Load Scenario File by maksud · Pull Request #46 · ORNL/ExaGO

maksud · 2026-04-23T17:19:55Z

Merge request type

New feature
Resolves bug
Documentation
Other

Relates to

This MR updates

Summary

Added an implementation of incorporating P and Q load scenarios. Created a file format for the load scenarios in: datafiles/case9/10_scenarios_9bus_load.txt For each scenario in the file, it changes the load values of the bus list mentioned in the file.

The format of each scenario is like this:

1,0.101,8_Load_1_185.0_1000,6_Load_1_105.5_2000

1 = Scenario Number

0.101 = Weight

8_Load_1_185.0_1000,6_Load_1_105.5_2000 = list of load values

8_Load_1_185.0_1000 = Set Bus 8 with P = 185.0 and Q = 1000
6_Load_1_105.5_2000= Set Bus 6 with P = 105.0 and Q = 2000

…diverging." This reverts commit e61b89b.

pelesh

I suggest target branch is the one you branched off.

I also made a few comments with suggested improvements.

pelesh · 2026-04-24T02:30:30Z

  char scenfile[PETSC_MAX_PATH_LEN];
+  char loadfile[PETSC_MAX_PATH_LEN];


Why are load files and scenario files different? Shouldn't there be single scenario file format including generation and load?

Yes, the loadfile variable name is not necessary actually. A single name is fine. I did not want to touch the variable name in the first implementation. I will drop the loadfile and use scenfile only.

Regarding Load and Generator together: It can be done but will require little bit more modification. The first cut focused on changing Load P and Q values. But I see I can use the same template to have different forecast types. Will try to submit the changes.

pelesh · 2026-04-24T02:32:14Z

+scenario_nr,weight,kv_list
+1,0.101,8_Load_1_185.0_1000,6_Load_1_105.5_2000
+2,0.102,6_Load_1_105.0_10
+3,0.103,8_Load_1_75.0_20


This is somewhat strange string format. Can't we use simple csv?

Okay, the issue is similar to the argument for CSR type of file format.

For example, say we have a test case, where we want to add a large load to each bus and say we have 30k eligible buses. So we have 30k such scenarios.

In old format, we need to have 30k columns for each CSV file, and all will have original model values except 1 column for each scenario of the existing CSV format.

In this new format, we are just keeping the bus which need to be changes.

But I agree the format is a bit strange and I am open to suggestions for input format. We can rewrite the parser accordingly.

How about we use JSON format?

@PhilipFackler, please chime in.

pelesh · 2026-04-24T02:33:23Z

+  PetscScalar *val1 = nullptr; /* forecast values */
+  PetscScalar *val2 = nullptr; /* 2nd set of forecast values */


What are forecast values and "2nd set of forecast values"? This is somewhat counterintuitive.

In current format, Forecast have a single value. But for a load, we want to change P and Q values simultaneously. So, either we can change the data type of val, or we can add add a second list of values. I used the second list here.

I am not sure we want to change P & Q simultaneously. They often change together but sometimes they don't as far as I know.

pelesh · 2026-04-24T02:34:41Z

  FORECAST_LOAD_P = 2,
-  FORECAST_LOAD_Q = 3
+  FORECAST_LOAD_PQ = 3


Why is FORECAST_LOAD_Q changed into FORECAST_LOAD_PQ?

I wanted to have the ability to change PQ value as a set. In later iterations, we can drop FORECAST_LOAD_P and use just FORECAST_GEN (after renaming FORECAST_WIND) and and FORECAST_LOAD.

I don't see why one would want to change P and Q as a set. They can and they occasionally do change separately.

@hambrickjc and @tsybinae: please chime in.

pelesh · 2026-04-24T02:36:17Z

+  ps->numits = opflow->numits;



Why numits? Why not spell it out number_iterations?

