[WIP] Automatic fix of incompatible target voltages #1115

geofjamg · 2024-11-06T15:03:06Z

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)

Does this PR already have an issue describing the problem?

What kind of change does this PR introduce?

What is the current behavior?

What is the new behavior (if this is a feature change)?

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

Yes
No

If yes, please check if the following requirements are fulfilled

The Breaking Change or Deprecated label has been added
The migration steps are described in the following section

What changes might users need to make in their application due to this PR? (migration steps)

Other information:

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

vidaldid-rte · 2024-11-15T12:35:16Z

src/main/java/com/powsybl/openloadflow/OpenLoadFlowParameters.java

@@ -276,6 +276,8 @@ public enum FictitiousGeneratorVoltageControlCheckMode {

    public static final String AREA_INTERCHANGE_P_MAX_MISMATCH_PARAM_NAME = "areaInterchangePMaxMismatch";

+    public static final String FIX_TARGET_VOLTAGE_INCOMPATIBILITY_PARAM_NAME = "fixTargetVoltageIncompatibility";
+


To be documented.
And maybe addiotional parameters given like TARGET_VOLTAGE_PLAUSIBILITY_THRESHOLD

However in my test networks, increasing CONTROLLED_BUS_NEIGHBORS_EXPLORATION_DEPTH had a terrible algotirhmic cost and didn't improve quality (defined as more convergence)

"And maybe addiotional parameters given like TARGET_VOLTAGE_PLAUSIBILITY_THRESHOLD" => yes
for CONTROLLED_BUS_NEIGHBORS_EXPLORATION_DEPTH indeed it should not be configurable, a value of 2 or 3 is probably enough.

src/main/java/com/powsybl/openloadflow/TargetVoltageCompatibilityChecker.java

vidaldid-rte · 2024-11-15T15:29:03Z

src/main/java/com/powsybl/openloadflow/adm/AdmittanceEquationSystem.java

+            PiModel piModel = branch.getPiModel();
+            if (FastMath.abs(piModel.getX()) < LfNetworkParameters.LOW_IMPEDANCE_THRESHOLD_DEFAULT_VALUE) {
+                if (bus1 != null && bus2 != null) {
+                    LOGGER.warn("Non impedant branches ({}) not supported in admittance matrix",


What is the impact of ignoring those branches ? Could this lead to failure to detect inconsistent voltage target in nearby stations connected by short lines ?
Wouldn't it be better to merge the buses in the graph ?

Fixed, I cut the impedance using the LfNetworkParameters.LOW_IMPEDANCE_THRESHOLD_DEFAULT_VALUE which is ok for what we are doing with admittance matrix.

vidaldid-rte · 2024-11-15T15:46:49Z

src/main/java/com/powsybl/openloadflow/adm/AdmittanceMatrix.java

+    public static AdmittanceMatrix create(AdmittanceEquationSystem ySystem, MatrixFactory matrixFactory) {
+        Objects.requireNonNull(matrixFactory);
+        try (var j = new JacobianMatrix<>(ySystem.getEquationSystem(), matrixFactory)) {
+            return new AdmittanceMatrix(ySystem.getEquationSystem(), j.getMatrix());


Wouldn't it be cleaner to inherit from JacobianMatrix ?
Here if a future evolution of JacobianMatrix closes the Matrix object the code will break because it assumes today that the matrix is still alive.

Admittance matrix has nothing to do with Jacobian matrix even if we use it to compute admittance.

Well - you reused the matrix factory part. And steal the matrix of a closable - in addition a matrix with a DirectBuffer in it. That works today, and in your code structure but is not really a reusable object. Not a big deal and not reason to disapprove. But I can't stay silent when I see this :-)

vidaldid-rte · 2024-11-15T15:57:10Z

src/main/java/com/powsybl/openloadflow/adm/AdmittanceMatrix.java

+        var i1yEq = equationSystem.getEquation(bus1.getNum(), AdmittanceEquationType.BUS_ADM_IY).orElseThrow();
+        var i2xEq = equationSystem.getEquation(bus2.getNum(), AdmittanceEquationType.BUS_ADM_IX).orElseThrow();
+        var i2yEq = equationSystem.getEquation(bus2.getNum(), AdmittanceEquationType.BUS_ADM_IY).orElseThrow();
+        if (e == null) {


This is definitely not MT safe. Not only that it could lead to double allocation. But because the matrix is reused.
Can you document the limitation in the class ?

Indeed but I don't think it is expected to be thread safe as all of other similar OLF linear algebra utilities.

If I didn-t know the underlyings I would expect it to be threadsafe. Maybe a general problem to OLF.
But running it in MT mode requires a lot of expertis today (see previous PR an MT security analysis).
Anyway, out of scope of ths feature.

vidaldid-rte · 2024-11-15T16:03:14Z

src/main/java/com/powsybl/openloadflow/TargetVoltageCompatibilityChecker.java

+
+        var ySystem = AdmittanceEquationSystem.create(network, new VariableSet<>());
+        try (var y = AdmittanceMatrix.create(ySystem, parameters.getMatrixFactory())) {
+            for (LfBus controlledBus : controlledBuses) {


Don't you compute the impedance between each pair of bus twice ? if you iterate on all buses then on all neighbours ?

