SmartDS Solar Extreme/High Scenario Issue

[adam-morse]

Hi Tarek,

I am having issues doing time series simulations of any of the GSO solar high or extreme scenarios. It seems like there must be some kind of issue with the InvControl definitions as these don't appear in the medium/low/none solar scenarios. I don't have any issue running high or extreme scenarios with the opendss_no_loadshapes files (ie snapshot simulations).

Opendssdirect is giving me DSSException: (#482) Error Encountered in Solve: Access violation. [file: "...\SmartDS_GSO\urban-suburban\scenarios\solar_high_batteries_high_timeseries\opendss\uhs1_1247\uhs1_1247--udt20176\Master.dss", line: 24]

I already changed all the kvarlimit properties to kvarmax and VV_RefReactivePower to RefReactivePower to be consistent with current OpenDSS properties. I didn't see that there were similar property updates needed for InvControl.

Any idea on how this might be fixed?
Adam

[tarekelgindy]

Hi Adam,

There's a couple of things that come to mind, although I don't have any immediate gotcha for this.
The first obvious one is to double check that the location for the timeseries profiles are referenced correctly in the LoadShapes.dss file.

Another thing to look at is if this problem occurs when you run a single timepoint. The default in the file you mentioned runs an entire year's simulation and loads all of the solar and load data into memory for that. It's unlikely that for a single feeder it's causing memory problems. But I almost always run simulations by using OpenDSSDirect to load the model into memory and then set the load and solar values for each timepoint that I want to run. This was what was used to run the simulations of these models to generate the data provided in the analysis folder. The process used is described here: https://github.com/openEDI/documentation/tree/main/SMART-DS#using-opendssdirect

However, the most likely thing though is incompatibility between the models and the newer version of OpenDSS (version 9.0+). As you correctly noted, the OpenDSS models use different keywords in the inverter models. It might be worth trying to run the original models using Legacy mode, which allows the old inverter settings to be run without problems:
https://dss-extensions.org/OpenDSSDirect.py/opendssdirect.html#opendssdirect.Basic.IBasic.LegacyModels

If you're running in python with opendssdirect imported as "dss", you can run the command:
dss.Basic.LegacyModels(True)

before running the model.

We've used this successfully in the past to run OpenDSS models of the SMART-DS systems. Do you know what version of OpenDSS you're using and whether you're running it in the GUI or through OpenDSSDirect? I could try and see if I can replicate the issue.

It's probably worth us updating the inverter models to be compatible with the latest OpenDSS versions (and testing it) with a new release.

Let me know if any of these help - I'd be interested to hear if any of these fixes address the problem.

[adam-morse]

Appreciate the quick response.

Unfortunately when I run dss.Basic.LegacyModels(True) I get DSSException: (#5016) LegacyModels flag is not supported in this version. I've been running the current version of opendssdirect=0.9.4. When I run SmartDS time series in the OpenDSS GUI my computer locks up and I prefer using the python interface anyway.

There is a note in what you cited about LegacyModels no longer being supported beyond DSS C-API v0.13.0. The earliest version of opendssdirect I was able to install this morning was 0.7.0. When I try installing earlier versions I get installation errors that are related to dss_python distribution version 0.10.7. I did get the same LegacyModels error for version 0.7.0...

I've been successful using the code you provided in the documentation to run snapshot simulations for different moments of time. Thank you for providing that documentation and I've been processing the code posted on OEDI.

I am curious to run the timeseries simulations to plot and analyze the monitor data and view changes over time. Running the smartDS feeders/substations typically doesn't take too long on my PC.

I'd definitely be curious if you can run the time-series simulations (for high and extreme PV scenarios) without the error I'm seeing...