Most commercial storage projects that underperform are not badly built. They are sized against the wrong number — usually nameplate capacity instead of the load event that actually drives the electricity bill.
This framework is the sequence we use with facility owners, EPCs and integrators. It works on 15-minute interval data and a tariff sheet. It produces two outputs: a power rating in kW, an energy rating in kWh, and a simple payback you can sanity-check before requesting a formal quotation.
1. Decide which problem you are buying a battery to solve
Commercial storage earns money or avoids cost in several distinct ways, and they pull sizing in different directions.
| Objective | What it needs | Sizing implication |
|---|---|---|
| Demand charge reduction (peak shaving) | High power for short, predictable windows | Power-led sizing; capacity can be modest |
| Time-of-use energy arbitrage | Daily charge/discharge cycling | Energy-led sizing; efficiency and cycle life dominate |
| PV self-consumption | Absorbing midday generation, discharging in the evening | Sized to array output and evening load |
| Backup / resilience | Instant transfer on outage, critical loads | Critical-load inventory drives power; duration drives energy |
| Capacity deferral | Avoiding a transformer or feeder upgrade | Peak-load coincidence with upgrade trigger |
Decide the primary objective in writing before sizing. A system sized for arbitrage will be out of position for peak shaving, and vice versa.
2. Build the load profile — this is the only non-negotiable input
Request 15-minute interval data for at least a full year from the utility or the site’s meter, plus the tariff sheet. Three derived numbers matter:
- Peak demand — the highest 15-minute average, and when it occurs
- Average load — energy consumption divided by the period
- Load factor — average load ÷ peak demand
A low load factor means short, expensive peaks and a strong peak-shaving case. A high load factor means the savings have to come from energy arbitrage and self-consumption instead. Where interval data is unavailable, a temporary power logger over four to six weeks is a better investment than any sizing spreadsheet built on assumptions.
3. Size power first, then energy
Step 1 — Set the target reduction in kW
Read the tariff structure. If demand is charged monthly on the highest 15-minute average, and the site’s peak is 800 kW with the next 90th-percentile peak at 610 kW, then shaving to roughly 610 kW removes the top, most expensive slice of demand without touching normal operations.
Target discharge power (kW) = current peak demand − target peak demand
Step 2 — Convert to usable energy
Usable energy (kWh) = target discharge power (kW) × discharge duration (h) ÷ round-trip efficiency
The duration comes from how long the peak band actually lasts, not from a default number. If your peaks are two hours long, a four-hour battery is over-specified for peak shaving.
Step 3 — Convert usable energy to nameplate energy
Nameplate energy (kWh) = usable energy (kWh) ÷ depth of discharge
Depth of discharge is set by the cell chemistry, the BMS and the warranty terms. A system that is warranted to cycle at 90% depth of discharge needs less nameplate capacity than one limited to 80% for the same delivered energy — which is why comparing systems on price per nameplate kWh alone is misleading.
Worked example — illustrative inputs
The figures below are illustrative inputs, not market averages. Replace them with your own tariff and load data before using the output.
Inputs
- Peak demand: 800 kW; target after shaving: 600 kW
- Peak band duration: 4 hours per day, ~250 days per year
- Demand charge: $12 / kW / month
- Energy rate: $0.22 / kWh peak, $0.09 / kWh off-peak
- Round-trip efficiency: 88%; depth of discharge: 90%
Calculations
Target discharge power = 800 − 600 = 200 kW
Usable energy = 200 kW × 4 h ÷ 0.88 = 909 kWh
Nameplate energy = 909 kWh ÷ 0.90 = 1,010 kWh
Demand charge saving = 200 kW × $12 × 12 months = $28,800 / year
Arbitrage margin = 909 kWh × 0.9 cycles × 250 d
× ($0.22 − $0.09) = $26,588 / year
Total illustrative benefit ≈ $55,400 / year
Simple payback = installed cost ÷ annual benefit. At an illustrative installed cost of $300 per kWh for 1,010 kWh, payback sits at roughly 5.5 years before degradation and O&M. The sensitive variables are obvious and worth challenging before any commitment: the avoided demand charge, the peak/off-peak spread, the number of cycling days, and the installed cost per kWh.

4. Adjust for what the first-pass calculation leaves out
- Degradation. Capacity and round-trip efficiency fall over the life of the system. Model year 1 and year 10 separately rather than using a single annual figure.