Also, would it accerlerate things if there was an option to check a pair only if at least one of the two buses is remotely controlled ? [This is where we have convergence issues]

Yes, we are checking potentially twice. It could be improved indeed.

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

sonarcloud · 2024-11-17T11:28:54Z

Quality Gate failed

Failed conditions
46.9% Coverage on New Code (required ≥ 90%)

See analysis details on SonarQube Cloud

geofjamg · 2024-11-17T11:40:18Z

@vidaldid-rte @SylvestreSakti I added in addition to "incompatible target v" check a new "unrealistic target v" which rely on target voltage to calculate voltage sensitivities (already existing in OLF) to estimate to controller bus voltage given a controlled bus targe voltage and discard when too far from 1pu. On my real test case it works very well.

vidaldid-rte · 2024-11-19T16:42:47Z

src/main/java/com/powsybl/openloadflow/RemoteVoltageTargetChecker.java

+        return findElementsToDiscardFromVoltageControl(network, parameters, new SparseMatrixFactory());
+    }
+
+    public static Set<String> findElementsToDiscardFromVoltageControl(Network network, LoadFlowParameters parameters, MatrixFactory matrixFactory) {


Is it possible to add a reason ?
Incompatible with XXX
Irrealistic

(It seems at this time there is even no log to understand the discard recommendation)

vidaldid-rte · 2024-11-19T16:51:25Z

src/main/java/com/powsybl/openloadflow/RemoteVoltageTargetChecker.java

+        return (EquationTerm<AcEquationType, AcEquationType>) controllerBus.getCalculatedV();
+    }
+
+    private void checkUnrealisticTargets(RemoteVoltageTargetCheckerParameters parameters,


Very good idea !

geofjamg and others added 14 commits November 4, 2024 20:34

Bad target voltage detection

b6d59c7

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

6aa5b8a

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

e62d1b8

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

965ae71

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

ea0be1f

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

b211915

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

ab8e4b1

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Merge branch 'main' into bad_target_v_detection

50dd365

Wip

021766d

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

fe48f54

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Add checker

486201c

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Fix

f03e043

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Fix

bf367f8

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Add perf log

499e6fb

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

vidaldid-rte reviewed Nov 15, 2024

View reviewed changes

src/main/java/com/powsybl/openloadflow/TargetVoltageCompatibilityChecker.java Outdated Show resolved Hide resolved

vidaldid-rte reviewed Nov 15, 2024

View reviewed changes

geofjamg added 10 commits November 15, 2024 21:38

Merge branch 'main' into bad_target_v_detection

ba46ea8

Add unrealistic target voltage check

a4563b1

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

660c2e4

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Wip

e978193

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Clean

14acfb3

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Clean

7ea9f6f

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Clean

9488317

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Improve logs

cb87f85

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Add method to directly check on IIDM network

70c4326

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Fix

ac807c9

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

geofjamg added 3 commits November 17, 2024 11:00

Fix

6fa1020

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Revert

1557c2c

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

Support low impedance branches in adm matrix calculation

bf94e9c

Signed-off-by: Geoffroy Jamgotchian <[email protected]>

vidaldid-rte mentioned this pull request Nov 19, 2024

Reduce irrealistic voltage created by remote voltage control #1124

Open

vidaldid-rte reviewed Nov 19, 2024

View reviewed changes

vidaldid-rte mentioned this pull request Nov 22, 2024

Check too far voltage remote control #1119

Open

5 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[WIP] Automatic fix of incompatible target voltages #1115

[WIP] Automatic fix of incompatible target voltages #1115

geofjamg commented Nov 6, 2024

vidaldid-rte Nov 15, 2024

geofjamg Nov 17, 2024

vidaldid-rte Nov 15, 2024

geofjamg Nov 17, 2024

vidaldid-rte Nov 15, 2024 •

edited

Loading

geofjamg Nov 17, 2024

vidaldid-rte Nov 19, 2024

vidaldid-rte Nov 15, 2024 •

edited

Loading

geofjamg Nov 17, 2024

vidaldid-rte Nov 19, 2024

vidaldid-rte Nov 15, 2024

geofjamg Nov 17, 2024

sonarcloud bot commented Nov 17, 2024

geofjamg commented Nov 17, 2024

vidaldid-rte Nov 19, 2024

vidaldid-rte Nov 19, 2024

		@@ -276,6 +276,8 @@ public enum FictitiousGeneratorVoltageControlCheckMode {

		public static final String AREA_INTERCHANGE_P_MAX_MISMATCH_PARAM_NAME = "areaInterchangePMaxMismatch";

		public static final String FIX_TARGET_VOLTAGE_INCOMPATIBILITY_PARAM_NAME = "fixTargetVoltageIncompatibility";

[WIP] Automatic fix of incompatible target voltages #1115

Are you sure you want to change the base?

[WIP] Automatic fix of incompatible target voltages #1115

Conversation

geofjamg commented Nov 6, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vidaldid-rte Nov 15, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vidaldid-rte Nov 15, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sonarcloud bot commented Nov 17, 2024

Quality Gate failed

geofjamg commented Nov 17, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vidaldid-rte Nov 15, 2024 •

edited

Loading

vidaldid-rte Nov 15, 2024 •

edited

Loading