ie3-institute / simona

simona is an agent-based discrete-event power system simulation model developed @ie3-institute

SIMONA

The agent-based simulation environment SIMONA provides a simulation toolbox to run and implement large-scale agent-based electricity grid simulations with focus on distribution grids. As a result, close-to-reality time series are generated from various system participants and grid assets that can be used to analyze a given power grid. Application cases are for example distribution grid planning purposes but also flexibility analysis or coupled sector interdependency analysis. The framework contains several out-of-the-box available models for a wide variety of grid participants as well as their operational behavior.

More information are provided in the project's documentation.

Usage and Contribution

SIMONA is part of several ongoing research projects and will be part of future research projects. Hence, the codebase is continuously under development from different perspectives, needs and developers.

We invite everyone to use SIMONA for their own research or for usage in a research project. If you use SIMONA for your own projects or research, please provide a reference to this repository. Furthermore, if you publish your scientific work please give appropriate credit by citing one of the introduction papers of SIMONA.

We're also happy for any feedback and contributions. For details on how to contribute, please take a look at the CONTRIBUTING.md file in the root directory of this repository.

Questions

For all SIMONA related questions please feel free to contact people involved in the development and maintenance of SIMONA. For the moment, these people are:

• Feismann, Daniel - [email protected]
• Peter, Sebastian - [email protected]
• Oberließen, Thomas - [email protected]
• Sen Sarma, Debopama - [email protected]
• Bao, Johannes - [email protected]
• Hohmann, Julian - [email protected]
• Kittl, Chris - [email protected]
• Hiry, Johannes - [email protected]

simona's Issues

Simplify SimonaConfig with shared param object

Parameter structures like sqlParams, csvParams are very similar (if not identical) across input types (currently primary and weather).

Shared parameters like timePattern could also be provided by a common trait/superclass.

Depending on carueda/tscfg#126, these can be implemented as optional shared objects.

Ignore snapshot dependencies

... for dependabot.

Correct calculation logic of three winding transformer

Support for experiment runs

Currently, we cannot run multiple configuration files in one simulation run. However, it would be nice to sequentially process multiple configurations instead of just one per execution.

Such a multi-configuration support should:

• be configurable (point to exact location with one config per line)
• create new logs in the corresponding config folder + disable console logging automatically and point to the log location to avoid a too clumsy console (e.g. "Started simulation XYZ. Full log can be found here: <PATH_TO_LOG>)
• Optionally (by config arg) aggregate results

Fix unique generation of ActorNames

On very large datasets the current implementation of the actor naming uuid tends to be not unique anymore. Therefore we need to find an alternative for the unique name generation.

Possible approaches:

• remove substring(0,8) (might clutter logs...)
• find another method to generate unique identifiers (maybe hash values?

Affected line:

Fix ConfigFailFastSpec:throw an exception if an influxDb1x is configured, but not accessible

The ConfigFailFastSpec for influxDb1x is currently disabled as it seems there is an dependency issue with a kotlin package (kotlin-stdlib) which is used in PSDM and subsequently in influxDB-java connector. This may also affect the functionality of the influxDb functionality in simona as well and is therefore considered as a major issue.

To solve this issue I assume solving the dependency issue should be sufficient. As this may kills simona's influxDB1x capabilities it is assumed to be a high priority bug and should be addressed asap!

I disabled the test for now.

Tasks:

• solve issue
• enable test again(!)

Let participant agents die on failed registration

Currently, the don't...

Add publish and add list of references to the README

We need to put all publications with regard to SIMONA at one place and link them in the reference to simplify to cite SIMONA.

Synchronous test for grid agent in center position

As of #111, the GridAgent resp. DBFSAlgorithm uses the dispatcher from actor context. For execution of power flow calculation.

However, using the TestFSMRef for synchronous testing, the configured dispatcher is explicitly overridden to `CallingThreadDispatcher' utilizing only the calling thread. However, this poses the risk of dead locks, which we obviously fell in.

Synchronous test currently is disabled, as we intend to refactor to akka typed and therefore the test also needs refactoring. However, if that is done, re-implement a synchronous test!

Setup shadow jar release deployment

In contrast to your other projects, simona is a "standalone" application which cannot be run effortless as "light" jar. Therefore we need to find a way to not only release simona @ maven central, but also to release fat/shadow jars here on github. Currently, the CI/CD pipeline does not support such a feature. Therefore such a setup, maybe even in conjuction with a docker deployment to ship the simona jar inside a docker container, is required.

