Monte Carlo

The Monte Carlo method is a statistical technique that involves running multiple simulations with randomly sampled input parameters to estimate the distribution of the output variables.

In Africa, navigating decision-making amidst rapid demand growth and substantial infrastructure changes poses a significant challenge. To tackle this uncertainty, the Monte Carlo method can be employed to construct a stochastic interpretation of deterministic model scenarios. This approach enhances the ability to account for variability and assess potential outcomes, providing a more robust foundation for informed decision-making.

To use the Monte-Carlo method in PyPSA-Earth, you have to activate the monte_carlo option in the configuration file (config.yaml), set add_to_snakefile to true in the monte_carlo section. This will enable the monte-carlo method.

There are a few additional options that needs to be set in the monte_carlo configuration options.

Set options

• samples: The number of samples to be used in the monte-carlo simulation.

• samppling_strategy: The method used to sample the input parameters. Either of pydoe2, chaospy, or scipy.

• seed: The seed for the random number generator. It is useful to set the seed to a fixed value to ensure reproducibility of the results.

Set uncertainties

The uncertainties section in the configuration file is used to specify the parameters to be sampled in the monte-carlo simulation. The uncertainties section is a dictionary with the keys being the pypsa object value to be sampled and the values split into type and args of which type is used to select the distribution and args used to specify the parameters of the selected distribution type.

The following is an example of the uncertainties section in the configuration file:

uncertainties:
    loads_t.p_set:
      type: uniform
      args: [0, 1]
    generators_t.p_max_pu.loc[:, n.generators.carrier == "onwind"]:
      type: lognormal
      args: [1.5]
    generators_t.p_max_pu.loc[:, n.generators.carrier == "solar"]:
      type: beta
      args: [0.5, 2]

Note

To understand the different distribution types and their parameters, check Scipy Documentation

Tip

To create reallistic uncertainties, it is important to pay attention to the distribution that is being applied to each pypsa object parameter.

Workflow

To perform a dry-run of the monte-carlo simulation after setting the config options, use the following command:

.../pypsa-earth % snakemake -j 1 solve_all_networks_monte -n

To create a DAG of the monte-carlo simulation workflow, use the following command:

.../pypsa-earth % snakemake -j 1 solve_all_networks_monte --dag | dot -Tpng > monte_carlo_workflow.png

The monte-carlo simulation can be run using the following rule:

.../pypsa-earth % snakemake -j 1 solve_all_networks_monte

Note

Increasing the number of cores can make the process run faster. The numbers of cores can be increased by setting the -j option to the desired number of cores.