Remove transformer voltage controls when no PV buses at non controlled side #482

annetill · 2022-03-18T16:09:16Z

Please check if the PR fulfills these requirements (please use '[x]' to check the checkboxes, or submit the PR and then click the checkboxes)

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 ? If so, link to this issue using '#XXX' and skip the rest
Yes, fixes #481.

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

annetill · 2022-03-18T16:26:38Z

AC sensitivities support is missing.

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

src/main/java/com/powsybl/openloadflow/ac/AbstractTransformerVoltageControlOuterLoop.java

geofjamg · 2022-03-21T10:12:37Z

src/main/java/com/powsybl/openloadflow/ac/AbstractTransformerVoltageControlOuterLoop.java

+            if (componentOnNotControlledSide != null) {
+                Optional<LfBus> generatorControlledBus = componentOnNotControlledSide.stream().filter(LfBus::isVoltageControlled).findAny();
+                if (!generatorControlledBus.isPresent()) {
+                    branch.setVoltageControlEnabled(false);


Do not forget to restore it in the outer loop cleanup!

It is done as the condition is only to have a voltage control, no?

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

src/test/java/com/powsybl/openloadflow/sensi/ac/AcSensitivityAnalysisTest.java

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

annetill · 2022-04-12T12:30:17Z

For me: all is handled except in an AC sensitivity analysis, after a contingency, if a transformer becomes enable to control, the predefined results are not filled.

annetill · 2022-04-12T12:30:17Z

For me: all is handled except in an AC sensitivity analysis, after a contingency, if a transformer becomes enable to control, the predefined results are not filled.

…ontrolled Signed-off-by: Hadrien <hadrien.godard@artelys.com>

geofjamg · 2022-05-17T06:03:06Z

We should close this PR as when #522 will be done, we will only keep new the way the simulate tap changer (which is the only safe and correct way)

sylvlecl · 2022-05-17T06:54:56Z

We should close this PR as when #522 will be done, we will only keep new the way the simulate tap changer (which is the only safe and correct way)

Won't you have KLU issues in sensitivities, in incremental mode, if you don't merge this one ?

Also, if the goal is to leave all modes, not only incremental, this one will be necessary. I am not sure it's a good idea to remove other modes, since I suspect the incremental mode will be very very slow ?

annetill · 2022-05-17T06:58:40Z

We should close this PR as when #522 will be done, we will only keep new the way the simulate tap changer (which is the only safe and correct way)

Won't you have KLU issues in sensitivities, in incremental mode, if you don't merge this one ?

Also, if the goal is to leave all modes, not only incremental, this one will be necessary. I am not sure it's a good idea to remove other modes, since I suspect the incremental mode will be very very slow ?

I have tested the incremental mode on CGMES IOP files, and it is clearly a beta feature. I really need to spend more time on it, but I don't have this time for the moment. In that PR, @Hadrien-Godard and me are also working on voltage sensitivities. So we will continue.

geofjamg · 2022-05-17T07:05:08Z

We should close this PR as when #522 will be done, we will only keep new the way the simulate tap changer (which is the only safe and correct way)

Won't you have KLU issues in sensitivities, in incremental mode, if you don't merge this one ?
Also, if the goal is to leave all modes, not only incremental, this one will be necessary. I am not sure it's a good idea to remove other modes, since I suspect the incremental mode will be very very slow ?

I have tested the incremental mode on CGMES IOP files, and it is clearly a beta feature. I really need to spend more time on it, but I don't have this time for the moment. In that PR, @Hadrien-Godard and me are also working on voltage sensitivities. So we will continue.

Every unrelated changes on voltage sensi have to move in another PR

geofjamg · 2022-05-17T07:06:21Z

We should close this PR as when #522 will be done, we will only keep new the way the simulate tap changer (which is the only safe and correct way)

Won't you have KLU issues in sensitivities, in incremental mode, if you don't merge this one ?

Also, if the goal is to leave all modes, not only incremental, this one will be necessary. I am not sure it's a good idea to remove other modes, since I suspect the incremental mode will be very very slow ?

Not significantly slower than with this mode.
Only one or two additional LU factorizations, in case we target directly the final tap.

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

sylvlecl · 2022-05-17T08:24:28Z

Not significantly slower than with this mode.
Only one or two additional LU factorizations, in case we target directly the final tap.

