Support DC switch resistance in load flow by SGI-Landry · Pull Request #1423 · powsybl/powsybl-open-loadflow

SGI-Landry · 2026-06-04T16:23:18Z

Please check if the PR fulfills these requirements

The commit message follows our guidelines
Tests for the changes have been added (for bug fixes / features)
Docs have been added / updated (for bug fixes / features)
A PR or issue has been opened in all impacted repositories (if any)

Does this PR already have an issue describing the problem?

Fixes #1371

Does this PR introduce a breaking change or deprecate an API?

Yes
No

Signed-off-by: Sylvestre Prabakaran <sylvestre.prabakaran@rte-france.com>

SGI-Landry · 2026-06-05T08:38:39Z

@bperr can you please review this activity? CI fails but this is normal, as long as tests run locally.

bperr · 2026-06-05T11:51:21Z

+/**
+ * @author Landry Huet {@literal <landry.huet at supergrid-institute.com>}
+ */
+public class LfDcSwitchImpl extends AbstractLfDcLine {


Should AbstractLfDcLine be renamed to AbstractLfDcBranch ? It may now represent something else than a DC line

It is a good idea. It has consequences in many files, however. In many places we then need to rename 'DcLine' to 'DcBranch'.

bperr · 2026-06-05T11:52:42Z

+
+    @Override
+    public void updateFlows(double i1, double i2, double p1, double p2) {
+        // TODO: DcSwitch has no DcTerminal yet in the IIDM model - revisit when terminals are added.


This mean that the result of the load flow cannot be propagated to the iidm Network object. Is there no workaround for this ?

You can still see the voltages at each node where the DcSwitches are connected. You do lose the data on the current. Fixing this requires however another activity in PowSyBl Core and it raises more questions. I will create a GitHub issue for this.

bperr · 2026-06-05T11:55:08Z

@@ -611,11 +613,25 @@ private static void createBranches(List<LfBus> lfBuses, LfNetwork lfNetwork, LfT

    private static void createDcLines(LfNetwork lfNetwork, LoadingContext loadingContext, LfNetworkParameters parameters) {


Similarly, this may be renamed to createDcBranches

Like many other methods and variables.

bperr · 2026-06-05T12:03:03Z


        List<LfDcBus> lfDcBuses = new ArrayList<>();
        createDcBuses(dcBuses, parameters, lfNetwork, lfDcBuses, loadingContext);
+        network.getDcSwitches().forEach(loadingContext.dcSwitchSet::add);


I understand that the DcTopologyVisitor interface does not allow to fill dcSwitchSet as for DC lines and ground so this is a workaround.
However I think your current implementation is incorrect as it adds all DC switches of the network, rather than the ones of the DC component. If your network has two DC components, the switches will be added twice.

You can either add a filter here using numDcc, or group DC Switches by (numCC, numDCC) (as done for AC switches)

bperr · 2026-06-05T12:29:47Z

@@ -0,0 +1,66 @@
+/**
+ * Copyright (c) 2025, SuperGrid Institute (http://www.supergrid-institute.com)


bperr · 2026-06-05T12:30:05Z

@@ -0,0 +1,102 @@
+/**
+ * Copyright (c) 2025, SuperGrid Institute (http://www.supergrid-institute.com)


bperr · 2026-06-05T12:34:31Z

+                .setOpen(false)
+                .setR(rSwitch)
+                .add();
+        addStandardConverters(network);


Since #1406 you have helper functions in AbstractLoadFlowNetworkFactory.java to create elements faster.

Let's discuss this. I don't see how this is less verbose.

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

… switches. Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

jeandemanged · 2026-06-29T12:53:11Z

Hi@SGI-Landry , cc @SylvestreSakti
Could you please update your branch ? (for inclusion in next release)
we do have powsybl-core RC now, and final powsybl-core should be ready soon

SGI-Landry · 2026-06-30T07:04:07Z

Hi@SGI-Landry , cc @SylvestreSakti Could you please update your branch ? (for inclusion in next release) we do have powsybl-core RC now, and final powsybl-core should be ready soon

Hello @jeandemanged I have just updated the branch. I have discovered a bug in the Core part of this activity, with a consequence in the load flow computation (see powsybl/powsybl-core#3953).

Please let's not merge the activity in the release train, but rather wait until the issue in Core is fixed.

jeandemanged · 2026-06-30T10:06:27Z

Hello @SGI-Landry ,
DC switches were made in IIDM without Terminals just like AC switches.
Not sure powsybl/powsybl-core#3953 is really a bug, rather a feature.
The needed feature is the ability to store DC flow results on DC switches. Modifying DC Switches to be DC Connectable and having DC Terminals could be one solution - but involves more work.
Some alternatives by decreasing complexity:

an IIDM Extension of DC Switch to store these results. I think the DcTerminals in DcSwitches powsybl-core#3953 proposal has an issue: DC Switches without resistance will never have results, and yet the DC Terminals are there. With extension you would put an extension only where results are relevant solving that problem.
putting results as simple IIDM properties on the DC Switches with non zero resistance. This mechanism is already used e.g. in LfBoundaryLineBus to store solved voltage magnitude and angle as properties.
not exporting the results for DC Switches in IIDM

My call would be for one of the above options (not writing, or writing as properties). If this cannot be done for this release train, I think it would make sense to do a small PR to have OLF throwing an exception on networks with DC Switch with non-zero resistance.

