Parsing profile with 15 minute timestepping by tolga-akan · Pull Request #513 · Multi-Energy-Systems-Optimization/mesido

tolga-akan · 2026-07-15T15:21:59Z

Currently timestepping is by default 1 hour.
The goal is to support also 15min timesteps in the data, but only if the data is also required with 15minute timesteps.

Updated/check profile parsing such that it parses 15min data as 15minute.
Updated/check profile averaging for x hours/days are taking accordingly when bigger timesteps are requested.
Add test that checks the profile parsing for 15min timesteps and that the right amount of variable with correct timesteps are created
Create a test where the optimization is tested with 15min timesteps and check that there are no other strange things happening with indexing. This test should not happen with the grow workflow, but should be on a test problem and should also include a producer and electricity profile.

tolga-akan · 2026-07-16T08:16:39Z

@FJanssen-TNO @KobusVanRooyen Ready for reviewing

FJanssen-TNO · 2026-07-24T11:50:50Z

+        f"Supported: 3600 s (1 h) or 900 s (15 min)."
+    )
+
+    steps_per_hour = 3600 // timestep_spacing_seconds  # 1 for 1-h input, 4 for 15-min input


rename to "org_steps_per_hour" then it is clear that this was the number of steps per hour for the original timeserie.

suggested renaming is done

Name has not been updated as far as I can see.

You are right. I am surprised. I remember that I renamed. Probably I have reverted by mistake. Now i have renamed.

tolga-akan

@FJanssen-TNO PR is not ready yet. I will continue after vacation.

tolga-akan · 2026-07-24T14:35:52Z

+        f"Supported: 3600 s (1 h) or 900 s (15 min)."
+    )
+
+    steps_per_hour = 3600 // timestep_spacing_seconds  # 1 for 1-h input, 4 for 15-min input


suggested renaming is done

FJanssen-TNO · 2026-07-27T07:17:49Z

@tolga-akan There is good reason why this is happening. Timestep 0 is considered the history in MESIDO, so when you are creating a profile and you are taking the average over a longer period, you need to start your timeseries first with a another value, in the method you used, the first timestep is therefore copied. After all the averages should be correct.

tolga-akan · 2026-08-11T13:02:12Z

@tolga-akan There is good reason why this is happening. Timestep 0 is considered the history in MESIDO, so when you are creating a profile and you are taking the average over a longer period, you need to start your timeseries first with a another value, in the method you used, the first timestep is therefore copied. After all the averages should be correct.

yes you are right. there is no issue wit that check now

tolga-akan · 2026-08-11T13:02:29Z

@FJanssen-TNO Ready to review

FJanssen-TNO · 2026-08-12T11:41:25Z

+        f"Supported: 3600 s (1 h) or 900 s (15 min)."
+    )
+
+    steps_per_hour = 3600 // timestep_spacing_seconds  # 1 for 1-h input, 4 for 15-min input


Name has not been updated as far as I can see.

FJanssen-TNO · 2026-08-12T11:46:48Z

+        input_n_days = 0.25  # length of input profile in days
+        input_n_steps = input_n_days * 24 * (3600 / input_timestep_seconds)


please just use 8hrs*3600/input_timestep_seconds

actually i liked the way I have done before, because it was parametrizing the variable with input_n_days. input_n_days is also used at the later moment in the code. anyways, i updated the line as you recommend.

FJanssen-TNO · 2026-08-12T11:52:50Z

+                len(datetimes),
+                expected_n_output_intervals + 1,  # +1 for closing sentinel
+            )
+            dts = [datetimes[i + 1] - datetimes[i] for i in range(len(datetimes) - 1)]


Why is a nested forloop used for this? can it not be done with a vector operation?
You could use np.diff()
Also instead of datetimes you could just extract problem.times(), that will provide you with timestamps in seconds. Then the actual_step_seconds no longer needs to calcualted separately.

