Substation Demand¶

When you build a model with the Study Area tool, TEMPO can estimate an electricity demand for each substation. It starts from the population of the area, splits the total across the substations, and gives each one an hourly (or finer) load shape.

This is a starting point for regions where you have no measured load data. Replace it with real demand series when you have them.

Turning it on¶

In step 2 of the Study Area network builder (Substations), switch on Add estimated demand.

The estimate needs a population figure. TEMPO takes it from the OpenStreetMap data of the places you selected and adds the figures up if you selected several. If none of them reports a population, the panel says so and no demand is created.

How the total is calculated¶

annual demand (kWh) = population × per-capita demand (kWh per person per year)

Per-capita demand defaults to 3,500 kWh per person per year. Change it to fit your region; national statistics are a good source.

TEMPO converts the annual total into an average power in MW and splits it across the substations:

Split option	How it works
Evenly	Every substation gets the same share
Weighted by voltage (size proxy)	Each substation's share is proportional to its voltage in kV. If no substation has a voltage tag, TEMPO falls back to an even split

Load shape¶

The load shape decides how demand moves through the day, week and year. You can pick one standard load profile or mix several.

One profile¶

Pick a profile from the Load shape (BDEW SLP) list. These are the German BDEW standard load profiles:

Group	Profiles
Flat	Constant load, no time variation
Residential	Household dynamic (`h0_dyn`), Household (`h0`)
Commercial	General (`g0`), offices (`g1`), evening-heavy (`g2`), continuous 24/7 (`g3`), shops (`g4`), bakery (`g5`), weekend-heavy (`g6`)
Agriculture	General (`l0`), dairy and livestock (`l1`), other (`l2`)
BDEW25 (2025)	Household (`h25`), Commercial (`g25`), Agriculture (`l25`)

Sector mix¶

Tick Mix to blend four sectors by weight instead:

Sector	Profile	Default weight
Household	`h0_dyn`	50 %
Commercial	`g0`	35 %
Industry (24/7)	`g3`	10 %
Agriculture	`l0`	5 %

Weights are normalised, so they don't have to add up to 100. A weight of 0 drops the sector.

Calendar and resolution¶

Calendar sets which country's public holidays demandlib takes into account when it builds the BDEW profiles. The default is DE because the BDEW curves are German. The built-in approximation (see below) ignores this setting.
Resolution is the timestep of the generated series: 60, 30 or 15 minutes. TEMPO also sets the model's resolution to match.

A small chart under these settings previews one week of the shape (Monday to Sunday).

demandlib or the built-in approximation¶

The real BDEW curves come from demandlib, which TEMPO runs in its own Python environment. It is optional.

With demandlib installed, TEMPO generates the actual BDEW profiles. The preview is labelled BDEW (demandlib).
Without it, TEMPO uses a built-in synthetic approximation of each profile, with daily, weekly and seasonal patterns. The preview is labelled synthetic (approx.) and the panel shows an Install demand profiles button.

You can install demandlib from that button, or tick it as an optional component in the setup screen on first launch.

What gets created¶

When you click Import all:

Each substation gets a power_demand technology.
With a load shape other than Flat, TEMPO writes a time series file called osm_substation_demand.csv. It has a datetime column and one column per substation, named after the substation. Values are in MW and negative, following Calliope's demand convention. They are absolute values, so no resource_scale is needed and the same file works with every engine.
With Flat, no file is written. Each substation gets a constant demand instead.

The file appears in the TimeSeries view like any other series, where you can look at it and edit it.

Changing model dates later¶

The generated series covers the model's start and end dates at the chosen resolution. If you change the dates or the resolution afterwards, open the series in TimeSeries. A banner shows whether the series still matches the model. If it doesn't, the banner turns amber and tells you how many rows the series has against how many the model expects. Click Regenerate from demandlib to rebuild it with the same settings. If demandlib is not installed, the button uses the built-in approximation instead.