations like Cape Verde, Rwanda, Zimbabwe, South Africa, Egypt, Ghana, and Kenya have signaled their commitments to EV adoption through targets and incentives5. For example, Kenya reduced excise duty on EV imports from 20 to 10%, aiming for a 5% share of EVs in vehicle imports by 2025 6.

Despite this momentum, adoption rates remain comparatively low due to unique challenges like the capacity of their electricity grids to support EVs. African utilities notoriously score poorly on grid efficiency and reliability metrics, with high system losses (median of over 20%), and system average interruption frequency or SAIFI (median of 81 disruptions per year)7. The World Bank identifies grid capacity as a significant barrier to EV adoption, citing issues like feeder and transformer overloading, voltage deviations, and power losses. Distribution grids will need reinforcements for EV deployment8. While power system impacts may extend to generation dispatch or transmission, these are expected to be less severe and more manageable. Comprehensive planning and collaboration among utilities, system operators, regulators, and policymakers are crucial to minimize negative impacts and maximize the benefits of EV adoption.

Given the unique contexts of African cities, tailored methods are essential for assessing the grid implications of EV adoption and designing locally appropriate risk mitigation plans. For instance, in many African cities, privately owned, informally operated “public” transport vehicles, known as paratransit systems, such as matatus in Kenya and danfos in Nigeria, dominate road-based public passenger transport9, with their share ranging from 65% in Yaoundé to as high as 98% in Dar es Salaam, profoundly shaping transportation demand across the continent10,11. Thus, data, models, assumptions, and insights from studies on EV adoption in industrialized cities12,13,14,15,16,17,18,19,20,21,22 may not be directly transferable to African cities. Therefore, this study fills a critical research gap by providing the first integrated analysis of context-specific EV multi-fleet expansion in African cities using local mobility data, focusing on Nairobi, Kenya, a rapidly growing metropolitan area projected to reach 10 million residents by 2050 28.

A comprehensive assessment of the interaction between transportation networks, mobility patterns, electricity infrastructure, consumer behavior, and power demand management, is required to understand the impact of EVs on a city’s power system and demand management strategies. However, existing studies exploring the impacts of EVs on power grids mainly focus on either single-fleet adoption, particularly private EVs, often neglecting spatial mobility patterns23,24,25, demand management26,27, or consumer behavior considerations18,28. There is a need for a holistic analysis of how multiple vehicle fleets, diverse charging behaviors, and power demand management intersect to affect grid performance. Yet, to our knowledge, no study on EV adoption and grid impacts provides a holistic integrated assessment of this interaction. Therefore, our study fills this gap by developing a comprehensive model (Fig. 1) that simulates the traffic patterns of multiple fleets – including private, commercial, and paratransit vehicles, driver charging decisions, EV charging demand, and distribution transformer utilization to explore the effects of progressively scaling up different EV fleets on electricity supply utilization and estimate transformer overloading, a critical grid reliability indicator. Additionally, we evaluate the techno-economic implications of charging demand management strategies, including avoiding battery energy storage costs and premature transformer replacements.

Flowchart showing the integrated model and simulation framework.

Our analysis reveals that while paratransit and commercial EV adoption may enhance grid conditions, widespread private EV uptake could strain the grid, significantly increasing peak electric load and transformer aging, leading to higher outages and early replacement costs of up to USD 6.5 million within five years. This strain worsens with high EV owner-range anxiety. However, smart charging strategies for private and commercial EVs could mitigate these costs. Paratransit vehicles, charging during low-demand periods, benefit less from these strategies. This study offers essential insights for electric utility companies, transport planners, regulators, and policymakers, guiding data-driven policies and decisions for effective electric mobility transitions in African cities.

We developed a stochastic model to predict EV movement patterns and estimate spatiotemporal charging demand. Identifying high-demand areas is crucial for targeted infrastructure upgrades, optimizing power distribution system assets, and demand management. Our spatial unit of analysis is a city zone, a small geographic area corresponding to a neighborhood as defined by the Uber Movement Project29 for aggregating and anonymizing trip data. Our study area includes 400 city zones, ranging from 0.8 to 8 km², with a mean spatial resolution of 2.7 km².

Using granular socio-economic data, we modeled the spatial distribution of private EV ownership, predicting that early adopters, thus home charging demand, would be concentrated in affluent areas, consistent with trends in countries with significant EV penetration30,31,32. For light-duty commercial fleets, charging demand is expected to be highest in commercial zones, reflecting their typical use for goods transport and errands33. Paratransit vehicles, like matatus and buses, operate on set routes connecting central and peri-urban areas (Supplementary Fig. 5)34.

Our model uses travel surveys and traffic data to predict EV charging demand patterns. We simulate daily round-trips for private EVs with randomly assigned destinations, departure, and travel times. Commercial EVs charge at their origin within commercial zones, traveling to multiple destinations daily based on probabilistic travel patterns. Discussions with electric bus operators informed our assumption that paratransit EVs charge during off-peak operation hours35, with infrastructure near termini.

Our results (Fig. 2a) show that most private EV trips average 33 km/day, consistent with a previous study on Nairobi’s urban transportation36. Commercial EVs cover about 40–70 km/day, with 3 to 5 stops, consistent with a prior pilot study of four light-duty commercial EVs within Nairobi37. Paratransit EVs cover 120 to 220 km/day, operating multiple short or fewer long trips, aligning with a previous study on matatu travel patterns33.

(a) Distribution of daily electric vehicle kilometers traveled by fleet. (b) Distribution of the average proportion of vehicle battery capacity drawn daily by fleet. (c) Comparison of daily charging events across fleets as a function of the daily distance coverage. (d) Spatial distribution of daily charging demand by fleet.

We estimated daily EV energy requirements using a probabilistic model of vehicle range and battery capacity. Simulating vehicle movement and charging over ten days, we found that private EVs use up to 25% of their battery capacity daily, with some using up to 100%. Commercial EVs use 30–50% of their capacity daily. Therefore, most private and commercial EVs can complete daily trips on a single charge. At the same time, paratransit vehicles often require multiple charges per day -- only about 15% of matatu and 60% of bus daily operations can be completed on a single charge (Fig. 2b).

