ASHRAE Guideline36, Section 5, Air-Side AFDD

[MatthewSteen]

Close #252.

AFDD Shapes

• 5.16.14 Multiple-Zone VAV Air-Handling Unit, Automatic Fault Detection and Diagnostics
• 5.17.4 Dual-Fan Dual-Duct Heating VAV Air-Handling Unit, Automatic Fault Detection and Diagnostics
• 5.18.13 Single-Zone VAV Air-Handling Unit, Automatic Fault Detection and Diagnostics
• 5.22.6 Fan Coil Unit, Automatic Fault Detection and Diagnostics

[MatthewSteen]

CI error appears to be unrelated.

@gtfierro FYI there a couple points here that appear to be missing from Brick.

1. Air_Differential_Temperature_Sensor
2. Entering_Air_Temperature_Sensor

[gtfierro]

I pushed a commit which changed how the outside air fraction shape got modeled, but this looks great!

[gtfierro]

Thanks! I think this looks good to me

[TShapinsky]

Do we know what the cause of the failing tests are? Is it just that this branch is missing fixed from develop and needs to be updated?

[TShapinsky]

@MatthewSteen does the notebook timeout for you locally? or just take a long time to run?

[MatthewSteen]

@TShapinsky fails locally too, all with ERROR traitlets:client.py:845 Timeout waiting for execute reply (600s).

Failing Notebook: notebooks/Existing-model-validation-example.ipynb
---------------------------------------------------------- Captured stderr call ----------------------------------------------------------
A cell timed out while it was being executed, after 600 seconds.
A cell timed out while it was being executed, after 600 seconds.
The message was: Cell execution timed out.
Here is a preview of the cell contents:
-------------------
generated_templates = validation_context.as_templates()
for t in generated_templates:
    print('-'*80)
    print(t.body.serialize())
    templ_bindings = {}
    for parameter in t.all_parameters:
        param_value = notebook_input(f"Please enter the value for parameter \"{parameter}\":")
        templ_bindings[parameter] = BLDG[param_value]
    graph = t.evaluate(templ_bindings)
    medium_office_model.add_graph(graph)

[MatthewSteen]

It's the first line in cell [7]...

generated_templates = validation_context.as_templates()

...that's taking >600s to run.

BuildingMOTIF/tests/integration/test_notebooks.py

Lines 10 to 11 in 6751fb4

	def test_notebook(notebook: Path):
	run = NotebookRun(notebook, 600)

[gtfierro]

Weird! I run that cell on my machine and it was relatively quick -- certainly nothing like 10 minutes. I'll try to reproduce in a clean environment

[gtfierro]

My tests show that a couple cells later is the call that times out:

conformance = medium_office_model.test_model_against_shapes(
    shape_collections=shape_collections,
    shapes_to_test=shapes_to_test,
    target_class=BRICK["AHU"])

That method involves a SPARQL query and some SHACL validation. I'm pretty sure the SPARQL query is the slow part. We may want ot use a smaller graph for this test

[MatthewSteen]

Ok, should we merge and revisit the fixtures in the future? I think it makes sense to use smaller models. Perhaps the SmallOffice model or some/all of the G36 systems.

[TShapinsky]

@MatthewSteen @gtfierro did we ever decide on how to mitigate this causing test failures. This is now causing all prs to fail testing...

[gtfierro]

I think it just needs a smaller test file or a faster SPARQL engine (but this second option is much harder to do for now). I don't have cycles to handle it this week but I can try next week

[MatthewSteen]

I was seeing the same failure without the changes I made here. Also seeing inconsistent test results in general so I'm trying to determine the cause.

After the release we should discuss the integration testing we want to have, which could be a topic for a dev mtg. One thought is we could (re)move the notebooks dir and move the integration testing notebooks under tests/integration/fixtures/. Then we could convert our Tutorials to ipynbs so that users can run them as example notebooks.

[gtfierro]

There may be a quick fix to this....there is a SPARQL query inside the "hot loop" of test_model_against_shapes which can actually be done outside the loop. I'm testing now to see if this fixes performance