This task includes a proper version management of shadow jars of main and dev branches - which means normal or snapshot shadow jar version naming and must also enable the currently ignored DeploySpec.scala

Tasks

• fix and enable DeploySpec.scala

Repeat simulation multiple times

Switch to scala 3 + java 17

Scala 3 has several improvements over the currently used scala version. Even more, in the mid-term I expect that we will run into several problems if we stick to scala 2.

We may consider using this version change to also switch to the latest java LTS which is currently java 17.

For a full switch, we depend on updates of the following tools to support scala 3

• Scoverage
• Scapegoat

We should also consider using this change for a new, actually the first public, release.
Therefore it might make sense to solve #17 before.

PrimaryDataService.scala is unused and could possibly be removed

Looks like edu.ie3.simona.agent.participant.data.primary.PrimaryDataService is not used anywhere.

Make SimonaConfig.BaseRuntimeConfig serializable.

With the new implemented functionality within tscfg, generated classes can now inherit from traits.
This means that it's now possible to make BaseRuntimeConfig and subclasses serializable.

Implement agent for combined heat and power plants

Model is ready yet, but there is no specific agent for it.

As the ChpModel produces electric and thermal heat output at the same time, please take special care of the scaling factor. This is not yet considered in the existing other agents.

Add config for transformer control groups

Configuration has to comprise for each control group:

• List of controlled transformers (by uuid)
• List of measurement devices to account for (by uuid)
• Specification of control scheme (e.g. limit violation prevention vs. preventive control)
• Parameters for control scheme (thresholds, dead bands etc.)

Adapt to changed `PvInputModel`

As of PowerSystemDataModel the parameter height changed to elevationAngle

Implement SQL input for time series mapping

Analogous to #34, SQL input of time series mappings is needed in PrimaryServiceProxy.

Fix MavenCentral deployment credits

Jenkins complains about insecure string interpolation when deploying to maven central. While the credentials are masked and therefore this should be safe to use, we should investigate another way to pass in the credentials there.

Should be doable fast forward, as we use credentials beforehand already and therefore one may just apply the concept of an existing credentials implementation.

If you start working on this, feel free to contact me for details.

Fix location of "vn_simona" test grid

The grid nodes are currently partly in Dortmund and Berlin

Capture meta information for power flow

Feature implemented by @johanneshiry

Introduce distributed simulation

As part of my master's thesis, implement the capability to run SIMONA as a distributed simulation. Local execution should still be possible.

Change transformer calculation

As of PowerSystemDataModel > 3.0 the no load susceptance of transformers (both two and three winding) will come negative. The automatic negation here should be adapted.

Adapt DBFS communication protocol

... to account for multiple connections between same GridAgents.

PlantUML logging of messages for easy debugging

@johanneshiry made me aware of the following:

Within the Beam project they log messages in a PlantUML format including states of the agents and ticks of their change.
This seems to be a genius way for easy debugging of complex message protocols.

Here are their corresponding docs and here is the corresponding PR

This might be a worthy feature if our message protocols get more complex. As of Johannes' quick assessment it seems like the code and concept can be relatively easily applied to SIMONA

Consider setting java version to 11

We currently depend on an outdated and possibly insecure version of jGraphT (1.4) as with jGraphT 1.5 java 11 is required. Therefore, as this should not change much, we should consider setting the projects java version from 1.8 to 11

May be considered in conjunction with PSDM#307

Implement agent for biomass power plants

Model is ready yet, but there is no specific agent for it.

As the BmModel produces electric and thermal heat output at the same time, please take special care of the scaling factor. This is not yet considered in the existing other agents.

Transferring model documentation from internal wiki

Some documentation has not been transferred from internal simona wiki yet.
Information should be checked for up-to-dateness and adapted for readthedocs.

• PvModel
• BmModel
• LoadModel: Move to Models section
• Transformer/3W Transformer
• LoadProfiles

Hierarchical scheduling for distributed computing

Introduce hierarchical scheduling as the second part of my master's thesis

Run log filter threshold from xml

Take over the filters from defined standard file appender from logback xml-config, if applicable. Else log with "INFO"-Level

Properly account for scaling factors in participants

Currently, the scaling factor is not properly accounted for. It should be checked, where it can be integrated best and what logic is best (first scaling active power and then calculate reactive power based on that one or scaling both results?)

Simplify PrimaryServiceProxy with new version of PSDM

With recent changes to ColumnScheme in PSDM, the PrimaryServiceProxy can be simplified a bit.

Implement measurement and control models

Fix weather source default resolution

Implement SQL input for primary data

In PrimaryServiceWorker, currently only input from CSV is implemented.

Depends on PSDM#467.

Implement Kafka output for SIMONA results

Currently, only node and line results are required.

• Also, provide Dockerfile for SIMONA with variable input data

Simplify initialization of PrimaryServiceWorker

PrimaryServiceProxy#toInitData unpacks source-related primary config and re-packages it into an InitPrimaryServiceStateData, which is then unpacked in PrimaryServiceWorker#init again. This seems a bit unnecessary.

One could take WeatherSource#apply as an example, where configs are turned into WeatherSourceWrappers directly.

This becomes more relevant once #34 is implemented.

Change Getting Started Documentation to Public-Version

Reference in Getting-Started Documentation is linked to internal SIMONA version

Installation
You can find and download the source code of the latest stable SIMONA version here. Go ahead and clone the repository using git:
$ git clone https://git.ie3.e-technik.tu-dortmund.de/SIMONACrew/SIMONA.git

Respect for voltage angle in distributed power flow

Currently, when an actor receives the slack node voltage from it's superior grid agent, the angle of the voltage is not accounted for.

Adapt to new simonaAPI snapshot

Switch to akka typed

The current implementation relies on akka classic which does not include type safety for message passing between actors. In the mid-term it would be very helpful to have akka typed support in order to have a more safe and stable protocol handling. This also might require some protocol adaptions and it may be helpful to have a documentation of the existing protocols available beforehand.

Tasks

• add documentation for existing agent and actor communication protocols
• adapt existing protocols for akka typed
• adapt the code to support akka typed

Fix invalid thread allocation in GridAgent

The current implementation of the GridAgent uses its own implementation of an ExecutionContext. However, this is a) not needed and b) leads to a huge amount of allocated Threads when an extensive grid is simulated on a large computational machine. Therefore, either the custom ExecutionContext should be replace by the dispatcher context OR the (somehow unrequired) Future should be removed from this actor as well.

Cleanup gradle.properties when spotless is fixed for java 17

Same as here

Participants write out results

... that have be messaged to the grid agents and therefore have been part of the power flow calculation.

The power flow is carried out in a fixed time resolution, whereas the participant act on different time scales. Therefore, for power flow calculation, the participant averages the simulation results over a certain time frame and messages it back to the grid agent. This result is not yet enclosed within the output data.

Add smart charging logic

... for electric vehicles in conjunction with a distinct mobility simulation.

Upgrade to Java 17

In order to re-enable SonarQube analysis during CI, Java needs to be upgraded.
As part of #53

Re-organizing test resources to packages

Test data can be moved to the respective packages that it is needed in. This way, it is more obvious to which test the data belongs and vice versa. Also, getClass.getResource() works well with this, as only the path relative to the own package path needs to be stated.

For example, the contents of it-data/primaryService can be moved into some directory in edu/ie3/simona/service/primary. The full path then does not need to be stated when using getClass.getResource().

As discussed with @t-ober, and @johanneshiry @ckittl you are welcome to give your opinion about this as well.

Move parking lot request out of calculatePowerWithSecondaryDataAndGoToIdle

Currently, EvFreeLotsRequest are handled within calculatePowerWithSecondaryDataAndGoToIdle of EvcsAgentFundamentals.

Free lot requests should be handled inside its own method.

Implement termination message for external simulation

Once the main simulation ends, it should inform the external sim about it via a termination message. Needs to implemented in SimonaAPI as well.

Fix bad groovy formatting

Spotless groovy formatting leads to weird results with multiline field definitions (here from RandomLoadModelTest):

	def loadInput =
	new LoadInput(
	UUID.fromString("4eeaf76a-ec17-4fc3-872d-34b7d6004b03"),
	"testLoad",
	OperatorInput.NO_OPERATOR_ASSIGNED,
	OperationTime.notLimited(),
	new NodeInput(
	UUID.fromString("e5c1cde5-c161-4a4f-997f-fcf31fecbf57"),
	"TestNodeInputModel",
	OperatorInput.NO_OPERATOR_ASSIGNED,
	OperationTime.notLimited(),
	Quantities.getQuantity(1d, PU),
	false,
	NodeInput.DEFAULT_GEO_POSITION,
	GermanVoltageLevelUtils.LV,
	-1
	),
	new CosPhiFixed("cosPhiFixed:{(0.0,0.95)}"),
	H0,
	false,
	Quantities.getQuantity(3000d, KILOWATTHOUR),
	Quantities.getQuantity(282.74d, VOLTAMPERE),
	0.95
	)

Could be solved by playing around with the formatting settings and enabling them in spotless.gradle.