Our results show that up to 60% of the private and commercial EV fleets would need to charge daily, with less than 20% of private EVs that cover less than 20 km/day charging concurrently on any given day. In contrast, e-buses on average, may need at least two charging events daily, while e-matatus covering over 150 km/day may need on average five daily charging events(Fig. 2c). This suggests the need for strategic placement of charging infrastructure for electric paratransit fleets to accommodate multiple daytime charging events with minimal to no disruptions to their operations. In contrast, home charging only could cater for most private EVs.

Our analysis further indicates that private and commercial fleet charging demand will be more widely distributed, covering 94% and 87% of the study area, respectively, compared to only 20% for the paratransit fleet (Fig. 2d). However, the paratransit fleet exhibits higher charging demand density. At a 30% fleet conversion rate, 50% of its spatial coverage will require over 30 MWh daily, whereas only 20% of zones covered by private and commercial fleets show similar demands. This underscores the need for higher capacity charging and grid infrastructure to meet the energy requirements of paratransit vehicles.

We analyze the impact of EV charging demand on Nairobi’s 2017 baseline electricity consumption. Our findings show that a modest 5% penetration for EVs could trigger a notable 2% rise in daily aggregate demand, with about 65% of all EVs requiring daily charging. As penetration increases to 30%, daily aggregate demand surges by 11%, reaching up to 24% at 100% EV penetration. These findings align with previous studies, which observed 5–20% aggregate demand increases under various EV adoption scenarios27,38. While private EVs significantly increase the load due to their large fleet sizes, paratransit EVs also play a substantial role due to their high-power consumption.

Beyond aggregate demand impact, we explore how EV charging influences the load profile, particularly its effect on peak demand and the load factor—a key metric for energy utilization efficiency. A higher load factor indicates more consistent electricity usage and better load management, enhancing power generation and distribution efficiency. In Nairobi, electricity consumption typically peaks in the evening (7–10 pm), and is lowest in the early morning (1–6 am). Uncoordinated in-home private EV charging would significantly exacerbate peak demand, contributing more than twice the load compared to other EV fleets, thereby reducing the load factor. Conversely, uncoordinated charging demand from the commercial and paratransit EV fleets would increase electricity consumption during daytime off-peak hours (Fig. 3). Notably, the paratransit EV fleet’s daytime load increases result in consistently higher load factors across different EV adoption rates (Fig. 4).

Transformer loading with simulated electric vehicle charging demand for individual fleets.

(a) Distribution of estimated transformer age (b) Remaining operational lifespan of transformers after five years with 5% EV penetration compared to the baseline (c) Remaining operational lifespan of transformers after ten years with 30% EV penetration compared to the baseline.

We examine the potential impact of EV adoption on Nairobi’s electric power distribution transformers, focusing on about 8,400 distribution substations serving approximately 1.6 million households, small commercial customers, and select industrial customers. Typically operating efficiently and safely within 30–125% of their rated capacity, transformers risk overheating, reduced lifespan, and damage when overloaded. Unmanaged EV charging could lead to transformer and feeder overloads, necessitating costly replacements and substation upgrades23,27.

We analyze the change in the proportion of overloaded transformers (125 − 150% loading) and critically overloaded transformers (over 150% loading) due to unmanaged EV charging across various fleet types and penetration levels. We also quantify the duration of overloading events, which significantly affects transformer lifespan. Initially, around 92% of transformers operate below 90% of their rated capacity, while 8% handle heavy loading ( 90 − 125%). With just 5% EV penetration, around 1% of transformers, mostly below 25 kVA, become overloaded.

Achieving the IEA Stated Policy Scenario target of 30% penetration by 2035 39 would result in daily overloading events lasting an average of 8.5 h for about 5% of transformers, including larger ones up to 500 kVA (Fig. 5). Private EVs, if left unmanaged, would significantly contribute to transformer overloading, affecting about 7% of transformers with daily overloading events lasting around 6 h on average. The substantial power consumption of paratransit EVs could also necessitate early replacement of 1 to 5% of transformers unless dedicated transformers are procured for these vehicles.

Impact of a smart charging strategy on Nairobi’s load profile.

We estimate the cost of premature transformer replacements due to EV charging demand. While our model does not comprehensively assess all factors affecting transformer lifespan, it provides valuable insights. Under normal conditions (80% average loading), transformers have an estimated lifespan of 30 years40, but actual longevity depends on factors like temperature variations, material quality, and maintenance41,42,43,44.

Most transformers in our study area have been operational for 10 to 16 years (Fig. 6a). We analyze the remaining lifespan of each transformer after five and ten years without EV integration, incorporating a probabilistic assessment of transformer loss-of-life correlated with daily loading. Under current conditions, without the added load from EV charging, we estimate that approximately 3% of transformers would require replacement within five years, and around 17% would need replacement within ten years. However, with a 5% penetration of EVs, the proportion of transformers needing replacement within five years could increase to approximately 6% (Fig. 6b). If EV penetration rises to 30%, this figure is projected to reach 20% within ten years (Fig. 6c).

(a) Impact of managed charging of a proportion of the electric fleet on the system load factor. Notes: The percent change in load factor at a zero percent level of coordinated charging represents the decrease (negative change) or increase (positive change) in system load factor due to uncoordinated EV charging. The horizontal dotted line represents no change in the system load factor due to EV charging load. (b) Value of peak demand mitigation by managed charging of a proportion of the electric fleet equivalent to the avoided cost of installing a Battery Energy Storage System.

Transformer replacements and upgrades can incur substantial costs, with an average unit cost of $121.56 per kVA (see Methods). By the five-year milestone, additional expenditures of $300,000 to $6.5 million may be necessary for transformer replacements and upgrades to accommodate EV charging demand, depending on penetration rates (Fig. 7a). Meeting Kenya’s 5% EV adoption target would require about $300,000 within five years, while meeting IEA’s projected EV stock of 30% of all vehicles in the Stated Policies Scenario39 would necessitate around $1.5 million. Fleet-specific scenarios indicate lower replacement costs for transformers serving paratransit EV fleets and higher costs for those serving private EVs.

(a) Additional expenditure in transformer upgrades and replacements at the 5-year mark due to additional loading from unmanaged electric vehicle demand; (b) Cost savings from a simple managed charging strategy.

Meeting peak electricity demand often requires using less efficient, more expensive generation sources, driving up consumer costs. This strains the grid, making it less resilient to unexpected disruptions or extreme weather events45. Simultaneous EV charging during peak hours exacerbates this strain. Thus, peak demand management, which curtails energy use during peak load periods while integrating EVs, offers opportunities to optimize the grid, cut infrastructure costs, and promote sustainability. Our study assesses the benefits of integrating smart charging strategies into future EV adoption plans in Nairobi.