Ok, from the text of the PR I thought it was only +/-1 increments, but I understand it's not the case.

sylvlecl · 2022-05-19T07:52:59Z

The previous discussion made me think that instead of the topological analysis, we could rely on a numerical analysis for the identification of those transfos.
Those would be the ones for which voltage sensitivity to the ratio is zero (epsilon).

flo-dup · 2022-05-25T11:03:12Z

src/main/java/com/powsybl/openloadflow/ac/AbstractTransformerVoltageControlOuterLoop.java

+                Optional<LfBus> generatorControlledBus = componentOnNotControlledSide.stream().filter(LfBus::isVoltageControlled).findAny();
+                if (!generatorControlledBus.isPresent()) {


You can replace these two lines by if (componentOnNotControlledSide.stream().noneMatch(LfBus::isVoltageControlled)) {

flo-dup · 2022-05-25T12:25:44Z

src/main/java/com/powsybl/openloadflow/ac/AbstractTransformerVoltageControlOuterLoop.java

@@ -47,4 +51,41 @@ protected OuterLoopStatus roundVoltageRatios(OuterLoopContext context) {
        }
        return status;
    }
+
+    public static void checkControl(LfNetwork network) {
+        var connectivity = network.getConnectivity();


I'm wondering if using the EvenShiloachGraphDecrementalConnectivity here is consistent, as the init is costly.

If we launch a security analysis, that's not a question as the init will be reused.

But if we only launch a loadflow, that's not really appropriate, and the naive approach is 4 times faster.

If we keep it this way, the performance issue has been fixed by #537. It was because we get the main connected component many times. And EvenShiloach algorithm recomputes the main connected component each time it is asked (it was first implemented for use cases where only the small components were needed).

I'm wondering if using the EvenShiloachGraphDecrementalConnectivity here is consistent, as the init is costly.

If we launch a security analysis, that's not a question as the init will be reused.

But if we only launch a loadflow, that's not really appropriate, and the naive approach is 4 times faster.

This piece of code is ran both for a simple LF and a SA. We probably should use naive implementation for the LF and EvenShiloach for the SA but in both cases it has to be fast so it's a good thing that you improved performance for EvenShiloach.

I also think that performance issue was more visible for the SA than simple LF.

Signed-off-by: Geoffroy Jamgotchian <geoffroy.jamgotchian@gmail.com>

annetill · 2022-05-30T07:41:00Z

Performances are great now!

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

flo-dup · 2022-05-30T08:53:02Z

src/main/java/com/powsybl/openloadflow/network/LfNetwork.java

-                            branch.getId());
-                    disabledTransformerCount++;
-                }
+            if (componentOnNotControlledSide != null && componentOnNotControlledSide.stream().noneMatch(LfBus::isVoltageControlled)) {


Is there any chance we're looping many times over a big set of LfBus? Then we should save the results for each component.
But it would mean that there is a big (main) connected component with many controller branches with the corresponding controlled buses outside that component. And no other controlled bus. Not very likely I suppose.

Yes, it's likely:
typically you will have a very large component for EHV (400-225 kV), on which will be connected most transformers that regulate on the lower level (90-63 kV in France).

I added a commit to save the result for each component, it's supposed to be a bit faster... or at least more robust. It seems that in practice it's not faster as in big components, the noneMatch encounters always quickly a controlled bus.

src/main/java/com/powsybl/openloadflow/network/LfNetwork.java

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Signed-off-by: Florian Dupuy <florian.dupuy@rte-france.com>

sonarqubecloud · 2022-05-30T13:27:03Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

98.2% Coverage
0.0% Duplication

Remove transformer voltage controls.

f3b81d9

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

annetill requested review from geofjamg and sylvlecl March 18, 2022 16:09

Support of contingencies.

041bfff

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Add a unit test to show how AC sensis behave.

50aa78a

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

sylvlecl approved these changes Mar 21, 2022

View reviewed changes

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

geofjamg requested changes Mar 21, 2022

View reviewed changes

annetill and others added 3 commits March 21, 2022 14:42

Fix.

bbbc949

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Merge branch 'main' into fix-regleurs-enable-to-control

c2e5ffe

Merge branch 'main' into fix-regleurs-enable-to-control

87e7861

geofjamg changed the title ~~Remove transformer voltage controls when no PV buses at non controlled side~~ [WIP] Remove transformer voltage controls when no PV buses at non controlled side Apr 2, 2022

annetill commented Apr 6, 2022

View reviewed changes

src/test/java/com/powsybl/openloadflow/sensi/ac/AcSensitivityAnalysisTest.java Show resolved Hide resolved

annetill added 4 commits April 12, 2022 09:39

Merge.

e998470

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Fix AC sensitivity analysis for pre contingency computation.

bdb8d6c

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Fix code smells.

70aadf9

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Fix code smell.

e654eaf

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

SKIP instead of ZERO for sensitivity when controlled bus is no more c…

11214c3

…ontrolled Signed-off-by: Hadrien <hadrien.godard@artelys.com>

Merge.

