Save simulation output

[wigging]

Added some utility functions to save the simulation output to CSV, NumPy .npy, and compressed NumPy .npz files. Also added an example of using the save functions.

I did a comparison of the disk usage for these different file formats (see below). The compressed .npz format can save a lot of storage space for large pack simulations. There are other file formats such as Feather and Parquet but using those formats would introduce more dependencies to liionpack.

Below are disk storage results for a 16p2s single charge/discharge simulation. Simulation outputs written to file are time, terminal voltage, pack terminal voltage, pack current, open circuit voltage, cell resistance, and cell current.

Filetype	Disk space
csv	552 KB
npy	188 KB
npz	88 KB

Below are disk storage results for a 400p125s single charge/discharge simulation. Simulation outputs written to file are time, terminal voltage, pack terminal voltage, pack current, open circuit voltage, cell resistance, and cell current.

Filetype	Disk space
csv	803 MB
npy	246 MB
npz	33 MB

[codecov]

Codecov Report

Merging #145 (fb555eb) into main (a7d8b02) will increase coverage by 0.05%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #145      +/-   ##
==========================================
+ Coverage   97.12%   97.18%   +0.05%     
==========================================
  Files          11       11              
  Lines        1184     1206      +22     
==========================================
+ Hits         1150     1172      +22     
  Misses         34       34              

[TomTranter]

@wigging thanks looks good but can you add tests?

[wigging]

Sure, but how do you make a unit test for a function that creates a folder and writes to files?

[TomTranter]

Sure, but how do you make a unit test for a function that creates a folder and writes to files?

You can use os.path.isfile() to assert that a file is created

[wigging]

Sir @TomTranter, I uploaded some tests. Let me know what else is needed. Regarding the online documentation, will the function docstrings be used to update the docs or do I need to manually update the documentation?

[wigging]

If this pull request is not useful for liionpack then please close the request. Otherwise, let me know what else needs to be done before accepting the request.

[TomTranter]

@wigging sorry I haven't got round to looking at the code yet. Will try to asap.

[TomTranter]

This is a good start and I'm happy to merge. Some thoughts would be we should add headers and a time column to the csv. Think about using hdf5 as well. Also we should add some options to do journaling for the solve process but this may be better to save the simulation objects themselves rather than output variables. Worth thinking about some more

[wigging]

The csv function saves each output item to its own csv file into a csv-results folder. This includes the time output which is Time_[s]. Similarly, the npy function saves each output item to its own npy file. The npz function saves the output into one npz file.

For saving to csv, I guess it would be better to save everything into one csv file instead of separate files.

For the NumPy formats, I think I'll just use the npz format and remove the ability to save to npy. I don't see any benefits to use npy instead of npz.

Supporting hdf5 output would add another dependency to the project. Liionpack already uses NumPy so not a big deal to save output in a NumPy format.

[wigging]

It's been awhile since I ran the code in this pull request. Disregard my comment about saving to a single csv file. The csv function saves to multiple files because some of the outputs are 2D arrays. So it's not feasible to save all the output into a single csv file.