New Strategy Targets Medium-Sized Cities for Smarter, Greener EV Charging Infrastructure
In the race to electrify transportation and meet aggressive carbon neutrality goals, one critical gap has quietly persisted: the planning and deployment of charging infrastructure in medium-sized cities. While megacities like Beijing, Shanghai, and Shenzhen have seen rapid rollout of public EV chargers—with dense networks lining highways, shopping malls, and residential towers—smaller urban centers often fall into a planning gray zone. Too large to be treated like rural outposts, yet too small to attract the same level of investment or policy attention as tier-1 hubs, these mid-tier cities have become the silent bottlenecks in China’s broader EV transition.
A newly published study is challenging that status quo. Led by researchers Shouwen Wang and Zhaobin Zhu, the work presents a novel planning framework that doesn’t just optimize where to place chargers—but why, how much, and for whom. More importantly, it introduces a concept rarely emphasized in engineering models: carbon-reduction-equivalent benefits—a metric that translates environmental impact directly into financial logic. Far from a theoretical exercise, the model was tested in Xiling District of Yichang City, a representative medium-sized urban area in central China, and validated using IEEE-standard power system simulations. The findings are both pragmatic and provocative: when charging infrastructure is designed with regional demand potential and decarbonization value in mind, the economic case strengthens, grid impacts soften, and user satisfaction rises—not just incrementally, but systemically.
So why do medium-sized cities matter so much in the EV ecosystem? The answer lies in their unique positioning. Unlike megacities—which often face land scarcity, saturated grids, and complex stakeholder coordination—or small towns, where EV adoption remains embryonic, medium-sized cities (typically with populations between 500,000 and 3 million) sit at a strategic inflection point. They possess the administrative capacity to implement municipal EV policies, the industrial base to support local supply chains, and the spatial flexibility to experiment with innovative infrastructure layouts. Crucially, many serve as regional hubs—connecting rural hinterlands to larger urban economies—and thus act as gateways for EV adoption beyond the coastal corridors.
Yet today, charging deployment in these cities remains reactive rather than strategic. Planners often replicate templates designed for megacities—installing chargers near municipal buildings or along arterial roads—without assessing where demand will emerge, or how it will evolve. The result? Underused stations in low-traffic zones, congestion at a few overburdened hubs, and a growing mismatch between charger locations and actual driver behavior. Drivers report frustration over “charger deserts” in mixed-use neighborhoods or long queues at highway rest stops—all symptoms of a top-down, supply-driven rollout that ignores ground-level patterns.
What sets this new study apart is its fusion of two traditionally siloed perspectives: static and dynamic charging demand.
Static demand refers to predictable, location-based charging—think residential overnight charging, workplace top-ups during office hours, or mall parking-lot sessions while shopping. It’s relatively easy to model: overlay census data, land use maps, and employment zones, and you get a rough heat map of where people live, work, and spend time.
Dynamic demand, by contrast, captures the randomness of real-world mobility—longer trips, detours, unexpected errands, or last-minute range anxiety. This is where EVs behave less like parked appliances and more like moving nodes in a fluid transportation network. A driver might fully charge at home but still seek a 20-minute top-up before a weekend drive to the mountains; a delivery van may need mid-route fast charging after completing morning routes.
Most prior planning models lean heavily on one or the other. Some use static zoning—leading to overbuilding in office districts but gaps in entertainment zones. Others simulate traffic flows but ignore socioeconomic context, producing technically optimal but socially impractical sites (e.g., a fast charger placed on a high-speed bypass where no one actually stops).
Wang and Zhu’s breakthrough lies in integrating both. They construct a “charging demand potential spatial distribution index”—a composite measure that folds together five key inputs:
- Charging demand carrying capacity: a localized gauge of how many EVs a district can realistically support, based not just on grid capacity, but also policy incentives (e.g., local EV subsidies), net population inflow, and high-tech industrial activity.
- Land-driven indicators: accounting for the real cost and availability of land—especially critical in cities where commercial, residential, and industrial zones have vastly different lease rates and zoning restrictions.
