ev-fleet-planning

Overview

Guides organizations through fleet electrification decisions with engineering rigor and financial precision. The skill quantifies what it takes to transition a vehicle fleet from internal combustion to electric: the chargers, the electrical upgrades, the utility costs, and the timeline. It bridges sustainability goals and infrastructure reality across six analyses: load impact, charger sizing, site electrical assessment, managed charging, total cost of ownership (TCO), and distribution grid impact. Use it to support capital budget requests, utility coordination, and fleet procurement decisions.

Workflow

<Workflow - Fleet Electrification Analysis description="Produce a phased, feasibility-checked, cost-justified fleet electrification plan from fleet and site data." tools=[get_current_time, web_search, url_fetch, file_read, run_python, file_write, open_in_session_tab] triggers=["User asks to plan fleet electrification", "size EV chargers", "model fleet charging load", "assess site electrical capacity", "compare EV vs ICE total cost of ownership", "optimize managed charging", "assess grid impact of fleet charging"]

0. [Agent] Verify reference data before any calculation. Call get_current_time, then list which time-sensitive values the analysis needs (tariffs, fuel and electricity prices, incentive amounts, technology costs). Fetch each from the matching source in references/data-sources.md with web_search or url_fetch. Validate: every time-sensitive value has a source fetched this session or provided by the user. If fails: per Rule 1, stop and ask the user to confirm or provide the value.

1. [Agent] Ingest fleet inventory and characterize duty cycles. If the user supplied a file, read it with file_read. Record class, annual/daily miles, fuel type, and MPG per vehicle. Categorize by replacement priority. Compute daily energy as daily_miles / EV_efficiency (mi/kWh). Identify classes with EV equivalents today. Validate: every vehicle has class, daily miles, and a computed daily energy. If fails: request fleet data, or use class-average assumptions and state the uncertainty explicitly.

2. [Ask user] Confirm site electrical inputs: service voltage and amperage, distribution transformer kVA rating, existing peak demand, panel space, and distance from transformer to parking. Validate: the user supplies these or explicitly asks for conservative assumptions. If fails: use a conservative assumption (50% of transformer headroom already consumed by existing load) and label it.

3. [Agent] Size charging infrastructure using references/charging-and-electrical-specs.md. Match charger type to dwell time and energy per class, set charger-to-vehicle ratios from shift patterns, and compute total connected load and coincident demand. Specify circuit sizes per NEC 625.41. Compare connected load against the headroom from step 2. Validate: each class has a charger level, count, coincident demand, and circuit spec, with a stated pass/fail against headroom. If fails: flag the infrastructure gap and quantify the upgrade requirement.

4. [Agent] Model the coincident charging demand profile. Run the Monte Carlo simulation in references/charging-and-electrical-specs.md via run_python (numpy), in bounded chunks under the 60-second limit, writing results incrementally. Generate P10/P50/P90 demand curves at 15-minute resolution, overlay the existing facility load, and assess transformer loading against IEEE C57.91 limits. Validate: P10/P50/P90 profiles exist and a resulting billing peak is computed. If fails: fall back to the deterministic worst case (all vehicles arrive at once at minimum SOC) and label it.

5. [Agent] Optimize managed charging with the valley-filling heuristic in references/cost-and-grid-models.md (no LP solver in the sandbox). Apply the site power constraint and the verified TOU rate. Compare managed vs unmanaged peak demand and annual cost, and state the OCPP communication requirement. Validate: managed and unmanaged peak demand and annual cost are both reported. If fails: use the simplest heuristic (charge off-peak only) and show its value.

6. [Agent] Build the TCO comparison using references/cost-and-grid-models.md with the values verified in step 0. Model the ICE baseline and the EV scenario over one horizon and discount rate, allocate infrastructure cost per vehicle, find the crossover year, and run sensitivity on fuel price, electricity rate, incentives, and maintenance. Validate: both TCO curves, a crossover year (or "none within horizon"), and a sensitivity range are reported, each traceable to a sourced input. If fails: report TCO with explicit assumptions and uncertainty ranges.

7. [Agent] Assess distribution grid impact using references/cost-and-grid-models.md. Compute the transformer aging factor and projected life reduction, the hosting capacity limit, and ranked mitigations. Produce a utility coordination summary. Validate: aging factor, hosting capacity, and at least one mitigation are reported. If fails: flag that a detailed grid study needs utility engineering data.

8. [Ask user] Confirm where to save the deliverable (a directory path). Do not assume a location. Validate: the user provides a writable path. If fails: offer to write into the skill's working directory and confirm.

9. [Agent] Produce the phased transition plan and write it to the confirmed path with file_write, then open it with open_in_session_tab. Phase 1: highest-TCO-gap vehicles within existing electrical capacity. Phase 2: medium priority with first upgrades. Phase 3: remaining fleet with any service upgrade. Include timeline, capital budget by phase, and incentive strategy. Append the Rule 12 disclaimer and a list of which figures were verified live versus assumed. Validate: the file exists at the path and names every phase, budget, and the verification/assumption list. If fails: deliver the Phase 1 plan with placeholders for later phases and note what is missing.

</Workflow - Fleet Electrification Analysis>