Behind-the-meter BESS design for C&I: sizing for demand charge reduction vs. resilience
C&I Energy Storage · Kriya Partners
A BESS sized for demand charge reduction is not the same system as one sized for resilience. The hardware might look similar on paper. The design brief is completely different.
The short version
One battery. Two completely different jobs.
Both use cases live behind the utility meter. Both store energy and dispatch it on demand. Beyond that, the similarities get thin fast.
A demand charge reduction system is an economic tool. It monitors 15-minute load intervals, identifies peaks before they set the monthly demand charge, and discharges to keep the facility under a target threshold. For large C&I customers, demand charges can represent 30 to 70 percent of the total monthly electricity bill. That is where the ROI lives.
A resilience system is an insurance asset. It earns its value when the grid goes down and critical operations keep running - not on normal days. The goal is maintaining defined loads for a defined duration: 4 hours, 8 hours, or longer depending on the operation and what a real outage costs.
When the use case isn't settled before design starts, the sizing defaults to one or the other. Usually demand charge reduction, because the ROI model is cleaner. The problem shows up when the outage comes.
Demand charge reduction
A kW problem sized for short, predictable events
Demand charges are billed on peak power draw - typically the highest 15-minute average in the billing period. One bad interval sets the demand charge for the whole month. A demand charge reduction BESS is designed to prevent that.
The primary sizing input is 12 months of 15-minute interval load data. You're identifying peak events: how high, how long, how predictable. The battery needs enough power (kW) to shave peaks below a target threshold for the duration of the peak window. Most C&I peak events run 30 to 90 minutes. That's the working duration, and it drives the kWh sizing.
The state of charge (SoC) strategy in a demand reduction system is financially optimized. The energy management system watches load forecasts and rate schedules, then positions the battery to be ready during anticipated peak windows. It doesn't need to be full all the time - only at the right level at the right time. Those are very different requirements.
Resilience design
A kWh problem - with harder constraints on controls and hardware
A resilience system starts with a different question: if the grid goes down right now, what needs to stay on and for how long?
That requires a critical load list - not all loads, only the ones that matter during an outage. Refrigeration, emergency lighting, key production lines, servers, HVAC for temperature-sensitive product or code compliance. Then a target backup duration: 4 hours to cover most local outages, 24 hours for a regional event, longer for an operation where any downtime is unacceptable.
Capacity sizing is kWh-led: critical load kW multiplied by required hours, adjusted for efficiency losses and end-of-life degradation. A facility with 150 kW of critical loads that needs 6 hours of backup needs roughly 900-1,000 kWh of installed capacity. The demand charge reduction system on the same site might be sized at 200-300 kWh.
But the kWh gap isn't even the most important difference. A resilience system has requirements a peak-shaving system doesn't:
- Inverter sized for full critical load including motor start surge currents Not the average discharge rate used in peak shaving. Starting HVAC compressors or refrigeration equipment during an outage creates momentary surge demands the inverter has to handle without tripping.
- Automatic transfer switching and islanding capability A grid-tied peak shaving system typically shuts down when the grid goes down. Resilience requires hardware and firmware that detects the outage, isolates from the grid, and continues supplying critical loads - automatically.
- SoC held near 100% at all times You don't know when the outage hits. The SoC strategy can't optimize for savings at the expense of readiness. The battery has to be available whenever the grid fails, not just during anticipated peak windows.
- EMS configured for reliability, not revenue The controls hierarchy changes completely. Reserve protection takes priority over any dispatch that would reduce available capacity. A financially optimized EMS and a reliability-first EMS are not the same software configuration.
Where projects go wrong
A peak-shaving system used as backup is not a backup system
A facility installs a BESS for demand charge reduction. The EMS does its job and discharges the battery during morning peak windows. Grid goes down at 2 PM. Battery is at 30 percent SoC. The system wasn't configured to island, so it shuts down anyway. The facility is dark.
The EMS didn't malfunction. The battery didn't fail. The system did exactly what it was designed to do. It just wasn't designed for this.
Outage costs for C&I operations range from a few thousand dollars per hour for small commercial facilities to tens of thousands per hour for manufacturing, food processing, cold storage, or any site where downtime directly stops production. The hardware cost difference between a peak-shaving system and a properly sized resilience system is almost always smaller than one extended outage at a facility where downtime is expensive.
The reverse mismatch costs differently. A system oversized and configured for resilience - held near full SoC, with capacity reserved and never dispatched for peak shaving - leaves demand charge savings on the table every month. The economic case weakens and often doesn't pencil the way the pro forma projected. Both mismatches are real. Both are avoidable with the right brief up front.
The design brief
Two different sets of questions - and they have to come first
The use case determines the brief. These aren't secondary details to sort out after a system is sized - they're the starting point.
Demand charge reduction brief
- What does the 15-minute load profile look like over 12 months?
- What is the utility's demand charge rate ($/kW/month)?
- How high are the peaks, and how long do they last?
- What target demand threshold makes the economics work?
- Is there a demand ratchet that amplifies the cost of missing a peak event?
- What are the available recharge windows between peak periods?
Resilience brief
- What are the critical loads and what is their combined kW draw?
- How long does backup need to last?
- What is the outage history at this site and in this region?
- Are there regulatory, insurance, or lease requirements for backup power?
- Does the facility have uptime requirements tied to contracts or compliance?
- Which loads have motor start surge requirements that affect inverter sizing?
Some projects need both. A system can be designed to reduce demand charges during normal operations while maintaining a protected resilience reserve - but that requires a controls strategy that balances both objectives, and the kWh capacity to support both without compromising either. It's a more complex brief, but it's not uncommon for sites where both use cases carry real economic weight.
The use case question isn't a preference. It determines everything downstream: inverter sizing, kWh capacity, SoC strategy, controls configuration, and whether the hardware includes the transfer switching needed to actually island from the grid. Get it wrong at the start, and the rest of the design is solving for the wrong problem.
Kriya handles BESS design, engineering, and permitting across all 50 states. We work through the use case question before anything gets sized - because the answer changes everything that comes after it.
