Quick answer
Size the battery around what the site needs it to do. Use kW to cover how high the peaks are and usable kWh to cover how long they last. Then compare real battery and PCS options, their full installed costs and the savings they produce, and choose the option that best meets the client’s payback, IRR or NPV goal.
Start with the job the battery has to do on the client's site. Power in kW tells you how much of the site's load it can cover at one time. Usable capacity in kWh tells you how long it can keep doing it.
If the site goes 100 kW over the target import level for half an hour, the battery needs about 50 kWh of usable energy before you allow for losses and reserve. If that same 100 kW excess lasts three hours, it needs about 300 kWh. The discharge power is the same, but the second job needs far more stored energy.
That is why you use the client's own half-hourly data, not a standard business profile. Elexon says its published load profiles represent an average customer in each profile class, so they cannot show you when this site peaks or how long it stays above the target.
For each size, check:
- How often the site goes over the target import level
- How long it stays there
- Whether solar surplus or a cheaper tariff period can recharge the battery
- Whether the next peak hits before the battery has recharged
- The site's import and export limits
- How much usable capacity is left after reserve and losses
- Whether the client plans to add EV chargers, machinery or extra shifts
Half-hourly data averages everything over the half hour, so when a big motor or a bank of chargers pulls hard for a minute or two, that spike gets buried in the average. If a short spike like that is what sets the PCS size, the half-hourly figure reads low and you size the PCS too small. Put a logger on the site for a week or two, catch the real peaks, and size the PCS off those before you lock the spec in.
Compare real packages and find where the extra spend stops paying back
Once you have the rough kW and kWh range, compare real packages. Run something like 100 kW/200 kWh, 100 kW/400 kWh and 200 kW/400 kWh through the same tariff and load model. Use the full price to put each one in, including:
- Extra battery cabinets
- The PCS
- Switchgear and cabling
- Civils
- Controls
- Commissioning
Then compare the annual savings, the payback and the lifetime return for each package.
The bigger system can save more money overall and still give the client a worse payback. That happens when the site has too few long peaks, the battery cannot recharge between them, or the site barely uses the extra capacity for much of the year.
Which way you go depends on what the client is buying. A client who wants the money back fastest takes the shorter payback, which usually means the smaller system. A client who wants the biggest return on the spend looks at net present value, and NPV usually points at a bigger system. So you rank the same options by payback, IRR or NPV, and size to the measure the client cares about most.
GridVolt's simulator takes that off your hands. It runs the combinations against budget prices per kW and per kWh, then ranks them by payback, IRR or NPV. You quote the size with the strongest case, and show the client what the next size up gets them and what it costs in payback.
Related questions
Do I need more battery power, or more usable capacity?
How do I get and check a client's half-hourly data before sizing a battery?