Chicago's 2050 Clean-Turn: Electrification Could Spike Grid Demand to 4 Gigawatts

By 2050, if Chicago electrifies aggressively, its cars and trucks will need so much electricity that the peak power demand could hit roughly 4 gigawatts—concentrated in a small urban core, at moments when the grid may be straining. That single number, hidden inside a sprawling simulation of the region's transport future, is the quiet earthquake this paper is trying to warn planners about. Electrification isn't just a transportation problem. It's an electricity-grid problem wearing a car's clothes.

The study, led by researchers at Argonne National Laboratory and the Chicago Metropolitan Agency for Planning (CMAP), builds the first integrated, agent-based model that lets all of these interventions interact—electrification, freight reengineering, road pricing, parking reform, and transit expansion—and watches what happens on a shared road network where cars, buses, trucks, and trains all fight for the same space (Alam et al., 2026). For a region that plans decades ahead, this is the difference between studying each policy in a sterile lab and letting them collide in a messy city.

The Science

At the heart of this work is POLARIS, a large-scale agent-based modeling framework developed at Argonne National Laboratory (Auld et al., 2016). Think of it as a city built inside a computer, populated not by averages but by individual synthetic people and households reconstructed from Census data, Public Use Microdata Areas, and the American Community Survey. Each of these agents wakes up, generates a sequence of daily activities, chooses where to go, how to get there, and when to leave—using behavioral models that were calibrated against real 2019 conditions, including transit boardings, mode shares, departure-time distributions, and trip-length profiles.

What makes POLARIS unusual is not just that it simulates travelers, but that it simulates firms too. Using the CRISTAL model (Stinson and Mohammadian, 2022), the framework generates a synthetic landscape of establishments and supply-chain relationships, from which freight trips emerge endogenously rather than being injected as fixed origin-destination matrices. This means truck traffic can respond to congestion, congestion can slow buses, and slow buses can push passengers onto cars—feedback loops that conventional decoupled models simply cannot see.

The other crucial feature is its iterative feedback structure. The simulation doesn't run once; it loops. If a simulated trip takes far longer than expected, agents revise their routing or mode choices in later iterations. Late arrivals spill into schedule adjustments, shortened activities, or cancellations. The system settles into a dynamic network equilibrium where choices reflect the evolving state of the network itself. This is how the model captures emergent, adaptive behavior—the way real people actually navigate an imperfect city.

To project to 2050, the team needed a network that didn't exist yet. They fused two sources: a 2019 POLARIS network with fine-grained detail inside Chicago, and CMAP's forecast Emme network with higher resolution in the suburbs (

). The fusion was semi-automated—most links and nodes merged algorithmically, but conflicts affecting connectivity required manual edits. The result was a hybrid 2050 network combining the urban core's specificity with suburban coverage.

The population itself shifts dramatically. By 2050, households in the region grow from 4.3 million to 5.3 million, and population from 10.4 million to 12.4 million (

). Growth is uneven: the Central Business District and Inner City see the steepest relative increases, roughly 2.65 times and 1.7 times in persons and households respectively (

).

Against this baseline, the team modeled nine scenarios in six thematic groups. The business-as-usual (BAU) scenario carries current trends forward with 15.2% telecommuting and modest electrification—20.5% of light-duty and 6.4% of medium- and heavy-duty trucks. Two electrification scenarios test moderate (55.1% LD, 10.1% MDHD) and high (94.0% LD, 50.1% MDHD) adoption. A freight bundle combines off-hour delivery, dedicated truck lanes, dock shore-powering, and rail diversion for trips over 500 miles. Three pricing scenarios test an expressway access fee, a flat motor fuel tax replacement, and a "smart" road user charge with spatiotemporally variable rates that exempt low-income households and cap their rate at $0.05 per mile (

). Finally, a transit scenario roughly doubles revenue trips through rail extensions and bus improvements.