That was my doing. opflow->numits was already defined, so I kept ps->numits. I agree we should rename both.

pelesh · 2026-04-24T02:45:36Z

+  // Body Example
+  // 1,0.1,8_Load_1_185.0_10,6_Load_1_105.5_20
+  while ((out = fgets(line, MAXLINE, fp)) != NULL) {
+    if (strcmp(line, "\r\n") == 0 || strcmp(line, "\n") == 0) {
+      continue; /* Skip blank lines */
+    }


This is unintuitive way to define and parse input. Since we are creating new input file and matching parser, we don't need to use C string operations.

CC @PhilipFackler

Okay, we can have a better parser to replace this code.

pelesh · 2026-04-24T02:46:53Z

+    vector<GenData> gendata;
+
+    tok = strtok(NULL, sep_comma);
+    sscanf(tok, "%lf", &weight);


I think mixing STL and C standard library is not the best way to proceed.

Ok, I will replace the STL from here.

maksud · 2026-04-24T05:03:29Z

I suggest target branch is the one you branched off.

I also made a few comments with suggested improvements.

I tried to change the target branch but it does not look great. Should I create another MR and close this one?

nkoukpaizan · 2026-04-24T18:03:04Z

I suggest target branch is the one you branched off.
I also made a few comments with suggested improvements.

I tried to change the target branch but it does not look great. Should I create another MR and close this one?

Yeah, this looks like a rebase to develop of nicholson/testing-cmake-scopflow-fix. Let me add a rebase of my branch and see if we can target that from here.

nkoukpaizan · 2026-04-24T18:22:28Z

I suggest target branch is the one you branched off.
I also made a few comments with suggested improvements.

I tried to change the target branch but it does not look great. Should I create another MR and close this one?

Yeah, this looks like a rebase to develop of nicholson/testing-cmake-scopflow-fix. Let me add a rebase of my branch and see if we can target that from here.

@maksud You should now be able to rebase you branch on nicholson/scopflow-fix-rebased to isolate your changes.

Maksudul Alam and others added 14 commits April 9, 2026 20:11

Adding Support for Load Scenarios

5a7587a

Adding a sample load scenario

818f9a2

Print total scopflow_Nc and distinguish with nc on rank 0.

6a1449b

Don't create scopflow vectors if solver is not IPOPT.

525daa2

Fixes to run scopflow with DCOPF?

1707362

Missing change to scopflow->subproblem_solver

0fb78c5

Apply pre-commmit fixes

542ef69

Decrease max_lazy_iter to get answers in finite time without diverging.

f6beca1

Revert "Decrease max_lazy_iter to get answers in finite time without …

9945012

…diverging." This reverts commit e61b89b.

Save numits.

ef220e1

Updating to include PQ Loads

42ced77

Commenting out printfs

97dfc85

Adds CTest

85b54b0

Apply pre-commmit fixes

32e1cbd

nkoukpaizan changed the base branch from develop to nicholson/testing-cmake-scopflow-fix April 23, 2026 18:05

nkoukpaizan assigned maksud Apr 23, 2026

nkoukpaizan added enhancement New feature or request sopflow Related to stochastic optimal power flow analysis. labels Apr 23, 2026

nkoukpaizan changed the base branch from nicholson/testing-cmake-scopflow-fix to develop April 23, 2026 18:18

pelesh requested changes Apr 24, 2026

View reviewed changes

pelesh requested review from PhilipFackler and nkoukpaizan April 24, 2026 02:50

maksud changed the base branch from develop to nicholson/testing-cmake-scopflow-fix April 24, 2026 04:57

maksud changed the base branch from nicholson/testing-cmake-scopflow-fix to develop April 24, 2026 05:02

nkoukpaizan changed the base branch from develop to nicholson/scopflow-fix-rebased April 24, 2026 18:15

		char scenfile[PETSC_MAX_PATH_LEN];
		char loadfile[PETSC_MAX_PATH_LEN];

		PetscScalar val1 = nullptr; / forecast values */
		PetscScalar val2 = nullptr; / 2nd set of forecast values */

Conversation

maksud commented Apr 23, 2026

Uh oh!

pelesh left a comment

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!

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!

maksud commented Apr 24, 2026

Uh oh!

nkoukpaizan commented Apr 24, 2026

Uh oh!

nkoukpaizan commented Apr 24, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants