Running MUSE

[Mia McLachlan]

Hi all, hope you had a good weekend!

MUSE has been installed correctly as far as we are aware. We have completed the example tutorial and also experimented with changing the inputs by expanding the technologies in the example to include solar. When visualising the results in Jupyter Notebooks, however, the output data does not match that presented in the tutorial. 

The inputs are identical so we wondered if there was a problem with the installation. Alternatively, we wondered if development of the model since the creation of the example tutorial may be causing the conflicting results. 

Many thanks,

Mia and Avi

[sara giarola]

Dear all,

there is a small mismatch in the version you are using (version 0.9) and the results in the documentation (~version 0.7). So yes, the results could be slightly different.

At this stage, please make sure that you get familiar with the possibilities you have of making changes and how you can make them as documented.

Best

[Mia McLachlan]

Thanks for the response, that makes makes sense. 

I am currently struggling with adding a second region. The documentation provides links for the complete input tables but when I try to follow the links they are broken. Do you have access to the full files?

Many thanks

Mia 

[Alexander Kell]

Apologies for the broken links. This is something we will fix once MUSE open source has been released.

Attached are the model inputs for each of the respective examples. They are in order and should be relatively straightforward, but let us know if you'd like any further help.

[Mia McLachlan]

Thanks Alex, I managed to fix the issues by comparing my inputs to these files. The simulation now runs successfully but indicates that the problem is infeasible. We also ran the same simulation on Avi's computer with the files from the tutorial directly copied across but the same message appears. Could there be something else we are doing wrong or may the update that Sara mentioned be causing the infeasibility?

Many thanks, 

Mia and Avi 

[Alexander Kell]

The infeasibility error is a common one. More information can be found in the FAQ section of the documentation: https://museenergydocs.readthedocs.io/en/latest/faq.html#i-get-a-demand-matching-error

But basically, you need to increase the constraints listed, as currently the low growth in supply can't meet the high growth in demand.

[Mia McLachlan]

Still getting this error a lot when trying to add a service demand. Do you have any tips on which constraints to change as upping the max on the gas supply doesn't seem to do the trick. Also is there a way to interpret which contraints are active from the error message?

Many thanks,

Mia

[Alexander Kell]

The best way to do this is to look at the last sector which was running.

If you scroll up above the error message, you should see something like the following:

-- 2021-10-14 16:49:10 - muse.sectors.sector - INFO

Running power for year 2045

... etc

If this is the first message you see, it means that the power sector wasn't able to expand any further. Therefore you should relax the constraints on the power sector. 

Another way of doing this is to plot the capacity and see if there is a particular technology which is rapidly expanding. A lot of the time, it is this technology which is causing the problems due to too strict constraints.

Best wishes,

Alex

[Mia McLachlan]

Thanks Alex, that has been really useful for relaxing constraints. I have another error-related question:

What sort of problem could cause "ValueError: Invalid input for linprog: c must not contain values inf, nan, or None"?

Best wishes,

Mia

[sara giarola]

This error is a zero value (generally in the cost values of a technology, LCOE) leading to nan in the code. The cause can be quite varied and generally depends on the input data. Can you try to add the sector individually to know which sector triggers it in a first instance?
We may ask you to upload your data after this initial work of diagnosis.

[sara giarola]

As you go through the results, please check that new capacity is coming online at some point. You can plot the sectors capacity until the model stops working.

--
You received this message because you are subscribed to a topic in the Google Groups "MUSE" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/muse-model/cGOoVPtmIus/unsubscribe.
To unsubscribe from this group and all its topics, send an email to muse-model+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/muse-model/d44acd3f-2deb-4882-ada3-1f7b7f7e0cf4n%40googlegroups.com.

[Alexander Kell]

Hi both,

We've found what could be a bug within MUSE, so thanks for raising the issue.

A quick solution would be to go into the settings.toml and replace the following lines:

lpsolver = "scipy"

to 

lpsolver = "adhoc"

Once you've done this, you'll have to go into the Agents.csv file and replace the column Objsort1 with "TRUE" for each row.

Then it should work.

Thanks,

Alex