Home Charging Energy Grows More Than Fourfold Under High Electrification

Label	Value
BAU	6.06
Medium Elec.	16.28
High Elec.	27.66

Home charging energy (Personal–Home) by scenario, from the paper's charging events table. Values in GWh.

What They Found

The headline energy result is stark. Under high electrification, total fuel mass—the literal weight of gasoline and diesel burned—declines by 68%. But that fuel is replaced by electricity, and the charging energy demand grows roughly 4 to 8 times above BAU, with peak power demand approaching 4 gigawatts concentrated in the urban core (Alam et al., 2026).

The charging data reveal where the grid pressure comes from. In BAU, nearly all charging happens at home: 896,604 home charging events consuming 6.06 GWh, versus just 39,384 public charging events. Under high electrification, home charging balloons to over 4.1 million events and 27.66 GWh, while depot charging for freight jumps from 14,484 events and 2.08 GWh to 105,000 events and 15.90 GWh. The charging profile is dominated by personal vehicles at home and freight at depots—not the public fast chargers that dominate public discourse (

Freight Depot Charging Leaps Sevenfold in the High-Electrification Scenario

Label	Value
BAU	2.08
Medium Elec.	3.28
High Elec.	15.9

Freight depot charging energy by scenario, from the paper's charging events table. Values in GWh.

). The average depot charging session draws 151.4 kWh over roughly 26 minutes; a dock session spreads just 18.4 kWh across 57 minutes while trucks load and unload.

Remarkably, freight electrification turns out to be a charging strategy hiding in plain sight. Freight dock charging—plugging trucks in while they idle at loading bays—consumes comparatively little energy but spreads over long dwell times, essentially using "dead" loading time to charge. This is precisely the kind of infrastructure insight that a decoupled model would miss.

On travel behavior, the results are more sobering for policy optimism. Most interventions produce only slight deviations from BAU in aggregate vehicle-miles traveled. The study attributes this to "the inherent inertia of automobile dependence even under significant policy reform." The smart pricing scenario stands out as the most effective lever for reducing auto driver passenger-miles and passenger-hours, cutting trip duration by 10% relative to BAU—consistent with higher variable costs discouraging longer discretionary trips.

The freight bundle resolves what looks like a paradox. It increases the number of truck trips while reducing total freight VMT. The explanation lives at the trip level: the bundle produces shorter and slower individual trips. Off-hour delivery fragments consolidated shipments into more frequent, smaller deliveries, and the rail diversion policy reassigns long-haul movements to rail, leaving shorter drayage trips for trucks. The average freight speed drops about 8%, and medium-duty trucks see the biggest rise in trip count. The freight bundle yields the lowest average speeds—around 31 mph for medium- and heavy-duty trucks versus BAU values near 33–39 mph—reflecting the operational friction of dedicated lanes and delivery windows (Alam et al., 2026).

Transit expansion delivers the strongest modal shift. The transit scenario boosts total daily ridership from about 2.18 million to 2.58 million—an 18.4% increase over BAU—while passenger-miles rise 14.0% and passenger-hours 6.1% (Alam et al., 2026). Notably, this expansion comes on top of a BAU that already includes substantial transit improvements: several Bus Rapid Transit lines, four new Red Line stations, a Metra extension, and a 37% increase in revenue trips. The transit scenario roughly doubles revenue trips again through rail extensions and a 101% increase in overall revenue trips.

Transit Expansion Lifts Daily Ridership by 18.4% Over BAU

Label	Value
01 BAU	2,175,455
04c Pricing Smart	2,228,766
05 Transit	2,576,684

Total transit ridership (daily) from the paper's performance-indicator table. BAU 2,175,455; Pricing Smart 2,228,766 (+2.5%); Transit 2,576,684 (+18.4%).

Why This Changes Things

The most consequential finding here is not about transportation at all—it's about the electricity grid. A 4-gigawatt peak demand concentrated in the urban core is the kind of number that keeps grid operators up at night. But the study's insight is that the load is not monolithic. Different charging segments behave differently: home charging dominates in volume but is spread through evenings; depot and dock charging load freight onto the grid during business hours; public charging is a small but growing slice.