We examine the impact of a simple smart EV charging strategy, delaying EV charging from peak hours (6 pm to 10 pm) to later slots, on Nairobi’s grid load factor and quantify cost savings from avoiding utility-scale Battery Energy Storage System (BESS) requirements (Fig. 4). Improving the load factor can enhance a utility’s operating efficiency, reduce energy waste, and lower electricity costs. EV adoption coupled with demand-side strategies, such as time-of-use pricing, smart charging infrastructure, or dynamic load balancing, can shift charging demand to off-peak hours.

In a scenario where half of all private vehicles are EVs, coordinating 20% of EV charging achieves an equivalent impact on the grid’s load factor as a 10 MW BESS costing US $11 million. Coordinating 50% of EV charging replicates the load factor improvement of a 30 MW BESS costing US $32 million. Coordinating charging among commercial and paratransit vehicles also positively impacts load factor, although these charging loads do not reduce load factor without managed charging. Grid upgrades may be necessary to accommodate rising EV demand without managed charging or a stationary BESS.

Demand-side strategies for EV charging can also reduce the need for costly transformer replacements due to overloading. We simulated managed charging at the transformer level, aiming to defer charging during peak demand. Savings in premature transformer replacements could range from $100,000 (5% EV penetration) to $600,000 (100% penetration) for private EVs. Due to significant off-peak charging, managing commercial EV charging could yield smaller savings, $20,000 to $100,000 (Fig. 7b). However, proper charging infrastructure siting is vital to meet commercial EVs’ daytime charging needs. Paratransit EVs, primarily operating during demand peaks and charging during daytime off-peak hours, show no cost savings from smart charging. Overall, utilities could save up to $610,000 in premature transformer replacements and upgrades with EV adoption planning accompanied by demand management strategies to shift peak time charging to off-peak hours.

EV consumer preferences are multifaceted, influenced by factors like price, range, technology, and cultural barriers. We hypothesize that Nairobi EV drivers’ response to range anxiety - the fear of running out of charge before reaching their destination - will be complex, similar to how Kenyan drivers often make trips to the fuel station when their tank is nearly empty and only partially refuel based on available cash. Yet, passenger vehicle owners in Kenya express skepticism about EVs meeting their mobility needs, with range anxiety as a primary barrier to adoption46.

As a case study, we analyze how the range anxiety of private EV drivers could influence charging patterns, and overall power system efficiency, highlighting the importance of understanding consumer behavior in EV planning. We simulate low-range and high-range anxiety scenarios using a probabilistic charging decision model. In the low-range anxiety scenario, drivers are less anxious, with a minimal probability of charging when the vehicle’s charge exceeds 30% and a high likelihood of disconnecting once it surpasses 70%. Conversely, in the high-range anxiety scenario, drivers exhibit heightened anxiety, with a higher probability of charging when the charge falls below 45% and a lower likelihood of disconnecting below 80%. We analyze the coincidence factor for private EVs at a 10% penetration rate, which measures the extent to which simultaneous charging events overlap by comparing the peak load from concurrent EV charging to the sum of the individual peak loads if each vehicle were charging independently. Our analysis reveals that, on average, a higher percentage of the private fleet will choose to charge on any given day under the high-range anxiety scenario compared to the low-range anxiety scenario (Fig. 8a).

Comparison of implications of the high and low driver range anxiety scenarios on charging patterns at a 10% private EV fleet penetration rate. (a) Percentage of the EV fleet that connects to charge in each simulation day. (b) Coincidence factor given by the total daily peak load divided by the sum of the daily peak load of the individual EVs. (c) Peak EV charging load for each simulation day.

Additionally, individuals with significant range anxiety exhibit a higher daily coincidence factor, potentially leading to peak charging demands up to two and a half times greater than those in the low-range anxiety scenario (Fig. 8b and c). This variation in charging behavior translates to a 1% difference in the power system’s load factor, equivalent to a 5 MW 4-hour BESS costing approximately US $5 million. These findings underscore the importance of considering consumer behavior when planning EV adoption, particularly for private fleets.

In this study, we analyze empirically informed movement and charging patterns of private, commercial, and paratransit EV fleets. Our findings highlight the implications of EV adoption for power system efficiency, distribution transformer overloading, grid reinforcement costs, and demand management strategies, including their susceptibility to consumer behavior influences.

Our results suggest that while private and commercial EV fleets can handle multiple trips on a single charge, only about 15% of daily trips of paratransit e-matatus and e-buses can be covered on a single charge, necessitating frequent recharging in 20% of the study area.

Regarding power system efficiency, our findings suggest that EV adoption in Nairobi could moderately increase electricity demand, using excess generation capacity47. However, unmanaged charging, especially from private and commercial EVs, could strain the power system, causing significant demand peaks during peak hours. In contrast, paratransit EVs, which primarily charge during off-peak hours, present a more favorable scenario without complex demand management.

Unmanaged EV charging risks overloading distribution transformers, leading to costly replacements and upgrades. Our study estimates substantial financial burdens for utilities due to overloading events, highlighting the need for demand-side management strategies. Implementing smart charging strategies could shift peak-hour demand to off-peak periods, enhancing system efficiency and reducing transformer replacement costs by 15–40%, particularly for private EV home charging. While commercial EV charging also benefits from smart management, paratransit EVs show limited advantages.

Consumer behavior, especially range anxiety, can significantly influence system efficiency and demand management. High range anxiety leads to more frequent charging, potentially doubling daily peak demand compared to low range anxiety. In Nairobi, drivers often maintain near-empty gas tanks, fueling as needed. If this behavior translates to EV low range anxiety, drivers may have less frequent charging events than high range anxiety drivers, potentially halving the peak demand. These results underscore the importance of a managed fleet conversion, and data collection on behavior patterns for system design.

Kenya’s National Transport and Safety Authority reports a 285% increase to 1,350 registered EVs over the past five years, driven by supportive government policies48 and investments49. Our study validates plans to increase electric buses in Nairobi50 by demonstrating potential system efficiency benefits from paratransit EV adoption. However, scaling up infrastructure to support electric buses and matatus remains challenging. Utilities will need to invest in distribution transformer upgrades, or electric paratransit companies will need dedicated transformers. These considerations also apply to neighboring countries, like Uganda, Rwanda, and Egypt, which plan to introduce electric buses in their major cities51,52.

Policies and incentives to encourage broad private EV adoption, such as Kenya’s reduced excise duty for electric vehicle imports from 20 to 10%, and investments in home charging infrastructure, should also support utilities in designing efficient demand-side management programs to steer private EV owners toward off-peak charging, therefore improving power system efficiency and avoiding premature transformer replacements. Commercial companies, like Kenya Power53, looking to electrify their fleets will also have to manage charging to avoid negatively impacting power system efficiency. Coordinated efforts between utilities, governments, and stakeholders are essential for strategic grid upgrades and efficient charging infrastructure placement to support a growing EV industry.

Our study has several limitations. Primarily, we lack empirical data necessary to validate our stochastic models that predict EV movement, charging patterns, and consumer charging behavior. Further, we rely on average statistics from Nairobi trip surveys, whereas GPS-tracked driving data would provide a more accurate assessment of the impacts of EV charging on the power system. Moreover, our model assumes uniform vehicle movements across various fleets, and equates EV movements with those of traditional vehicles, overlooking potential shifts in driving patterns that may arise with vehicle electrification. Although we consider consumer charging decisions, our model does not account for choices related to workplace charging access for private EVs or responses to time-of-use electricity prices. Additionally, our analysis does not factor in road conditions that affect vehicle energy consumption, nor does it account for the multiple confounding factors that contribute to transformer aging and failure. As more data on transportation, EVs, consumer behavior, and power systems become available in African cities, future work incorporating real-world data could significantly improve the accuracy and reliability of our findings.

Future studies could extend our approach to examine the upstream impacts of EV charging on the power system, including the limitations of low and medium voltage feeders and the transmission network to accommodate EV charging demand, contingent on the availability of comprehensive power system data. With the availability of real-world EV charging and movement data in Nairobi, future studies could more accurately assess consumer charging behaviors and their implications for EV batteries and the power system. Additionally, future research could integrate a pricing incentive model to shift EV charging from high-price to low-price hours, and also examine the cost-benefit trade-offs associated with demand management strategies and grid infrastructure enhancements. Though unconventional and unregulated, the use of motorcycles (colloquially called bodabodas) for transport is growing in popularity in many African cities54 but remains understudied. In Kenya, 74% of the current EV stock comprises electric two- and three-wheelers, yet little information exists on their urban mobility patterns. Future studies could explore the impact of this growing fleet on power systems. Our immediate focus in expanding this study will be on investigating the implications of EV adoption on the country’s renewable energy utilization by developing a high-resolution generation dispatch model and exploring environmental and economic policy scenarios under various EV adoption scenarios.

While EV adoption holds promise for sustainable mobility in Kenya and beyond, careful planning and demand-side management are imperative. Addressing consumer behavior, managing peak charging demand, and upgrading infrastructure are crucial steps for a smooth transition to electric mobility in Nairobi and other African cities.

This study is the first of its kind for African cities, exploring urban electric mobility transitions through high-resolution modeling. The implications of our findings are crucial for electric utilities, municipal transport planners, regulators, and policymakers in delivering successful electric mobility transitions across African cities. Practically, stakeholders like Kenya Power and the Energy Regulatory Commission of Kenya can immediately leverage our results. Our methods are replicable for other cities across the continent, aiding in efficient and effective planning of African energy and mobility transitions.

Our study takes a stochastic approach to understand and model the dynamics of EV usage in Nairobi (Fig. 1). We model spatial vehicle ownership in Nairobi at a high granularity. We simulate the daily movement patterns of various vehicle fleets using a Monte Carlo Simulation method, considering factors like when people arrive and leave, average commute times, and the distances vehicles cover. Monte Carlo Simulation is a technique that uses random sampling to model complex processes, providing numerical estimates by simulating multiple possible outcomes. We model the charging behavior of EVs by determining if and at what battery state of charge they choose to begin and complete charging. To achieve this, we apply a stochastic Markov chain charging behavioral algorithm adapted from a previous study55. As a result, we model the daily charging events for different vehicle fleets under various conditions. Building on this model, we run simulations to forecast the charging demand for different vehicle fleets for each spatial unit of analysis over ten days at a minute resolution. The objective of our model is to maximize efficiency—we optimize the number of charging points to minimize wait times. Finally, we analyze the potential impacts of different penetration levels of EVs on Nairobi’s power infrastructure and peak electricity demand. Data inputs to our model framework include information on travel behavior from Nairobi person trip and transport surveys, vehicle registrations from the Kenya Revenue Authority, vehicle movement data from Uber, and electricity demand and grid infrastructure data from the electric utility, Kenya Power (Supplementary Note 1). Next, we describe our modeling approach in detail.

Kenya has approximately 3,280,934 registered vehicles, comprising 70% passenger vehicles, 14% light-duty commercial vehicles, 11% heavy-duty commercial vehicles, and 5% paratransit vehicles. Using Nairobi’s Gross County Product (GCP) as a proportion of Kenya’s total economic activity56, we estimate that Nairobi has about 625,000 privately owned vehicles, 193,000 light-duty commercial vehicles, 11,000 paratransit minivans (matatus), and 600 paratransit buses.

To estimate EV owners in 400 different zones within our study area defined by the Uber Movement Project, we first assess vehicle ownership based on household vehicle ownership rates, household sizes, and total population data in each zone. Prior studies model vehicle ownership in relation to income with an S-shaped curve57,58,59. Income and wealth positively correlate with electricity consumption in Nairobi, as validated by the correlation between constituency-level electricity consumption and average wealth indices from the Demographic Health Survey60. ​​With granular electricity consumption data but only constituency-level wealth data, we therefore use the average electricity consumption of each zone as a wealth proxy to construct granular wealth indices. ​​We fit a log-logistic function (Eq. 1) to the zone wealth indices and vehicle ownership rates.

where V is the vehicle ownership rate, Vmax is the saturation level of vehicle ownership, Y is the wealth index, and l and s are two positive parameters defining the shape of the curve. We adjust these parameters to match the model’s cumulative vehicle count with Nairobi’s total vehicle count. We use a similar function to estimate the zone-level distribution of light-duty commercial vehicles (LDCVs), hypothesizing that LDCV numbers correlate with commercial activity, proxied by small commercial customers’ electricity consumption. The spatial distribution of private and commercial vehicle fleets is shown in Supplementary Fig. 1.

