Five-Year Peak Forecast with Behind-the-Meter PV, EVs, and a Data Center
An integrated utility serves a summer-peaking system. Its weather-normalized peak for the most recent year was 2,850 MW; a 1-in-10 weather adjustment adds 4%. Over the next five years, planners expect 450 MW of customer-sited PV nameplate capacity, 120 MW of new peak-coincident EV charging, and a contracted 300 MW data center that will energize in year three. The data center has an 85% peak coincidence factor and a contractual take-or-pay minimum of 200 MW. The utility uses a 15% planning reserve margin and currently has 3,200 MW of accredited capacity, but 200 MW of coal retires in year two. Its current resource plan assumes PV contributes 30% of nameplate at peak. A developer offers a 400 MW combined-cycle unit with a 24-month lead time and a 300 MW demand-response program that tests at 60% availability during system peaks. The commission has asked whether the utility can avoid new capacity.
1. Compute the year-five planning peak and capacity need under the utility's 30% PV accreditation assumption. Show each adjustment and explain why using 65% nameplate coincidence would be inappropriate for resource adequacy.
Read the worked answer
Start with the weather-normalized peak and adjust in a consistent order:
- Weather: 2,850 x 1.04 = 2,964 MW.
- EV charging: +120 MW = 3,084 MW.
- Data center: 300 x 0.85 = 255 MW; the 200 MW minimum is not binding because 255 > 200; total = 3,339 MW.
- PV: 450 x 0.30 = 135 MW accredited; planning peak = 3,204 MW.
- Reserve margin: 3,204 x 1.15 = 3,684.6 MW required; existing accredited capacity after retirement = 3,200 - 200 = 3,000 MW; need about 685 MW.
Using 65% would subtract 292.5 MW and understate the need by about 157.5 MW at the planning-peak level, or about 181 MW after reserve. Coincidence at the system peak is not the same as firm capacity: PV output is weather-dependent, correlated with other PV, and may be unavailable during a 1-in-10 peak. The 30% ELCC is the utility's approved planning assumption; the 65% figure is an operating estimate. Recompute with sensitivity to 20% and 40% PV ELCC and disclose the range.
2. Should the 300 MW demand-response program count toward the deficit? State the accreditation and contract conditions you would require.
Read the worked answer
Only if it is firm and dispatchable at the planning peak. A test at 60% availability supports an accredited contribution of 180 MW, not 300 MW; use 300 x 0.60 = 180 MW unless historical event performance justifies more. Conditions: telemetry, day-ahead and real-time dispatch, measurement and verification, penalties for non-performance, no double counting with emergency-only programs, and a minimum duration matching the loss-of-load event, for example four hours. Demand response can defer a combustion turbine, but it should not be the sole resource for a 685 MW deficit. A staged portfolio of demand response, short-term purchases, and the 400 MW combined-cycle unit with a 24-month lead time preserves flexibility. If the program is voluntary and price-triggered, treat it as a sensitivity, not firm capacity.
3. What is the largest forecast risk, and what decision rule should trigger procurement?
Read the worked answer
The data center is the largest single risk because 255 MW at peak is about 8% of the year-five peak and its timing, load factor, and on-site backup are uncertain. EV and PV coincidence are second. Use scenario analysis: low case with data center at 50%, EV at 50%, and PV ELCC at 40%; base case; high case with data center at 100%, EV at 120%, and PV ELCC at 20%. Decision rule: if the probability-weighted peak plus reserve exceeds 3,000 MW by more than the lead time can cover, or if the data center's energization is confirmed and its minimum take is at least 200 MW, issue an RFP for 300 to 500 MW of capacity with online dates in years three to five. Include a trigger review every six months and contract options rather than committing to the full combined-cycle unit.