Battery Dispatch Under a Cloud-Edge Peak Event
A 180,000 ft² hospital campus has a 1,200 kW peak, 450 kW of rooftop PV, and a 250 kW/500 kWh lithium-ion battery. The utility tariff has a $22/kW-month demand charge based on the highest 15-minute interval and TOU energy: $0.16/kWh on-peak (2-7 p.m. weekdays) and $0.08/kWh off-peak. PV produces 1,200 MWh/year, mostly from 9 a.m. to 3 p.m. The battery is currently dispatched to shave the monthly peak to 950 kW by discharging 250 kW for one hour. Last July, a cloud edge cut PV output by 300 kW in five minutes; the battery was at 20% SOC, and the campus peak reached 1,180 kW. The BAS also starts two 200-hp chillers at 1:55 p.m. The battery has a 10-year warranty, 70% end-of-life capacity, and 88% round-trip efficiency. The demand charge is ratcheted at 70% of the prior 11-month peak.
1. Calculate the monthly demand-charge savings from shaving the peak from 1,200 kW to 950 kW, and explain why the July event still produced a high bill. How much usable battery energy is needed to guarantee a 250 kW shave for 30 minutes at 90% depth of discharge?
Read the worked answer
Demand-charge savings = (1,200 − 950) × $22 = $5,500/month, or $66,000/year if sustained. The July event defeated the shave because the battery was at 20% SOC: 20% of 500 kWh = 100 kWh DC. At 88% round-trip efficiency, delivered energy ≈ 88 kWh, so at 250 kW the battery could support only 88 / 250 = 0.35 h, or about 21 minutes. The cloud-edge loss plus the 1:55 p.m. chiller start created a net-load spike after the battery was depleted.
Required energy for a 30-minute, 250 kW shave: 250 × 0.5 = 125 kWh delivered. DC energy = 125 / 0.88 = 142 kWh. At 90% depth of discharge, nameplate = 142 / 0.90 = 158 kWh. The 500 kWh battery has enough nameplate energy, but only if the SOC reserve is managed before the event.
2. Compare the annual value of using the battery for TOU arbitrage instead of demand shaving. Assume 250 weekdays and that the battery cycles 400 kWh daily.
Read the worked answer
Arbitrage value: 400 kWh DC × 0.88 = 352 kWh delivered. Price spread = $0.16 − $0.08 = $0.08/kWh. Daily value = 352 × $0.08 = $28.16. At 250 weekdays, annual value = $7,040.
Demand-shaving value is $5,500/month, or $66,000/year, far larger than arbitrage. The comparison is not purely either/or: a 30-minute shave needs only about 142 kWh DC, leaving capacity for arbitrage on days when the peak is secure. But the July event shows that arbitrage must not consume the SOC reserve needed for the 2-7 p.m. peak. Prioritize demand shaving; arbitrage is secondary and should be limited to energy above the reserve floor.
3. Recommend control and operating changes to prevent a repeat. Include setpoints, forecasting, and chiller staging.
Read the worked answer
Set a pre-peak SOC floor of 50-60% by 1:30 p.m. on weekdays, not 20%. Use a PV forecast plus a cloud-edge trigger to increase battery reserve or pre-cool before 2 p.m. Stagger the two 200-hp chiller starts, for example at 1:40 and 1:50 p.m., and use soft-start or VFD ramp control. Implement net-load-following dispatch with a 250 kW cap and a 30-minute duration guarantee, and alarm if SOC falls below the reserve before 7 p.m.
Also verify BAS-to-battery communication latency, inverter availability, and interval-data alignment with the utility meter. If the campus repeatedly approaches 1,180 kW after control fixes, evaluate a 300 kW/600 kWh battery or additional peak-shaving measures; do not assume the existing 250 kW inverter can cover every cloud edge and chiller-start coincidence.