The vehicle curb weight significantly affects EV battery capacity and energy consumption, which in turn determines the vehicle’s range61. We constructed a probabilistic model to randomly assign EV specifications from a choice of 87 EV models (Supplementary Table 2) to our simulated vehicles based on the distribution of tare weights of Kenya’s current registered vehicle stock. The mean curb weight for registered private vehicles is approximately 1,400 kg (standard deviation: 400 kg) and for light-duty commercial vehicles, about 1,700 kg (standard deviation: 400 kg). A Monte Carlo simulation uses the curb weight distributions to randomly assign each EV owner technical parameters such as range, battery capacity, energy consumption, and charge power. For each scenario, the number of EV owners sampled for each fleet type in each zone matches the scenario’s EV adoption rate (Supplementary Table 3).

We addressed the lack of granular transportation data in Nairobi by stochastically generating synthetic driving patterns. We used probability distributions of departure and commute times from travel surveys (Supplementary Fig. 2), and probability distributions of travel times, speeds and distances from mobility data to construct spatial Origin-Destination (OD) pairs for each simulated vehicle, subsequently estimating the arrival times, vehicle kilometers traveled (VKTs), and states of charge (SOCs) for each EV tailored to the specific fleet.

Trip destinations (workplace, school or other) were determined stochastically for private vehicle scenarios. Departure times and commute durations were also sampled stochastically. Destination zones were identified by sampling hourly inter-zone travel time probabilities, reflecting city traffic and commercial activity levels. The probability of traveling to a destination zone is given by jointly maximizing the probability matching the stochastically drawn commute time and travel time to a destination zone at hour, h, and the probability of high commercial activity in a destination zone. Once the destination zone is determined, we estimate the travel speeds, which we use as inputs to the EV energy demand estimation model. Return trips home were simulated by sampling departure times from the home trip probability distribution. We conduct a simple validation of our model’s trip destinations against the results of a trip generation and attraction model developed for Nairobi from a previous effort (Supplementary Fig. 3).

For commercial vehicle scenarios, daily VKTs were determined stochastically. The number of stops per trip was modeled as a linear function (Eq. 2) of VKT/day of the VKT/day fitted to data from a 2017 pilot study of 4 electric LDCVs in Nairobi 38.

The origin zones of electric LDCVs were stochastically determined based on commercial activity levels in each zone, with daily distances randomly drawn from the probability distribution of daily commercial vehicle kilometers traveled obtained from a 2015 Nairobi travel survey. Average trip distances between stops were determined stochastically, ensuring their sum equaled the total daily distance. Stop durations, ranging from 5 to 120 min, were sampled uniformly. Initial departure times were sampled from the “other” destination probability function (Supplementary Fig. 2). Destination zones for each stop were then determined by sampling hourly inter-zone travel time probabilities and a model of the relationship between the daily trip distance and the number of stops constructed based on data from the 2015 electric LDCV pilot study (Supplementary Fig. 4).

For paratransit vehicle scenarios, charging was assumed to occur only at outbound transit terminals outside the central business district. The spatial locations of outbound transit terminals and the paratransit routes are shown in Supplementary Fig. 5. Paratransit vehicles usually operate between 6:00 am and 11:00 pm. 6:00 am − 9:00 am and 4:00 pm to 8:00 pm are off-peak operating hours. We model departures on each route at the beginning of the day starting at 6.00 am following a gamma distribution of a Poisson process (Eq. 3).

where f(x) is the waiting time, t, density function until the nth EV departs, with an average departure rate of l. During peak hours, vehicles depart every 2 min. We assume a 15 min downtime at the terminals to pick up passengers. Travel durations on route segments were determined by sampling hourly probability distributions of speeds and segment lengths.

For the two-wheeler vehicles, we used information on the modal share of two-wheelers within Nairobi from the person trip survey to model the spatial distribution of privately owned motorcycles relative to four-wheeled vehicles (Supplementary Fig. 6). For commercial motorcycles, we used a similar model to that for commercial four-wheeled vehicles, based on commercial activity levels in each zone.

To estimate the energy demand of each EV trip, we use a drivetrain model based on the relationship between energy consumption (kWh/km), average travel speed (km/h), and trip distance (Eq. 4).

where Etrip is the energy consumed by the EV during the trip in kWh, dtrip is the route distance and f (Vavg) is the energy consumed per km (kWh/km) when the EV is traveling at speed Vavg (km/h). The function f (Vavg) depends on the EV’s battery type, resulting in different model energy-speed curves.

We leverage algorithms for energy consumption and speed based on publicly available lab dynamometer tests for eight commonly sold EVs62. These tests provided the average energy-speed curve represented by Eq. 5:

Energy consumption values in the technical specifications of each EV are typically given for an assumed speed of 110 km/h. Using these values and the average energy-speed curve, we generated energy consumption-speed curves for each of the 87 EV models (Supplementary Fig. 7).

We develop a user charging decision model using a stochastic Markov chain algorithm56. The model inputs are logistic functions that define and adjust driver behavior parameters, representing the probability of connecting to and disconnecting from charging at various states of charge (SOC). These functions are.

Here, pi is the probability of disconnecting when charging and qi is the probability of connecting to charge based on state of charge, i. kp and kq control the gradient changes in the logistic functions and reflect range anxiety levels. At the same time, xp and xq are the central points of the gradient and depend on the state of charge. For drivers with high range anxiety, xp and/or xq are larger, along with a larger kp/kq, indicating a higher sensitivity to SOC changes. Conversely, for low range anxiety, the values are smaller, indicating less sensitivity. For the low range anxiety scenario, we define xp = 60, xq = 15, kp = 0.3 and kq = 0.35 such that there is almost no chance of choosing to charge if SOC is above 30%, and nearly a 100% chance of disconnecting if SOC is above 70%. For the high range anxiety scenario, we define xp = 90, xq = 50, kp = 0.5 and kq = 0.5 such that there is nearly a 100% chance of choosing to charge if SOC is below 45%, and almost no chance of disconnecting if SOC is below 80% (Supplementary Fig. 8).

We simulate EV movement patterns over 10 days, creating daily charging profiles for each EV at a 1-minute granularity. We model charging for the private EV fleet as Level 2 private home charging, the commercial EV fleet as Level 2 workplace charging, and the paratransit EV fleet as public fast charging. Level 2 charging can operate between 6.2 and 19.2 kW. In this study, we assume Level 2 charging operates at 11 kW.

