Specifying the input data

[jacob-mannhardt]

Why do we have default values and extra parameter files?

The dimensionality of our parameters demands many parameter values to be specified.
Example: Parameter demand of 20 energy carriers in 30 nodes in 8760 hours in 15 years demands
20 x 30 x 8,760 x 15 = 78,840,000 data points for a single parameter.

However, the demand is zero or constant for most carriers, nodes, hours etc. Therefore, we only want to provide those values that are necessary to provide.

What is the general workflow of specifying the input data?

We deploy a three-step procedure to specify the input data:

1. Define a default_value in the attributes.json (see Structure of the attributes.json files #351). This default value will be used for all indices of the parameter.
2. If necessary, provide a csv file, named <parameter_name>.csv, with additional data (see below)
3. Overwrite the default value with the specified values for the specified indices
4. If desired, scale the hourly-dependent time series up and down with <parameter_name>_yearly_variation.csv (see below)

How can data be provided in the csv files for the additional parameter values?

The most straightforward way is to specify all additional indices, for example here for demand:

In this example, the time steps are defined as rows and the nodes as columns. If only a subset of time steps or nodes were given, the values of the remaining indices would be the default value.
The orientation of indices (rows or columns) does not matter. Note that at least one index must be given as rows. We could also provide both indices as rows, e.g.:

time, node, demand
0,AT,9.426895
1,AT,9.550369
2,AT,10.28023
...
0,BE,11.92593
1,BE,11.68575
2,BE,12.88876
...
0,BG,4.445882
1,BG,4.780947
2,BG,5.557322
...

Often, we want to keep values constant for one index but specify the values for the other index. As an example, we could have a country-specific demand that is constant for all hours of the year. Or we could have an hourly-dependent demand that is the same for all countries. In both cases, we only want to provide the values for one index. We can achieve this behavior by dropping the constant index.

Node-independent demand:

Time-independent demand:

As before, we can still provide a subset of indices.

How can I vary an hourly-dependent parameter over the years?

Assuming that our demand (hourly-dependent) shall increase by 5% each year in country AT and by 10% in country BE, we can specify this in the <parameter_name>_yearly_variation.csv file. This is only relevant for hourly-dependent parameters. The yearly variation is given in relative terms (= multiplier) and applied to the entire time series. Thus, we cannot change the profile of the time series, but we scale it up or down. We first build the annual time series from the default values and the extra .csv file. Then we apply the
The resulting file would look like this:

year, node,demand_yearly_variation
2020,AT,1
2021,AT,1.05
2022,AT,1.1025
...
2020,BE,1
2021,BE,1.1
2022,BE,1.21
...