- Cycling limits. Warranty terms cap throughput. If the arbitrage case needs two cycles a day, check the warranty allows it.
- Parasitic load. Thermal management and controls consume energy. Ask for the auxiliary consumption figure, not just the inverter efficiency.
- Demand charge ratchets. Some tariffs bill on a rolling 12-month maximum. A single bad month sets the floor for a year; the control strategy has to prevent it.
- Backup value. Resilience is real value but hard to monetise. State it separately so it does not inflate the payback calculation.
5. Site conditions that decide feasibility
- Space and floor loading. A 1 MWh cabinet system has a footprint and a point load. Confirm slab capacity and access routes before selecting a product.
- Thermal management and IP rating. Where will the enclosure sit — indoor plant room, outdoor pad, or shaded canopy? Ambient temperature and dust decide whether air cooling is sufficient.
- Electrical interface. Available transformer capacity, breaker and cable sizes, switchgear space, and whether the connection is on the low-voltage or medium-voltage side.
- Grid interconnection and protection. Anti-islanding, protection settings, and whether the utility requires an interconnection study.
- Fire and code requirements. Separation distances, ventilation, detection and suppression requirements vary by jurisdiction and by system type. These constraints can eliminate a design faster than the economics can.
- Permitting and interconnection lead time. Usually longer than the equipment lead time. Start it in parallel with procurement.
6. Choose the architecture deliberately
| Architecture | Typical fit | Trade-off |
|---|---|---|
| All-in-one cabinet (integrated battery, PCS, EMS) | 60–500 kWh, indoor or pad-mounted, single- or three-phase | Fast to deploy, less granular expansion |
| Stackable / modular multi-unit | 100 kWh to several MWh, phased growth | More wiring and coordination, flexible scaling |
| Containerised | 500 kWh and above, dedicated sites | Highest energy density, needs site works and permitting |
Also decide early whether you need three-phase output, parallel capability for future expansion, islanding with automatic transfer, generator coordination, and an EMS that can expose data to your existing building management system. These are firmware and switchgear decisions, not afterthoughts.
7. What to put in your RFQ
- Load profile summary: 12 months of peak demand, average load, load factor, and the shape of the peak band.
- Target discharge power and duration, and the objective driving them.
- Tariff structure and the rates you are applying.
- Site conditions: location, ambient temperature range, IP requirement, available space and access.
- Electrical interface: voltage, phases, transformer and switchgear details.
- Required certifications for your market.
- Monitoring, EMS integration and reporting requirements.
- Service expectations: commissioning support, remote diagnostics, spare parts, warranty terms.
A supplier who responds with a load-profile question rather than an immediate price is telling you something useful.
Five sizing mistakes worth avoiding
- Sizing on nameplate kWh and comparing systems on that basis alone.
- Assuming a four-hour battery when the peak band lasts ninety minutes.
- Ignoring round-trip efficiency and parasitic load in the savings model.
- Treating a demand-charge ratchet as if it reset every month.
- Committing to equipment before the interconnection and permitting path is understood.

FAQ
What is the smallest commercial storage system worth installing? It is set by the economics, not the hardware. If the site’s peak demand and tariff spread cannot generate meaningful annual savings, a smaller system will not fix that. Run the framework above on real interval data before selecting a size.
Can a storage system be expanded later? Yes, if the architecture and switchgear are chosen for it. Modular and stackable systems are usually the practical route to phased expansion; ask about parallel capability at the design stage rather than after commissioning.
How long does a commercial system take to deliver and install? Equipment lead time is usually the shorter part of the schedule. Interconnection studies, permits and site works typically dominate. Confirm both tracks before you set a commissioning date.
Do I need an EMS, or is the inverter enough? For peak shaving and arbitrage, the control logic is the product. An EMS that can read your meter, forecast the peak band and respect the battery’s warranty limits is what converts hardware into savings; a bare inverter without that logic will not capture them reliably.
What certifications should I require? Ask for the standards relevant to your market and application — for stationary storage these commonly include UL 1973, UL 9540 and UL 9540A evidence in North America, and IEC 62619 evidence elsewhere. Confirm with your authority having jurisdiction and your insurer.
Have your interval data ready? Send us 12 months of 15-minute load data, your tariff structure and your site constraints. We will return a system configuration with a power rating in kW, an energy rating in kWh, and the architecture we recommend — based on our commercial and industrial storage range.