The initial battery state of charge (SOC) at the start of the simulation is drawn from a normal distribution with a mean of 0.8 and a standard deviation of 0.04. The SOC at the beginning of subsequent days is based on the previous day’s charging decisions. For each trip, the arrival SOC is calculated using the departure SOC (SOCdep), the energy consumed during the trip (Etrip), and the EV’s battery capacity (EVcap ) as follows:

We then schedule charging considering the simulated arrival/departure times, user charging decisions, and EV charging requirements of all EVs. Each EV begins charging instantaneously at either its rated power or the maximum power allowed by the Level 2 charger, whichever is lower, and continues until the disconnect decision is made. This strategy optimizes the number of charging points required to prevent queuing. In all baseline simulations, unmanaged charging is assumed, meaning each EV starts charging as soon as it is connected to the grid upon arrival in its destination. We obtain an average charging profile for each vehicle and sum these to create a cumulative fleet charging demand profile for each zone. We evenly distribute the cumulative EV load across all transformers within the zone to estimate the EV charging load on each transformer due to the lack of spatially granular vehicle movement data. However, we acknowledge that this approach may underestimate the average transformer loading.

We model the expected transformer loss of life as a function of daily transformer loading, using data from a previous study on the impact of loading on transformer lifespan63. The resulting model for daily transformer loss-of-life (hours per day), f (LoLtx, load), is:

where LT is the percentage transformer loading. To estimate transformer loss of life over a year, we assume transformer loading follows a normal distribution with a mean, µk and a standard deviation, sk between 5% and 20% of the mean40. We account for overload duration, assuming a maximum transformer temperature of 130 °C, the industry standard limit for oil-immersed transformers44. The function for additional daily transformer loss-of-life due to overload, f(LoLtx, duration), is:

where Do, tx is the daily overload duration in hours. Using transformer commission dates and assuming a 30-year lifespan, we calculate the remaining lifetime, Lrem, at the time of this study. The transformer loss in years, Lprem is determined as:

We develop a distribution transformer replacement cost function using data from the Kenya Rural Electrification and Renewable Energy Corporation’s 2021 Procurement Database64, applying a 2021 exchange rate of 110 Kenyan shillings per USD. The unit cost f (ST ) in $/kVA of a transformer of size ST in kVA is:

Therefore, the transformer cost, Ctx, in $ is f (ST ) * ST. The cost of premature transformer replacement Ctx, rep in year k is43:

where Ctx is the cost of the transformer in $, Nexp is the expected transformer life in years, and the transformer loss in years, Lprem is also equal to Nexp − k.

Although we have data on transformer capacities and loading, we lack transformer load profiles, which are essential for assessing the impact of EV demand on transformer loading. To address this, we map each residential, commercial, and industrial electricity customer to their respective distribution transformer using customer and transformer location data. We assume that residential and small commercial customers in Nairobi follow a demand profile similar to the aggregate demand profile in Nairobi. We use smart meter data from 400 customers to develop an average demand profile for industrial customers.

We normalize both the aggregate Nairobi demand profile and the industrial demand profile. We subtract the normalized industrial profile from the normalized aggregate Nairobi profile to derive a final residential and small commercial customer profile. We estimate the total load profile for each transformer by weighting the normalized residential and small commercial profile according to the number of customers connected to the transformer and their average daily consumption. We then add the industrial customer demand profiles to the relevant transformer profiles. Baseline transformer loading, Txi, load is calculated as follows:

where Txi, peak is the peak demand of transformer i in kW and Txi, max is the maximum transformer capacity in kVA and pf is the average power factor. After adding the aggregate EV charging demand profile to the baseline transformer load profile, we recalculate the revised transformer loading values for each transformer.

Most of the data that supports the findings of this study are publicly available and sources are provided in Supplementary Note 1 of this paper. Restrictions apply to the availability of the detailed Kenya vehicle registration data, and the electricity consumption, generation, and transformer data, which were used under license for the current study and so are not publicly available. Additional data related to this paper may be requested from the corresponding author ([email protected]).

Boehm, S. et al. State of Climate Action 2023. https://doi.org/10.46830/wrirpt.23.00010 (2023).

International Energy Agency. Global EV Outlook 2023. https://iea.blob.core.windows.net/assets/dacf14d2-eabc-498a-8263-9f97fd5dc327/GEVO2023.pdf (2023).

Ramalope, D. et al. Renewable energy transition in Sub-Saharan Africa. https://climateanalytics.org/media/renewable_energy_transition_in_sub-saharan_africa.pdf (2022).

International Energy Agency. Africa Energy Outlook 2019. https://iea.blob.core.windows.net/assets/2f7b6170-d616-4dd7-a7ca-a65a3a332fc1/Africa_Energy_Outlook_2019.pdf (2019).

International Energy Agency. Global EV Policy Explorer. https://www.iea.org/data-and-statistics/data-tools/global-ev-policy-explorer (2024).

Ministry of Energy. Kenya National Energy Efficiency and Conservation Strategy. https://unepccc.org/wp-content/uploads/2020/09/kenya-national-energy-efficiency-and-conservation-strategy-2020-1.pdf (2020).

Energy Sector Management Assistance Program. Utility Performance. Kenya Power and Lighting Company (KPLC) (2024). https://utilityperformance.energydata.info/utilities/KPLC

Keskes, T., Li, Y., Jensen, H., Rytter, S. & Krzysztof, A. Electric Mobility and Power Systems: Impacts and Mitigation Strategies in Developing Countries. (2023). https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099050123125542493/p1746590c032c506708d4905fad210b582c

Falchetta, G. An Analysis of Paratransit Services in African Cities Thanks to Digitaltransport4africa’s GTFS Open Data. Digital Transport for Africa (2021). https://digitaltransport4africa.org/an-analysis-of-paratransit-services-in-african-cities-thanks-to-digitaltransport4africas-gtfs-open-data/

Jennings, G. & Behrens, R. The case for investing in paratransit: Strategies for regulation and reform. https://www.mobiliseyourcity.net/sites/default/files/2021-07/Investing%20in%20Paratransit%20-%20Jennings%26Behrens-August2017.pdf (2017).

Randall, L. et al. Active travel and paratransit use in African cities: mixed-method systematic review and meta-ethnography. J. Transp. Health. 28, 101558 (2023).

