ie3-institute / simbench2psdm

Tool to convert SimBench data sets to ie³'s power system data model

simBench2pdsm

Tool to convert SimBench data sets to ie3's power system data model. SimBench has been research project of Kassel university, Fraunhofer IEE, RWTH Aachen university and TU Dortmund university, aiming at provision of realistic models of electrical distribution grids including time series. The data is availabe at https://simbench.de.

There is another conversion tool available here, that makes the models available to the well known pandapower simulation software.

simbench2psdm's Issues

Take care of class NodePFResult

As of now I did not find an example file for power flow results of nodes.

The susceptance of the transformer conversion has to be negative

The reactance we calculate in the so called "no load experiment" is inductive as it relates to the coil of the transformer. Therefore the susceptance (its inverse) is negative by definition.

Implement concurrency strategy

It should be possible to convert multiple data sets concurrently.

Deliver result as archives

Currently, only a folder of csv files is delivered. Add the functionality to deliver them as archives.

Write converted grid model to csv

... somehow self-explanatory.

Add coverage support for ScalaTest

Reporting of test coverage is foreseen to be made with scoverage. This unfortunately has some issues dealing with Scala 2.13.

Map "subnet" to psdm conept of "subnet"

In PowerSystemDataModel, the subnet denotes a galvanically separated sub grid. However, in certain data sets of SimBench this seems not to be the case, e.g. "1-EHV-mixed--0-sw". In this specific example, subnet "EHV1" mixed 380 kV und 220 kV sub grids.

• Clarify the concept of SimBench's subnet
• Extend subnet mapping from simply mapping the field subnet, but also consider different voltage levels

Repersist the grid with fixed PSDM csv sink

Since the PSDM update: ie3-institute/PowerSystemDataModel#583 an issue within the csv sink.

• Update the PSDM dependency
• repersist all grids
• upload them to the shared drive

Config validator is only used in tests

The validity checks of the config validator seem to only be called in the ConfigValidatorSpec.

Output time series meta data

Currently after conversion we do not know anything about the load profiles, as all their metadata isn't persisted.
Write out the additional data.

Update jdk version to 17

JDK should be updated from 1.8 to 17. This will also fix the incompatibility with current SonarQube.

Update PSDM version

Address extensive memory usage

Currently, there are several grid models, that consume a lot of memory:

• 1-HVMV-mixed-all-0-no_sw
• 1-HVMV-mixed-all-1-no_sw
• 1-HVMV-mixed-all-2-no_sw
• 1-HVMV-urban-all-0-no_sw
• 1-HVMV-urban-all-1-no_sw
• 1-HVMV-urban-all-2-no_sw
• 1-MVLV-comm-all-0-no_sw
• 1-MVLV-comm-all-1-no_sw
• 1-MVLV-comm-all-2-no_sw
• 1-MVLV-rural-all-0-no_sw
• 1-MVLV-rural-all-1-no_sw
• 1-MVLV-rural-all-2-no_sw
• 1-MVLV-semiurb-all-0-no_sw
• 1-MVLV-semiurb-all-1-no_sw
• 1-MVLV-semiurb-all-2-no_sw
• 1-MVLV-urban-all-0-no_sw
• 1-MVLV-urban-all-1-no_sw
• 1-MVLV-urban-all-2-no_sw

Try to address those cases.

Let children of ProfileType be parsable enumerations

... instead of case objects handed back by a very cruel apply method. Applies to

• LoadProfileType
• StorageProfileType
• ResProfileType
• PowerPlantProfileType

Handle models parallely

By implementing a concurrent handling in Coordinator, e.g. together with kind of a worker pool or controllable amount of children.

Allow for removing of switches

Currently, switches are converted in any case.

Sometimes, a simulator might only support a bus-branch model without switches. Therefore, allow the removal of switches from target data set.

Negative transformer no load susceptance

When going over to PowerSystemDataModel version 3.0, we have to provide the no-load susceptance as a negative value.

Search for memory leaks

When working on big input models or in concurrent operation, the memory is exhausted quite quickly. In order to see, if this is just a very memory consuming application or it is caused due to memory leaks: Have detailed look at it.

Wrong line type selected

... e.g. when converting "1-EHVHV-mixed-all-0-no_sw".

Within the routine, the line types get converted and result in pairs of types with similiar / equal attributes, but different rated voltages. Specifically, there are types, with the same electrical parameters, one for 220 kV rated voltage and one for 380 kV.

However, unfortunately, when converting the lines themselves, the wrong type gets select. Let there be a line between two 220 kV-nodes, then a 380 kV line type is selected, although a 220 kV type is available.

Please do research on how that error can occur and try to develop a solution.

Increase test coverage

Switch over to newer PowerSystemDataModel version

Problems with unit comparison of "equal units"

In 78c168d when comparing values of kind:

      actual.getgM() should equalWithTolerance(
        Quantities.getQuantity(826.44628, MetricPrefix.NANO(SIEMENS))
      )

We got failing tests, even though they shouldn't.

The problem of the test failing was the unit even though it was of unit nano Siemens.
Looking into the unit actual had a noted symbol attribute while expected didn't.
I suspect that it has something to do with the StandardUnits.

In any case this comparison should work. Find out what this is about

Convert Simbench data model to PowerSystemDataModel

Tidy up download folder, when file is already apparent

Add argument / config parameter to convert DER processing to their corresponding asset in PSDM

At the moment, all infeed assets are converted to FixedFeedIn entities with their infeed timeseries attached. It would be nice to have to ability to convert the entities to their PSDM equivalent (e.g. pv -> pv) as configuration parameter.

This would allow using existing models to calculate infeed based on e.g. solar radiation for pv panels instead of using the pre-calculated simbench timeseries.

Write time series parallely

Time series are written in their own file, therefore, they can be treated in parallel. However, this feature is currently not supported by PowerSystemDataModel.

Implement data retreival strategy

It should be checked, if the data has to be freshly downloaded or if either a completely converted data set or at least the downloaded archives are apparent.

Write mapping from participant to respective time series to file

Faulty tests in Transformer2wTypeConverterSpec

Some of the tests of in the Transformer2wTypeConverterSpec shouldn't pass as the values the actual calculated result is compared against seems to be wrongly parameterized.

For the conductance (gM) the result (as per my pen and paper calculations and implementation of the parameter calculation for the powerfactory2psdm converter) should actually be a third of the noted 2480.5790 nS.

Why do the test pass:

We compare the difference of actual and expected result against the testing tolerance. As we miss to take the absolute value of that difference the tests conveniently still passes when the expected result is much greater than the actual result.

De-serialize simbench models

Add license header to scala files

simBench2psdm/gradle/scripts/spotless.gradle

Lines 16 to 18 in 4a3213d

	scala {
	scalafmt()
	}

licenseHeader ie3LicHead, "package.*\\n"

Choose, if time series shall be created

Harmonise CI

... with other ie³ repositories.

Fix inability to determine correct folder for time series

Removing RocketChat from Jenkinsfile

Respect correct sign for assets feeding in

By convention, fed in power has a negative sign. Currently, that is not the case.