0d551b3

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

annetill changed the title ~~[WIP] Remove transformer voltage controls when no PV buses at non controlled side~~ Remove transformer voltage controls when no PV buses at non controlled side May 17, 2022

Hadrien-Godard approved these changes May 17, 2022

View reviewed changes

annetill added the PR: next-release label May 18, 2022

Merge branch 'main' into fix-regleurs-enable-to-control

c1929d4

flo-dup requested changes May 25, 2022

View reviewed changes

geofjamg added 3 commits May 29, 2022 21:34

Refactor

d464b2b

Signed-off-by: Geoffroy Jamgotchian <geoffroy.jamgotchian@gmail.com>

Clean

3f34fb5

Signed-off-by: Geoffroy Jamgotchian <geoffroy.jamgotchian@gmail.com>

Improve logging

d169bbc

Signed-off-by: Geoffroy Jamgotchian <geoffroy.jamgotchian@gmail.com>

geofjamg approved these changes May 29, 2022

View reviewed changes

Merge branch 'main' into fix-regleurs-enable-to-control

e468ed2

annetill added 2 commits May 30, 2022 10:10

Apply review comment.

a37c3a8

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Fix code smell.

f23b250

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

flo-dup reviewed May 30, 2022

View reviewed changes

annetill and others added 2 commits May 30, 2022 12:17

Apply review comment.

55e075b

Signed-off-by: Anne Tilloy <anne.tilloy@rte-france.com>

Save no PV buses information using a Map

42f5a70

Signed-off-by: Florian Dupuy <florian.dupuy@rte-france.com>

flo-dup force-pushed the fix-regleurs-enable-to-control branch from 503f0c4 to 42f5a70 Compare May 30, 2022 13:24

flo-dup approved these changes May 30, 2022

View reviewed changes

flo-dup merged commit 9b6203b into main May 30, 2022

flo-dup deleted the fix-regleurs-enable-to-control branch May 30, 2022 14:37

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove transformer voltage controls when no PV buses at non controlled side #482

Remove transformer voltage controls when no PV buses at non controlled side #482

annetill commented Mar 18, 2022 •

edited by flo-dup

Loading

annetill commented Mar 18, 2022

geofjamg Mar 21, 2022

annetill Mar 21, 2022

annetill commented Apr 12, 2022

annetill commented Apr 12, 2022

geofjamg commented May 17, 2022

sylvlecl commented May 17, 2022

annetill commented May 17, 2022

geofjamg commented May 17, 2022

geofjamg commented May 17, 2022 •

edited

Loading

sylvlecl commented May 17, 2022

sylvlecl commented May 19, 2022

flo-dup May 25, 2022

flo-dup May 25, 2022 •

edited

Loading

flo-dup May 25, 2022

geofjamg May 29, 2022 •

edited

Loading

annetill commented May 30, 2022

flo-dup May 30, 2022

sylvlecl May 30, 2022

flo-dup May 30, 2022

sonarqubecloud bot commented May 30, 2022

		Optional<LfBus> generatorControlledBus = componentOnNotControlledSide.stream().filter(LfBus::isVoltageControlled).findAny();
		if (!generatorControlledBus.isPresent()) {

Remove transformer voltage controls when no PV buses at non controlled side #482

Remove transformer voltage controls when no PV buses at non controlled side #482

Conversation

annetill commented Mar 18, 2022 • edited by flo-dup Loading

annetill commented Mar 18, 2022

geofjamg Mar 21, 2022

Choose a reason for hiding this comment

annetill Mar 21, 2022

Choose a reason for hiding this comment

annetill commented Apr 12, 2022

annetill commented Apr 12, 2022

geofjamg commented May 17, 2022

sylvlecl commented May 17, 2022

annetill commented May 17, 2022

geofjamg commented May 17, 2022

geofjamg commented May 17, 2022 • edited Loading

sylvlecl commented May 17, 2022

sylvlecl commented May 19, 2022

flo-dup May 25, 2022

Choose a reason for hiding this comment

flo-dup May 25, 2022 • edited Loading

Choose a reason for hiding this comment

flo-dup May 25, 2022

Choose a reason for hiding this comment

geofjamg May 29, 2022 • edited Loading

Choose a reason for hiding this comment

annetill commented May 30, 2022

flo-dup May 30, 2022

Choose a reason for hiding this comment

sylvlecl May 30, 2022

Choose a reason for hiding this comment

flo-dup May 30, 2022

Choose a reason for hiding this comment

sonarqubecloud bot commented May 30, 2022

annetill commented Mar 18, 2022 •

edited by flo-dup

Loading

geofjamg commented May 17, 2022 •

edited

Loading

flo-dup May 25, 2022 •

edited

Loading

geofjamg May 29, 2022 •

edited

Loading