Article PubMed PubMed Central Google Scholar

Arias, M. B., Kim, M. & Bae, S. Prediction of electric vehicle charging-power demand in realistic urban traffic networks. Appl. Energy. 195, 738–753 (2017).

Article ADS Google Scholar

Borlaug, B. et al. Heavy-duty truck electrification and the impacts of depot charging on electricity distribution systems. Nat. Energy. 6, 673–682 (2021).

Article ADS Google Scholar

Mu, Y., Wu, J., Jenkins, N., Jia, H. & Wang, C. A spatial–temporal model for grid impact analysis of plug-in electric vehicles. Appl. Energy. 114, 456–465 (2014).

Article ADS Google Scholar

Neaimeh, M. et al. A probabilistic approach to combining smart meter and electric vehicle charging data to investigate distribution network impacts. Appl. Energy. 157, 688–698 (2015).

Article ADS Google Scholar

Needell, Z. A., McNerney, J., Chang, M. T. & Trancik, J. E. Potential for widespread electrification of personal vehicle travel in the United States. Nat. Energy. 1, 16112 (2016).

Article ADS Google Scholar

Purnell, K., Bruce, A. G. & MacGill, I. Impacts of electrifying public transit on the electricity grid, from regional to state level analysis. Appl. Energy. 307, 118272 (2022).

Article Google Scholar

Salah, F., Ilg, J. P., Flath, C. M., Basse, H. & van Dinther, C. Impact of electric vehicles on distribution substations: a Swiss case study. Appl. Energy. 137, 88–96 (2015).

Article ADS Google Scholar

Xu, Y., Çolak, S., Kara, E. C., Moura, S. J. & González, M. C. Planning for electric vehicle needs by coupling charging profiles with urban mobility. Nat. Energy. 3, 484–493 (2018).

Article ADS Google Scholar

Fachrizal, R. & Munkhammar, J. Improved photovoltaic self-consumption in residential buildings with distributed and centralized smart charging of Electric vehicles. Energies 13, 1153 (2020).

Jian, L., Yongqiang, Z. & Hyoungmi, K. The potential and economics of EV smart charging: a case study in Shanghai. Energy Policy. 119, 206–214 (2018).

Article Google Scholar

Mwasilu, F., Justo, J. J., Kim, E. K., Do, T. D. & Jung, J. W. Electric vehicles and smart grid interaction: a review on vehicle to grid and renewable energy sources integration. Renew. Sustain. Energy Rev. 34, 501–516 (2014).

Article Google Scholar

Clement-Nyns, K., Haesen, E. & Driesen, J. The impact of charging Plug-In Hybrid Electric vehicles on a residential distribution Grid. IEEE Trans. Power Syst. 25, 371–380 (2010).

Article ADS Google Scholar

Kapustin, N. O. & Grushevenko, D. A. Long-term electric vehicles outlook and their potential impact on electric grid. Energy Policy. 137, 111103 (2020).

Article Google Scholar

Leou, R. C., Su, C. L. & Lu, C. N. Stochastic analyses of Electric Vehicle charging impacts on distribution network. IEEE Trans. Power Syst. 29, 1055–1063 (2014).

Article ADS Google Scholar

Gan, W. et al. Coordinated Planning of Transportation and Electric Power Networks with the Proliferation of Electric Vehicles. IEEE Trans. Smart Grid. 11, 4005–4016 (2020).

Article Google Scholar

Muratori, M. Impact of uncoordinated plug-in electric vehicle charging on residential power demand. Nat. Energy. 3, 193–201 (2018).

Article ADS Google Scholar

Crozier, C., Morstyn, T. & McCulloch, M. The opportunity for smart charging to mitigate the impact of electric vehicles on transmission and distribution systems. Appl. Energy. 268, 114973 (2020).

Article Google Scholar

Uber Technologies, Inc. Uber Movement. https://movement.uber.com/explore/nairobi?lang=en-US (2020).

Hardman, S., Shiu, E. & Steinberger-Wilckens, R. Comparing high-end and low-end early adopters of battery electric vehicles. Transp. Res. Part. Policy Pract. 88, 40–57 (2016).

Article Google Scholar

Hardman, S., Fleming, K., Kare, E. & Ramadan, M. A perspective on equity in the transition to electric vehicle. MIT Sci. Policy Rev. 2, 46–54 (2021).

Ling, Z., Cherry, C. R. & Wen, Y. Determining the factors that influence electric vehicle adoption: A stated preference survey study in Beijing, China. Sustainability 13, 11719 (2021).

Mbandi, A. M. et al. Estimating on-road vehicle fuel economy in Africa: A case study based on an urban transport survey in Nairobi, Kenya. Energies 12, 1177 (2019).

Salon, D. & Aligula, E. M. Urban travel in Nairobi, Kenya: Analysis, insights, and opportunities. J. Transp. Geogr. 22, 65–76 (2012).

Article Google Scholar

Williams, S., White, A., Waiganjo, P., Orwa, D. & Klopp, J. The digital matatu project: using cell phones to create an open source data for Nairobi’s semi-formal bus system. J. Transp. Geogr. 49, 39–51 (2015).

Article Google Scholar

County Government of Nairobi. The Project on Integrated Urban Development Master Plan for the City of Nairobi in the Republic of Kenya. (2014). https://repository.kippra.or.ke/handle/123456789/1348

Romano, F. Drive Electric Study: 2 year study of e-Mobility. (2019).

Gilleran, M. et al. Impact of electric vehicle charging on the power demand of retail buildings. Adv. Appl. Energy. 4, 100062 (2021).

Article Google Scholar

IEA. Global EV Outlook 2024. (2024). https://www.iea.org/reports/global-ev-outlook-2024

Sen, P. K. & Pansuwan, S. Overloading and loss-of-life assessment guidelines of oil-cooled transformers. Rural Electric Power Conference. Papers Presented at the 45th Annual Conference (Cat. No.01CH37214) (2001), 1–8. https://doi.org/10.1109/REPCON.2001.949516 (2001).

Jardini, J. A., Schmidt, H. P., Tahan, C. M. V., De Oliveira, C. C. B. & Ahn, S. U. distribution transformer loss of life evaluation: a novel approach based on daily load profiles. IEEE Trans. Power Deliv. 15, 361–366 (2000).

Article Google Scholar

Montsinger, V. M. Effect of load factor on operation of power transformers by temperature. Trans. Am. Inst. Electr. Eng. 59, 632–636 (1940).

