Comparing scenarios fairly
A single result tells you little. The useful question is what changes when you change one thing about the house or its systems, and that answer only means something if everything else stays equal. This article explains how EnergyFlow saves and compares runs, and how to keep a comparison fair.
What a scenario is
Each time you run the simulation, the run is saved as a scenario. It has a run number, the inputs exactly as they were when you submitted it, and the results. Editing the wizard afterwards does not change a saved run, so an old run always shows what it was built from.
To make a new scenario, change an input in the wizard and run again, or use Re-run simulation on the results page. The new run gets the next number. You can label a run by double-clicking its name, and pin one run as your baseline, the reference the others are measured against. A project runs one simulation at a time, so start the next run once the active one has finished.
How runs are compared
When you have at least two runs, the results page shows a comparison under the Scenario history strip. You choose which run to view and which to compare it against. It shows:
- every input that differs between the two runs, for example “Battery capacity: 5 kWh → 10 kWh” (an illustrative pair of values)
- up to six key figures with their change from the other run: Annual cost, Heating demand (the heat the house needs delivered), Cooling demand, Heat pump COP (heat delivered per unit of electricity used), Grid import total and Self-consumption (the share of your own solar production that you use at home instead of exporting it, shown only when there is PV)
- detailed metrics in groups: Thermal demand (the heat and cooling the house asks for), Equipment electricity, Grid import, PV + self-consumption, and Comfort (indoor air temperature, shown as Ta mean, Ta min and Ta max). A pool, an electric car or solar hot water adds groups of its own. Groups with a change open by default, and Show all opens the rest
A separate comparison page, opened from the Compare button in your projects list, puts several of your saved runs side by side in one table, with the configuration of each run and every metric in a row. It marks the best value in a row in bold green where a lower or higher number is normally preferable.
Missing is not zero
If a run does not have a metric, the comparison shows it as absent, never as zero. A dash means the value was not recorded, as in an older run made before that metric existed. If one run has equipment the other lacks, such as a battery or cooling, the row says “added in this run” or “removed in this run” and gives no change or percentage. A move from no battery to a battery is not a move from zero. Absent values are also left out when the table picks the best value in a row.
Colours are not verdicts
Green and red follow whether lower or higher is usually better. That is a convenience. A smaller heating figure from a house that has become too cold is not an improvement, so read the Comfort group next to the energy figures.
Change one thing at a time
If you change the heat pump size and the insulation together and the annual cost falls, you cannot say which change did it. Change one input per run, label each run with what you changed (“baseline”, “added shading”), and compare each against the baseline. You can still compare packages of changes, as long as you read the list of differing inputs and know what the result bundles together.
What to hold fixed
Weather
A typical year (an ordinary-looking year assembled from many real ones, so no single heatwave or mild winter skews it) and a real year differ for reasons that have nothing to do with your choices. The comparison lists Weather basis (Typical year or Specific real year) and Weather year among the differing inputs, so you can see if they moved. To learn how much the weather alone matters, run the same house on two weather years and compare those two runs.
Tariff
Use the same prices in both runs. A dynamic tariff has prices that change hour by hour with the wholesale market. With one, the price series follows the weather year: a real-year run uses that year’s day-ahead prices (the hourly prices fixed the day before), and a typical-year run uses a fixed reference year. Changing the weather year therefore also changes the prices.
Dispatch method
Dispatch is how the simulated home energy management system (EMS) decides when the battery charges and discharges and when power is bought or sold. The model switches automatically to an optimiser when a dynamic tariff and a battery are both configured, and uses fixed rules otherwise. The rules follow a set order: solar power covers the house first, spare solar charges the battery, the battery covers a shortfall, and the grid supplies what is left. The optimiser plans each day against the hourly prices and finds the cheapest plan within limits such as battery size and charging power. Switching tariff type can therefore change the method as well as the prices. The recap on the results page says which was used.
Schedules and setpoints
Keep household size, presence pattern (one of three daily patterns of when people are home) and thermostat settings the same unless one of them is what you are testing. Those settings, called setpoints, are the temperatures the heating and cooling try to hold.
What a difference means
A difference between two runs is a difference in the model under the assumptions you set. It is not a guarantee, and it does not carry an error bar. Treat a small gap as a tie, not as a ranking. Real houses add installation quality, controls and habits that the model does not see.
Limits of the model
- Both runs share the same simplifications. Shared errors partly cancel in a comparison, but an effect the model leaves out can matter more to one option than to the other, and then the comparison is tilted.
- Cost compares grid electricity only. Fuel, equipment cost, maintenance and financing are not part of it.
- Runs made before a metric existed, or whose hourly data was not kept, show their key figures only. The page tells you when charts for a run are not stored.