When the default working mode of the storage devices is set to Cost optimization on the settings page, the box Choose your cost optimization mode appears below the dropdown. It decides how the EPSherpa EMS uses the battery to lower the energy bill:
| Mode | When to choose it |
|---|---|
| Predictive AI algorithm (default) | Recommended for most sites. The EPSherpa EMS plans the battery with forecasts of the weather, the consumption and the energy prices. |
| Control manually via fixed rules | You want predictable behaviour that you set yourself: charge and discharge at fixed price levels and charge levels, without forecasts. |
| Recharge at low cost and discharge only at the peak shaving discharge thresholds | Only offered for the cost optimization partition of the peak shaving plus mode. |
Cost optimization is only as good as the prices it works with. Check that the energy contract and grid costs are configured correctly in the app, see Pricing parameters.
After changing any of the settings below, scroll down and click Save.
The predictive AI algorithm analyzes the weather and consumption forecasts and combines them with the energy prices to maximize the return of the battery. It continuously calculates the most profitable plan based on the data it has at that moment. Forecasts change during the day, and what actually happens can differ from what was predicted, so a decision that looks wrong afterwards can have been the best one at the time it was made. See Predictions versus ex-post analyses.
| Strategy | Behaviour |
|---|---|
| Conservative | Opts for the safest predictions, so that there is always enough energy in the battery, possibly at a slightly lower return. This strategy always charges the battery to a minimum level before high prices start, even when that option is not ticked: it then uses the level entered next to that option (70% unless changed). |
| Balanced (default) | The middle ground between safety and return. |
| Aggressive | Hunts for the highest possible return, with a higher risk that the battery is charged less than ideal when the forecasts turn out wrong. |
| Parameter | Default | Explanation |
|---|---|---|
| Allow energy trading (stop trading below X% charged) | Off, 70% | Allows selling energy from the battery to the grid to maximize profit. When it is off, the battery is only discharged to cover the consumption of the site. Below the chosen state of charge no energy is sold, except in two cases: when negative purchase prices are expected in the next 12 hours, or when the current selling price is higher than the highest purchase price expected in the next 6 hours (the energy can then be bought back for less). |
| Factor battery wear into decisions (X cent/kWh per charging cycle) | On, 4 cent/kWh | The algorithm accounts for this wear cost for every kWh that passes through the battery, on top of the efficiency losses. The battery is only used when the price difference is worth it, so on days with relatively flat prices the battery may not be very active. The settings page warns when the value is above 6 cent/kWh, because the battery would then rarely be used. See Accounting for the battery wear and efficiency. |
| All PV of the installation is connected to the EPSherpa EMS | Off | Tick this only if all the PV in the electrical installation is connected to the EPSherpa EMS. The algorithm then knows that it can curtail PV that would otherwise exceed the grid export limit or inject at negative selling prices. When it is not ticked, the algorithm must assume that there can be PV it does not know of, and therefore that injection beyond the grid export limit or at negative selling prices remains possible. Leave it unticked if you are not sure. |
| Preferentially store excess production immediately in the battery | Off | Charges excess production into the battery immediately, even when the algorithm would rather charge later. This setting is ignored when trading is allowed and the purchase prices are high (otherwise selling would not always be possible), and when negative purchase prices are expected in the next 12 hours (charging at those prices is always better). |
| Make sure that the battery is charged to a minimum level before high prices start (X%) | Off, 70% | Charges the battery to at least this level right before the prices go from low to high, so that the charge is available during the high prices. A transition only counts when there is a noticeable difference between the night and morning prices, or between the afternoon and evening prices. No energy is sold to the grid below this level; it is meant for self-consumption only. |
"Make sure that the battery is charged to a minimum level before high prices start" can keep the battery from discharging during the night before a day with very low prices around noon. The same applies to the Conservative strategy.
The fixed rules take predictable actions: you set exactly what the battery does at which prices and charge levels. The rules do not use forecasts; for more adaptive control, use the predictive AI algorithm.
The price thresholds are set as a number of hours within each half of the day: from 00:00 till 12:00 and from 12:00 till 24:00. "The 4 hours of lowest prices" means the 4 cheapest hours of the current half day. This way the rules anticipate both the typical morning peak and the evening peak, also on days when the evening prices are much higher than the morning prices. As soon as the conditions of a rule are met, the battery charges or discharges, either for self-consumption or at the highest possible power, depending on the rule.
| Option | Behaviour |
|---|---|
| Never charge from the grid | The battery is only charged with excess PV production. |
| Only charge from the grid at negative purchase prices (default) | The battery charges from the grid at the highest possible power, up to 100%, only when the purchase price is negative, so you get paid for consuming. |
| Smart charging at low purchase prices | The battery charges from the grid at the highest possible power when the purchase price is among the X hours of lowest prices (default 4) and the battery is less than Y% charged (default 50%). At negative purchase prices it charges up to 100%. |
| Option | Behaviour |
|---|---|
| Always store all excess PV production immediately in the battery (default) | Every bit of excess PV production goes into the battery. |
| Only store PV production that would otherwise be sold at a low price | Excess PV production is only stored when the selling price is among the X hours of lowest prices (default 6), or negative. Otherwise it is sold to the grid. |
| Option | Behaviour |
|---|---|
| Always discharge for self-consumption (default) | The battery covers the consumption of the site, so that no energy is bought from the grid, except when the purchase price is zero or negative. |
| Only use at a high purchase price | The battery covers the consumption only when the purchase price is among the X hours of highest prices (default 4). It never discharges at a zero or negative purchase price. |
| Option | Behaviour |
|---|---|
| Never sell from the battery (default) | The energy in the battery is only used for self-consumption. |
| Sell at a high selling price | The battery discharges at the highest possible power when the selling price is positive and at least X cent/kWh (default 20) above the lowest selling price of the current half day. It stops below Y% charged (default 70%). |
This mode only appears for the cost optimization partition of the peak shaving plus mode. The partition is used for peak shaving only: it is recharged at the cheapest moments, but never discharged for cost reasons, even when prices are high. To keep the risk low, it always recharges within the next 6 hours. This mode has no further settings.
Two advanced settings influence the cost optimization of storage, see Advanced settings: the size of the self-consumption variability buffer, and whether batteries may be put in standby at low power setpoints.