Article Google Scholar

Sen, P. K. et al. Transformer overloading and assessment of loss-of-life for liquid-filled transformers final project report. https://pserc.wisc.edu/wp-content/uploads/sites/755/2018/08/T-25_Final-Report_Feb-2011.pdf (2011).

Watson, H. Permissible loading of oil-immersed transformers and regulators. Facilities Instructions, Standards, and Techniques. https://ntrl.ntis.gov/NTRL/dashboard/searchResults/titleDetail/PB95128567.xhtml (1991).

Veldman, E., Gibescu, M., Slootweg, H. & Kling, W. L. Scenario-based modelling of future residential electricity demands and assessing their impact on distribution grids. Energy Policy. 56, 233–247 (2013). (J G.).

Article Google Scholar

Conzade, J., Engel, H., Kendall, A. & Pais, G. Power to move: Accelerating the electric transport transition in sub-Saharan Africa. McKinsey & Company. https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/power-to-move-accelerating-the-electric-transport-transition-in-sub-saharan-africa (2022).

Kuhudzai, R. J. Electric mobility, cooking, energy storage, & decarbonization to anchor kenya power’s growth plan. Clean Technica. https://cleantechnica.com/2022/12/19/electric-mobility-cooking-energy-storage-decarbonization-toanchor-kenya-powers-growth-plan/ (2023).

Kenya Power. Powering E-Mobility in Kenya. https://www.kplc.co.ke/img/full/Final%20E-MOBILITY%20CONFERENCE%20REPORT%20(DRAFT%2010).pdf (2023).

Njogu, H. Accelerating adoption of electric vehicles for sustainable transport in Nairobi. Brookings. https://www.brookings.edu/articles/accelerating-adoption-of-electric-vehicles-for-sustainable-transport-in-nairobi/ (2023).

Page, T. Electric buses are driving a silent revolution in Nairobi. CNN. https://www.cnn.com/2023/02/10/africa/nairobi-kenya-electric-buses-roam-basigo-spc-intl/index.html (2023).

Rwanda Development Board. BasiGo partners with AC Mobility (Tap&Go) to bring Electric Buses to Rwanda. https://rdb.rw/basigo-partners-with-ac-mobility-tapgo-to-bring-electric-buses-to-rwanda/ (2023).

The World Bank. Electric Buses Will Help Cairo Move Toward a Greener Future. The World Bank. https://www.worldbank.org/en/news/feature/2023/08/29/electric-buses-will-help-cairo-move-toward-a-greener-future (2023).

Mutua, J. Kenya Power buys electric fleet in shift to clean transport. Business Daily Africa. https://www.businessdailyafrica.com/bd/corporate/companies/kenya-power-buys-electric-fleet-in-shift-to-clean-transport--4212424 (2023).

Olvera, L. D., Plat, D., Pochet, P. & Maïdadi, S. Motorbike taxis in the ‘transport crisis’ of West and Central African cities. EchoGéo. https://doi.org/10.4000/echogeo.13080 (2012).

Fotouhi, Z., Hashemi, M. R., Narimani, H. & Bayram, I. S. A General Model for EV drivers’ charging behavior. IEEE Trans. Veh. Technol. 68, 7368–7382 (2019).

Article Google Scholar

Humanitarian Data Exchange. Kenya - Gross County Product (GCP) by economic activities per county. HDX. https://data.humdata.org/dataset/kenya-gross-county-product-gcp-by-economic-activities-per-county (2019).

Lescaroux, F. Car ownership in relation to income distribution and consumers’ spending decisions. J. Transp. Econ. Policy JTEP. 44, 207–230 (2010).

Google Scholar

Dargay, J. M. The effect of income on car ownership: evidence of asymmetry. Transp. Res. Part. Policy Pract. 35, 807–821 (2001).

Article Google Scholar

Owoeye, T., Olanipekun, D. B., Ogunsola, A. J. & Kutu, A. A. Energy prices, income and electricity consumption in Africa: the role of Technological Innovation. Int. J. Energy Econ. Policy. 10, 392–400 (2020).

Article Google Scholar

Rutsein, S. O. & Johnson, K. DHS comparative reports no. 6. https://dhsprogram.com/pubs/pdf/CR6/CR6.pdf (2004).

Mruzek, M., Gajdac, I., Kucera, L. & Barta, D. Analysis of parameters Influencing Electric Vehicle Range. Procedia Eng. 134, 165–174 (2016).

Article Google Scholar

Argonne Downloadable Dynamometer Database, Generated at the Argonne National Laboratory, Advanced Powertrain Research Facility (APRF), under the Funding and Guidance of the U.S. Department of Energy (DOE). (2015).

Ritonga, A. S. et al. Analysis of the effect of loading on the transformers usage time. J. Renew. Energ. Electr. Comp. Eng. 1, 79 (2021).

Article Google Scholar

REREC. Opening Minutes For Supply of Transformers. Rural Electrification Renew. Energy Corporation. https://www.rerec.co.ke/tender-notice.php (2021).

Download references

We would like to thank Kenya Power for providing the electricity consumption, generation and transformer data. We would also like to thank Jit Bhattacharya from BasiGo and Mikael Gånge from Roam for their valuable insights on electric paratransit fleetoperations.

Department of Electrical and Computer Engineering, University of Washington, 185 W Stevens Way NE, Seattle, WA, 98195, USA

June Lukuyu

World Resources Institute - Africa, 14 School Lane, Westlands, Nairobi, Kenya

Rebekah Shirley

College of Information and Computer Sciences, University of Massachusetts Amherst, 140 Governors Drive, Amherst, MA, 01003, USA

Jay Taneja

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

Study conception and design: J.L, R.S, and J.T; contribution of new analytic tools and data analysis: J.L; interpretation of results: J.L, R.S, and J.T; manuscript preparation: J.L; manuscript review: J.L, R.S, and J.T.

Correspondence to June Lukuyu.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Below is the link to the electronic supplementary material.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

Lukuyu, J., Shirley, R. & Taneja, J. Managing grid impacts from increased electric vehicle adoption in African cities. Sci Rep 14, 24320 (2024). https://doi.org/10.1038/s41598-024-75039-3

Download citation

Received: 25 June 2024

Accepted: 01 October 2024

Published: 17 October 2024

DOI: https://doi.org/10.1038/s41598-024-75039-3

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative