Simulation timeline¶

A typical acud simulation timeline is a sequence of contiguous hours between the start and the end simulation datetimes, both included. For sizeable models with an extended timeline, this can lead to very long runtimes (hours) and massive hardrive memory consumption. A workaround for this is to run acud simulations with a snapshots-based timeline.

Snapshots-based simulation timeline¶

When running a simulation through a snapshots-based timeline, , you can specify a set of (non-contiguous) datetimes at which the optimization should be run. If required, these snapshots-based simulation results can then be extrapolated, assigning the nearest hourly results to non-computed datetimes (e.g. the extrapolated result of a non-computed datetime happening at 7a.m. will be the last computed result at 7 a.m.).

Snapshots are specified by:

Listing snapshots in a csv file with a single field called datetime, and
Referencing this csv file in a time.snapshots.file entry in the run configuration yml file.

Note that you also need to provide the date format in the time.snapshots.datetimeformat entry to ensure that the snapshot dates and times are correctly interpreted by acud. An extract of the time section of a run yml file including snapshots-based timeline specification is shown below.

time:
    startperiod: 2015-01-01 00:00
    endperiod: 2018-12-31 23:00
    stepsize: 24
    blocks: 3
    snapshots:
        file: csv/snapshots.csv
        datetimeformat: '%d/%m/%Y'

To further illustrate, these are the first lines of an example snapshots csv file. Note that the dates are non-contiguous.

datetime
01/01/2018
03/01/2018
10/01/2018
20/01/2018

Snapshots override start and end datetime specifications. If no snapshots specification is provided in the run config yml file, the simulation timeline will consist of the sequence of contiguous hours between the start and the end datetime, both included.

After completing a snapshots-based simulation, you can extrapolate the sparse timeline results to obtain a dense timeline solution using the xtra command, e.g. like this:

ad xtra run.yml