it can be done only via np.diff(problem.times()). redundant check is removed

FJanssen-TNO · 2026-08-12T11:56:09Z

+            np.testing.assert_array_equal(
+                expected_step_seconds,
+                actual_step_seconds,
+            )
+            np.testing.assert_array_equal(
+                expected_step_seconds,
+                np.diff(problem.times()),
+            )


I don't see the difference between these two checks.

redundant check is removed

FJanssen-TNO · 2026-08-12T11:57:21Z

+            parsed_input_data = pd.read_csv(
+                input_csv_path,
+                parse_dates=["DateTime"],
+                dayfirst=True,
+            )


Do this before the for loop over the different day_steps as it is the same both times.

you are right. Since I added the for loop later, I forgot to check if I call the variables unnecessarily many times. now the things can be called outside of the loop are moved out.

FJanssen-TNO · 2026-08-12T11:58:29Z

+                np.testing.assert_array_equal(
+                    len(averaged_profile),
+                    expected_n_output_intervals,
+                    err_msg=f"Unexpected averaged profile length for '{var_name}'",
+                )


This test is unnecessary, the next test will fail if the length of the arrays is different

yes. Thats correct. check is removed

FJanssen-TNO · 2026-08-12T12:00:05Z

+                if len(averaged_profile) == expected_n_output_intervals + 1:
+                    averaged_profile = averaged_profile[1:]


is this not required every time this test is done? Because then remove the if statement.

This if condition is unnecessary. I forgot to remove this. Previously I added this for a debugging purpose. Now if is removed

FJanssen-TNO · 2026-08-12T12:01:13Z


 ## Added
 - ESDL profile reading and writing for different types: influxdb and postgres, and inline (in ESDL file).
+- Parsing of input profiles with 15-minute interval


Move this to "changed": Parsing of input profiles with 15-minute interval now supported.

changelog is updated

FJanssen-TNO · 2026-08-12T12:02:07Z

+
+    def parameters(self, ensemble_member):
+        parameters = super().parameters(ensemble_member)
+        parameters["time_step_days"] = self.__day_steps


For what is this parameter needed?

thanks for highlight. parameter is not needed. I do not need to create time_step_days with __day_steps. Now I removed parameters method.

tolga-akan · 2026-08-12T14:06:20Z

@FJanssen-TNO Ready for reviewing

tolga-akan added 2 commits July 15, 2026 12:10

wip

2fc5973

input profile with 15 min timestepping is parsed

bb149a4

tolga-akan linked an issue Jul 15, 2026 that may be closed by this pull request

15 minute timestepping #491

Open

4 tasks

tolga-akan marked this pull request as ready for review July 15, 2026 15:22

tolga-akan requested review from FJanssen-TNO, KobusVanRooyen and jimrojerTNO as code owners July 15, 2026 15:22

rename a function and update some comments

d5c61f4

typo

c457504

FJanssen-TNO reviewed Jul 24, 2026

View reviewed changes

Updates based on PR review. WIP

792e15b

tolga-akan commented Jul 24, 2026

View reviewed changes

tolga-akan added 3 commits August 11, 2026 14:26

WIP

86a1c49

Merge branch 'main' into 491-15-minute-timestepping

6ab47f7

15 minutes averaging is added to the test

c722920

FJanssen-TNO reviewed Aug 12, 2026

View reviewed changes

tolga-akan added 2 commits August 12, 2026 15:58

updated based on pr review

c58b057

style check

df7320f

		input_n_days = 0.25 # length of input profile in days
		input_n_steps = input_n_days * 24 * (3600 / input_timestep_seconds)

		if len(averaged_profile) == expected_n_output_intervals + 1:
		averaged_profile = averaged_profile[1:]

Uh oh!

Conversation

tolga-akan commented Jul 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tolga-akan commented Jul 16, 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!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

tolga-akan left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

FJanssen-TNO commented Jul 27, 2026

Uh oh!

tolga-akan commented Aug 11, 2026

Uh oh!

tolga-akan commented Aug 11, 2026

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!

tolga-akan commented Aug 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

tolga-akan commented Jul 15, 2026 •

edited

Loading