Charge More Electric Trucks—Without Grid Upgrades

A single megawatt shifted by a few hours can be the difference between a “denied until upgrades” interconnection and a depot that’s energized this quarter.

That’s not hypothetical. In Northern California, modeling based on real feeder constraints and heavy-duty truck driving patterns shows that flexible service connections can free up enough distribution capacity to support 10 additional heavy-duty trucks per feeder in many cases—without rebuilding the local grid. Scaled across roughly 2,000 feeders, that’s 25,000–70,000 more electric trucks per year depending on duty cycle.

This matters beyond trucking. In our “AI in Agriculture: Precision Farming for Modern Growers” series, we usually talk about optimizing water, fertilizer, and field operations. But electrification is landing in agriculture fast—yard trucks, milk haulers, feed delivery, refrigerated transport, and even irrigation pumping. The grid constraints facing truck depots are the same constraints rural co-ops and investor-owned utilities face when farms add new electrified loads. The practical lesson: capacity is often available—you just need the intelligence to use it safely.

Flexible service connections: the fastest path to more charging capacity

Flexible service connections let a site connect sooner by agreeing to curtail or shift load during a small number of locally constrained hours. Instead of planning for a once-a-year “worst possible hour,” the utility and customer formalize a rule: “You can interconnect now, but during specific peak windows you’ll reduce charging so the feeder stays within safe limits.”

The traditional process is cautious for a reason. Utilities often evaluate:

• the highest feeder load hour of the year, and
• the maximum load the new site could theoretically draw

If that combination exceeds the feeder’s operating limit, the project gets pushed into the upgrade queue—often months or years—despite the fact that the “bad hour” might be rare, seasonal, or misaligned with the depot’s actual charging schedule.

My take: most interconnection delays for depots aren’t “no capacity,” they’re “no guarantees.” Flexible service connections convert uncertainty into an enforceable operating agreement. That’s why they’re so effective.

Why this is an AI problem (and an AI opportunity)

Utilities can’t manage what they can’t predict. Flexible service connections depend on knowing, with high confidence:

• when a feeder will be stressed,
• how much headroom is available, and
• how customer load can be reshaped without breaking operations.

That’s exactly where AI in utilities is proving its value: demand forecasting, feeder-level anomaly detection, and optimization-based scheduling that translates grid constraints into simple operating signals for customers (day-ahead, seasonal, or even hour-ahead).

In other words: flexible service connections aren’t just a tariff idea. They’re a data product.

What the Northern California numbers tell us (and why they’re believable)

RMI’s analysis combined:

• public feeder data (PG&E’s grid information), and
• representative charging patterns derived from telematics for regional-haul and urban delivery heavy-duty trucks.

They then ran an optimization: shift charging away from feeder peak hours in 1-hour increments within a defined “flexibility window,” while still delivering the energy trucks need by morning.

Here are the headline results that utility planners and fleet operators should care about:

• Shifting just 1 MW away from peak by several hours on the limited days required allowed most constrained feeders to support ~10 more heavy-duty trucks.
• Expanding the flexibility window from 2 hours to 6 hours increased the median additional trucks per feeder dramatically:
  • Regional trucks: from 3 to 22
  • Urban trucks: from 9 to 64
• Across PG&E territory, that scales to:
  • 8,000–41,000 extra regional trucks
  • 22,000–112,000 extra urban trucks

These aren’t “marketing numbers.” They’re engineering-adjacent outputs from constraint-based scheduling against real feeder peaks.

When the grid actually needs flexibility

In Northern California (and much of the US), the tough window is familiar: 4–9 p.m. in summer, when solar output falls and residential load rises.

A striking result from the modeling: for about 60% of feeders, a simple two-hour flexibility requirement would only be triggered during July. That means fleets would operate normally most of the year, with occasional, predictable “don’t charge hard right now” constraints during a handful of peak days.

For agriculture, this resembles irrigation demand management: you don’t stop pumping all season—you adjust during the few hours that stress the system.

Will load shifting mess up fleet (or farm) operations? Usually, no

Most heavy-duty trucks driving under ~300 miles/day are parked at the depot roughly from 4 p.m. to 5 a.m. That parking window overlaps the peak grid hours—and extends well beyond them.

RMI’s earlier work also found that, on average, California heavy-duty trucks can meet daily charging needs using less than half of their depot dwell time at 75 kW charging power (and many depots install higher power).

Put simply: there’s slack in the schedule. That slack is what flexible service connections monetize.

Managed charging is the real enabler (and it’s software-first)

If you’re picturing dispatchers manually turning chargers on and off, you’re overthinking it. The operational requirement is typically:

• set a charging completion target (e.g., “all trucks ready by 5 a.m.”),
• define a maximum site kW limit during constrained hours, and
• let an optimizer schedule which vehicles charge when.