This matters because Commonwealth Edison—Illinois' largest electric utility—is a named consumer of these results, alongside CMAP's next regional transportation plan. The study offers a concrete case for targeted charger upgrades rather than blanket infrastructure spending. If most freight charging happens at depots and docks during natural dwell times, investments should target those facilities, not just public fast-charging networks. The energy table makes this visible: depot charging under high electrification delivers 15.90 GWh from just 105,000 events—more than triple the public charging load—at a manageable 151.4 kWh per session spread over ~26 minutes.

There's also a structural honesty in the travel results that policymakers should hear. Most interventions barely move aggregate VMT because automobile dependence is deeply embedded in the region's spatial and social fabric. The study doesn't spin this into failure; it treats it as a constraint. The smart pricing scenario's 10% trip-duration reduction stands out precisely because variable, congestion-responsive pricing is one of the few levers that directly changes behavior at the margin. Meanwhile, transit's 18.4% ridership gain shows that supply expansion is the dominant lever for rail and bus—echoing earlier findings that service supply, not fare policy, drives ridership response (Verbas and Mahmassani, 2015).

The freight paradox is itself a policy lesson. Reducing total freight VMT while increasing trip frequency is not a contradiction; it's a reconfiguration. Shorter, slower, more frequent deliveries relieve long-haul truck traffic at the cost of more local trips. Whether that's a net win depends on what you optimize for—total miles, urban congestion, emissions, or delivery reliability. The framework makes those trade-offs explicit rather than implicit.

And there's a fairness thread running through the pricing scenarios worth noticing. The smart pricing design exempts low-income households—those at or below 80% of regional median income—and caps their rate at $0.05 per mile regardless of location or time of day. This responds directly to evidence that cordon and distance-based pricing disproportionately hurt lower-income households unless paired with discounts or revenue recycling (Venezia, 2023). The study treats equity not as an afterthought but as a design constraint built into the pricing instrument.

What's Next

This is a simulation, not a prediction, and its authors are careful to say so. The BAU baseline already embeds substantial planned transit investment, so the transit scenario's 18.4% ridership gain is measured against an already-improved system, not a stagnant one. The electrification results assume the charging infrastructure was optimized for each scenario, which is both a strength (realistic infrastructure deployment) and a boundary (different siting choices could shift the load profile).

Several open questions remain. The study models freight dock charging and depot charging but doesn't fully resolve where the system-level grid upgrades must land to absorb that 4 GW peak—that's presumably downstream work with ComEd. The individual freight levers (off-hour delivery, dedicated lanes, rail diversion) were run as a bundle, and their separate sensitivity analyses are omitted here, so we can't isolate which intervention drives which effect. And the smart pricing scenario's equity design, while thoughtful, treats a $0.05 per-mile cap as a floor; the distributional consequences across income groups beyond that threshold aren't fully explored.

What the paper does—and what's genuinely new—is to hold all these policies in one room at the same time. Conventional regional planning too often evaluates electrification, freight reform, pricing, and transit in isolated silos, each with its own model and its own assumptions, then naively sums the results. POLARIS lets them interact: trucks slow buses, buses push riders to cars, pricing pushes long trips off the road, and all of it moves electrons through a grid that isn't ready. That integrated view is the methodological contribution that will outlast any single number in this paper.

The takeaway for the world beyond Chicago is that the clean-energy transition in transportation is a coordination problem, not a substitution problem. Swapping gasoline for electrons doesn't just change the fuel; it changes the infrastructure, the timing of demand, the economics of freight, and the geography of load. The cities that plan for that integration early—that design their grids around depot charging and their pricing around fairness and their fleets around the network they actually have—will navigate 2050 far more gracefully than those that electrify first and think second.

The 4 GW question is Chicago's question, but the structure of the answer travels anywhere a city plans a decade or two ahead. Ask not just what your vehicles will run on, but where the power comes from, when it is drawn, and who pays. Those are the questions this framework finally lets a region ask all at once.