- Charging station market potential: modeled using economic gravity principles—how GDP and geographic proximity influence inter-district charging spillover (e.g., shoppers from neighboring towns boosting demand in a central commercial hub).
- User regional distribution probability: derived from large-scale travel surveys (notably the U.S. National Household Travel Survey, adapted for Chinese contexts), then refined using Markov chain modeling to estimate the likelihood of trips originating from home (H), work (W), or other (O) zones—yielding a baseline probability matrix: 47.81% from home, 12.81% from work, 39.38% from other locations.
- EV driving characteristics: including daily mileage distributions, plug-in/out times, and state-of-charge (SOC) depletion rates—enabling estimation of when and how much energy drivers need.
The result is a granular, probabilistic canvas: not a rigid blueprint, but a living map of potential—a forecast of where demand is likely to concentrate over the next 5–10 years, even before EV penetration peaks.
But knowing where to build is only half the battle. The other half is why. And here, the study makes its most policy-relevant pivot: it embeds carbon-reduction-equivalent revenue directly into the economic optimization model.
Traditionally, charging station planning weighs costs (land, construction, maintenance, grid losses) against revenues (electricity sales, service fees). Environmental benefits—while lauded in press releases—are treated as intangible externalities. This new framework monetizes them.
Specifically, the authors calculate CO₂ reduction by comparing the emissions of equivalent internal combustion engine (ICE) vehicles over the same distance—using a standard differential of ~913.5 kg CO₂ per MWh avoided. Then, they factor in the prevailing carbon trading price (set at ¥0.012/kg in the study’s scenario—conservative by today’s rising market trends). This “equivalent revenue” is added to the station’s income stream, effectively turning decarbonization into a line item on the balance sheet.
Why does this matter? Because it changes the calculus for investors and municipal planners. Under a pure cost-recovery model, a new fast-charging hub in a mixed-use district might appear marginally viable—high land cost, uncertain usage. But when carbon credits are included, the net present value improves significantly. In the Yichang case study, the carbon benefit alone contributed over ¥800,000 annually to a 7-station network—enough to offset nearly 15% of annual operating expenses.
Critically, the model avoids the trap of “more is better.” Using immune algorithm-based optimization (a bio-inspired method known for avoiding local optima and premature convergence), the team generated dozens of candidate layouts with station counts ranging from 6 to 10. Then, instead of picking the one with highest revenue or lowest cost, they applied an entropy-weighted TOPSIS evaluation—a multi-criteria decision tool that ranks alternatives by proximity to an “ideal” solution across all dimensions: cost, revenue, grid impact, and user satisfaction.
The winner? Seven stations.
Not six (the “cheapest” option), not ten (the “highest coverage” setup), but seven—the balanced optimum. And remarkably, this matched the target outlined in Yichang’s own municipal EV infrastructure plan, lending real-world credibility to the method.
Digging deeper into the placement logic reveals another insight: optimal sites clustered not in the city center, but in the southwest quadrant—specifically near a hospital, a tech park, and a large shopping complex. Why? Because this zone uniquely interweaves all three activity types: people live nearby (residential density), work there (office parks), and play there (retail/entertainment). Such “trip generators” produce higher-frequency, multi-purpose travel—and thus more varied, reliable charging demand—compared to mono-functional zones. A workplace-only district sees surges at 8 a.m. and 5 p.m.; a residential area peaks overnight. But a mixed zone hums steadily throughout the day.
Moreover, shorter average trip distances in such areas reduce range anxiety, making AC Level 2 chargers more viable alongside DC fast chargers—lowering both capital cost and grid stress.
To test grid compatibility, the researchers mapped the seven proposed sites onto the IEEE 33-node distribution network—a widely accepted benchmark for power flow analysis. Using backward-forward sweep methods, they simulated voltage profiles, line losses, and loading under peak charging demand. The results were reassuring: voltage deviations remained well within ±5% limits, and no branch exceeded thermal ratings. In fact, the entropy-TOPSIS-selected plan caused less voltage fluctuation than alternatives—proof that environmental and engineering goals need not conflict.