Let us know

SGI-Landry · 2026-06-30T14:00:04Z

Hello @SGI-Landry , DC switches were made in IIDM without Terminals just like AC switches. Not sure powsybl/powsybl-core#3953 is really a bug, rather a feature. The needed feature is the ability to store DC flow results on DC switches. Modifying DC Switches to be DC Connectable and having DC Terminals could be one solution - but involves more work. Some alternatives by decreasing complexity:
* an IIDM Extension of DC Switch to store these results. I think the [DcTerminals in DcSwitches powsybl-core#3953](https://github.com/powsybl/powsybl-core/issues/3953) proposal has an issue: DC Switches without resistance will never have results, and yet the DC Terminals are there. With extension you would put an extension only where results are relevant solving that problem.

* putting results as simple IIDM properties on the DC Switches with non zero resistance. This mechanism is already used e.g. in LfBoundaryLineBus to store solved voltage magnitude and angle as properties.

* not exporting the results for DC Switches in IIDM
My call would be for one of the above options (not writing, or writing as properties). If this cannot be done for this release train, I think it would make sense to do a small PR to have OLF throwing an exception on networks with DC Switch with non-zero resistance.

Let us know

Thanks for your answer, and thanks @SylvestreSakti for putting the exception.

The issue is more than just recording the results. Continuing the discussion on powsybl/powsybl-core#3953

SylvestreSakti added 3 commits April 27, 2026 16:22

Bump powsybl-core version to 7.3.0-SNAPSHOT

dd49d34

Signed-off-by: Sylvestre Prabakaran <sylvestre.prabakaran@rte-france.com>

Bump IIDM version to 1.17 (#1391)

6a2dc71

Signed-off-by: Sylvestre Prabakaran <sylvestre.prabakaran@rte-france.com>

Disabling 4 failing tests (to investigate) to keep CI working

1fda17d

Signed-off-by: Sylvestre Prabakaran <sylvestre.prabakaran@rte-france.com>

SGI-Landry requested a review from bperr June 4, 2026 16:23

SGI-Landry self-assigned this Jun 4, 2026

SGI-Landry added PR: do-not-merge PR: next-release needs_core_rc labels Jun 4, 2026

bperr requested changes Jun 5, 2026

View reviewed changes

SylvestreSakti and others added 8 commits June 8, 2026 18:42

Merge branch 'main' into ci/core-7.3.0-SNAPSHOT

62b58a4

First step for resistance in DcSwitch.

d173bed

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Add tests for the load flow computation.

50edaa0

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Fixes including sign error in test.

d168d72

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Fix case with open switch.

bc5ab1a

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Clean up comments and add diagrams.

69df787

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Add documentation.

ae3f567

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

fix rebase issue.

ec1ee12

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

SGI-Landry changed the title ~~Lh dc switch resistance~~ Support DC switch resistance in load flow Jun 9, 2026

SGI-Landry added 3 commits June 10, 2026 10:27

Fix rebase issue.

ba81a6b

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Rename LfDcLine to LfDcBranch for consistency between DC lines and DC…

543684b

… switches. Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Fix Copyright year.

aa9ff93

Signed-off-by: Landry Huet <landry.huet@supergrid-institute.com>

Base automatically changed from ci/core-7.3.0-SNAPSHOT to main June 23, 2026 07:10

SGI-Landry force-pushed the LH-DcSwitch-resistance branch from fe5033c to aa9ff93 Compare June 30, 2026 06:39

SylvestreSakti mentioned this pull request Jun 30, 2026

Throw exception when having DcSwitch with non zero resistance #1447

Merged

6 tasks

SylvestreSakti removed PR: do-not-merge PR: next-release labels Jun 30, 2026

SylvestreSakti added Release 09/2026 and removed needs_core_rc labels Jun 30, 2026

		@@ -611,11 +613,25 @@ private static void createBranches(List<LfBus> lfBuses, LfNetwork lfNetwork, LfT

		private static void createDcLines(LfNetwork lfNetwork, LoadingContext loadingContext, LfNetworkParameters parameters) {

		@@ -0,0 +1,66 @@
		/**
		* Copyright (c) 2025, SuperGrid Institute (http://www.supergrid-institute.com)

		@@ -0,0 +1,102 @@
		/**
		* Copyright (c) 2025, SuperGrid Institute (http://www.supergrid-institute.com)

Uh oh!

Conversation

SGI-Landry commented Jun 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

SGI-Landry commented Jun 5, 2026

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jeandemanged commented Jun 29, 2026

Uh oh!

SGI-Landry commented Jun 30, 2026

Uh oh!

jeandemanged commented Jun 30, 2026

Uh oh!

SGI-Landry commented Jun 30, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

SGI-Landry commented Jun 4, 2026 •

edited

Loading