That optimizer can be rule-based, but the best results come from AI-driven forecasting that accounts for uncertainty:

• route variability,
• arrival times,
• battery state-of-charge,
• ambient temperature impacts, and
• feeder constraints that shift by season and local load.

For fleets serving agricultural supply chains—grain hauling, dairy logistics, cold storage—this is directly applicable. The “vehicle” could also be an electrified refrigeration unit or an irrigation pump. The control logic is the same: hit an energy target by a deadline while respecting a grid limit.

What about atypical schedules? Use batteries (strategically)

Not every fleet parks overnight. Some run late shifts, some do mid-day returns, some need rapid turnarounds.

When flexibility windows conflict with operations, on-site battery storage becomes the buffer:

• charge the battery off-peak,
• discharge during constrained peak hours,
• keep trucks charging without exceeding the feeder limit.

For farms, this is also familiar: storage turns intermittent cheap power (or on-site solar) into firm capacity during the hours the utility cares about.

How utilities operationalize flexible connections: the PG&E pattern

PG&E’s FlexConnect program is the clearest example of “real life” implementation at scale. The key operational element is visibility—the utility uses grid monitoring and informs participants day-ahead when constraints are expected.

A real-world proof point cited: PepsiCo added ~20 additional electric trucks on regional-haul routes by using this approach—consistent with the modeled additional-truck ranges.

Other utilities are moving, but mostly in pilots (examples include SCE and National Grid). Some programs are trending toward seasonal limits (simpler to administer) rather than day-ahead signals (more efficient, but data-heavy).

My stance: if your utility doesn’t have feeder-level forecasting and monitoring, it will default to blunt seasonal rules. That still helps, but it leaves capacity on the table. AI-driven operations is how you get the best of both worlds: reliability and utilization.

The AI playbook: turning feeder constraints into actionable signals

Flexible service connections are a policy wrapper around an optimization workflow. Here’s a practical “AI in utilities” blueprint that works for trucking depots and rural electrification projects tied to agriculture.

1) Forecast feeder peaks at the right granularity

The most useful forecasts are not system-wide. They’re feeder-level and weather-aware, with special handling for:

• summer evening ramps,
• distributed solar variability,
• local commercial patterns, and
• electrification clusters (depots, cold storage, pumps).

Even a one-hour resolution can unlock value; finer intervals can unlock more.

2) Translate forecasts into capacity envelopes

Instead of telling customers “don’t charge,” define a simple envelope:

• Max kW from 4–9 p.m. on constrained days
• Or a daily kWh cap within a time window

This makes compliance auditable and automation-friendly.

3) Optimize customer load (and automate compliance)

For fleets and large agricultural loads, the optimizer should:

• prioritize vehicles/loads by deadline and urgency,
• allocate power fairly across chargers,
• adapt when arrivals are late or temperatures change.

This is where AI scheduling beats static rules, because it responds to real operations.

4) Measure performance and improve the model

Utilities should track:

• number of constrained events,
• customer compliance,
• avoided upgrade deferrals (months/years), and
• incremental MWh served without violations.

This feedback loop is what makes the program scalable.

Snippet-worthy definition: A flexible service connection is an interconnection agreement that swaps rare peak-hour curtailment for faster energization and avoided distribution upgrades.

What fleet operators and ag businesses should do next

If you’re planning an electric truck depot—or electrifying farm-adjacent operations—treat interconnection like a data project, not just an electrical one.

Here’s a practical checklist I’ve found works well:

1. Ask the utility about flexible interconnection options (day-ahead limits, seasonal limits, or managed charging pilots).
2. Document your true operating window (arrival/departure distributions, not just a single schedule).
3. Install charging management software that can enforce a site-level kW cap automatically.
4. Design for optional storage (even if you don’t buy it immediately): leave space, conduit, and switchgear flexibility.
5. Model two scenarios before committing to upgrades:
   • “always available power” (upgrade-heavy)
   • “flexible power + controls” (software-heavy)

For utilities and energy service providers, the lead-generating opportunity is clear: customers don’t want a lecture on feeder ratings—they want a path to energization with predictable rules.

Where this fits in precision agriculture

Precision farming is really precision operations: sensing, forecasting, and optimizing constrained resources. We apply that mindset to nitrogen, irrigation, and equipment routing. The grid is just another constrained resource.

Flexible service connections bring that same logic to electrification: use intelligence to shape demand instead of overbuilding supply. As electric trucking expands into agricultural supply chains in 2026 and beyond, the winners won’t be the teams with the biggest transformers. They’ll be the teams that can coordinate charging, storage, and grid constraints with software.

If you’re building a plan for electrified logistics—or even evaluating electrified pumping and cold storage—what would change if your utility let you connect now, as long as you agreed to stay under a feeder limit for a few dozen hours a year?