Perhaps the most compelling finding emerges from a counterfactual simulation: What if we ignored demand potential and carbon benefits?
The study ran a side-by-side comparison—same city, same algorithm, but with a conventional cost-revenue model. The outcome was stark. Without demand-potential weighting, stations tended to cluster along major roads, neglecting high-potential interior neighborhoods. Without carbon-equivalent revenue, the model favored fewer, larger stations—maximizing throughput per unit cost but increasing user travel distance and grid congestion at single points.
When plotted on a benefit-cost curve, the “enhanced” model (with demand + carbon) consistently outperformed the baseline—especially at lower station counts. Fewer, smarter stations, guided by potential and valued for their emissions savings, delivered higher net societal benefit than more numerous but poorly sited ones.
This has profound implications for national policy. China’s “Dual Carbon” goals—peak emissions by 2030, carbon neutrality by 2060—cannot be met by power sector decarbonization alone. Transportation accounts for nearly 10% of national CO₂ emissions and is still rising. EVs are the linchpin—but only if their supporting infrastructure is designed for decarbonization, not just electrification.
Medium-sized cities, often overlooked in national EV narratives, may hold disproportionate leverage. They’re large enough to move the needle on regional emissions, yet agile enough to pilot new models. A successful rollout in Yichang could be replicated in dozens of similar cities across central and western China—Zhuzhou, Nanchong, Jiaozuo—creating a “middle-belt” of EV readiness that bridges the coastal-inland divide.
Of course, challenges remain. Data transparency is one: the model relies on municipal statistics (population flows, land prices, EV registrations) that aren’t always standardized or publicly available. Behavioral assumptions—like trip patterns derived from U.S. surveys—need local validation. And while carbon pricing is growing, its stability and scalability are still uncertain.
Still, the framework offers a robust starting point—not a rigid prescription, but a process: understand local demand anatomy, quantify environmental value, simulate trade-offs, and select the balanced optimum, not the extreme one.
For auto OEMs, this means rethinking market-entry strategies. Rather than targeting only megacities or offering generic “nationwide charger access” promises, brands could partner with medium-sized municipalities on co-branded infrastructure—positioning themselves as enablers of local sustainability. Imagine a BYD or NIO charging hub co-located with a city’s green mobility center, featuring solar canopies, battery buffers, and real-time emissions dashboards showing “CO₂ saved today: 217 kg.” That’s not just infrastructure—it’s marketing, policy, and environmental stewardship in one.
For grid operators, the message is clear: proactive engagement with urban planners is no longer optional. As EV loads shift from predictable nighttime home charging to more distributed, daytime public charging, distribution networks must adapt. Dynamic hosting capacity assessments, targeted feeder upgrades, and demand-response integration with charging networks will become essential—not as cost centers, but as value-enabling investments.
And for drivers? The promise is simpler: fewer empty stations, shorter detours, fairer pricing, and the quiet satisfaction of knowing every kilowatt-hour charged isn’t just powering a car—but actively erasing carbon from the atmosphere.
The electrification of transport is no longer a question of if, but of how well. In that regard, this study doesn’t just propose a new algorithm—it offers a new philosophy: infrastructure as an expression of place, purpose, and planetary responsibility. Medium-sized cities, long seen as the “middle children” of urban policy, may just be the ones to lead the way.
—
Shouwen Wang¹,², Zhaobin Zhu³,⁴
¹ School of Law & Public Administration, China Three Gorges University, Yichang 443002, China
² School of Economics, Huazhong University of Science and Technology, Wuhan 430074, China
³ College of Electrical Engineering & New Energy, China Three Gorges University, Yichang 443002, China
⁴ Provincial Graduate Workstation, Hubei Energy Co., Yichang 443002, China
Journal of China Three Gorges University (Natural Sciences), 2023, 45(4): 86–94
DOI: 10.13393/j.cnki.issn.1672-948X.2023.04.014