Simulation performance of "Price Signal Forecast" battery dispatch

Hello SAM team,

First of all, thanks for implementing the cost optimization approach for battery dispatch in the latest release. However, I am facing some significant slowdown using this feature and I am wondering if there is something unexpected going on and if there are any ways to mitigate this (even at the expense of optimization quality).

Using the SAM app, the simulation takes about ten times longer using price signal forecasting. I also tried this with PySAM (batt_dispatch_choice=5) the performance hit there is about 7 times. Is this a know issue or is there something about my setup that could be causing this? If it is known, is there any configuration available to speed it up, even if the model is simplified?

Thank you

These are the simulation times I get on an out-of-the-box PV-Battery Residential model.
 
With Peak-Shaving Look Ahead Dispatch:

---

Simulation report

---

Total time: 12729 msSSC time: 12586 msSSC version: 250 (OS X 64 bit GNU/C++ Nov 29 2020 03:26:17)Models (5): belpe pvsamv1 grid utilityrate5 cashloan 
With Price Signal Forecast:

---

Simulation report

---

Total time: 136044 msSSC time: 135899 msSSC version: 250 (OS X 64 bit GNU/C++ Nov 29 2020 03:26:17)Models (5): belpe pvsamv1 grid utilityrate5 cashloan 

Hi Nicolas,

Those run times are consistent with what I'm seeing on my machine. I will check with the model developers and see if they have any suggestions.

Best regards,
Paul.

Thank you Paul, looking forward to that.

On a related note: is there any available documentation for how this dispatch mode works? This document is very helpful for the peak-shaving algorithms, but I have not found something similar for the price signal optimization.

Best,
Nicolas

Hi Nicolas,

We hope to make some improvements to the underlying simulation code in the SAM Simulation Core (SSC) to speed up the battery model, particularly for the dispatch options that require handling large amounts of data. These improvements would be for the next release of SAM in the Fall of 2021. Unfortunately there is not a way to improve the performance without modifying the model internally.

As for documentation of the dispatch options, we do not have a detailed description of the price signal forecast dispatch option, which is brand new as of SAM 2020.11.29. I will see if we can get a description posted on the Battery Publications page soon: sam.nrel.gov/battery-storage/battery-publications.html .

Thanks,
Paul.

I was curious if there has been any update on this. My performance speed for doing the PV+Fuel cell+Bat and trying any combination except for Fixed output for fuel cell operation is taking over a day to simulate. It is also going past 100% (right now it's 114.8% and progressing by 0.1% every couple of min. It was fast up until 100% then slowed down to a crawl. 
What % should I expect it to get to since we're past 100 already?

Hi Scot,

The simulation should not take that long on a reasonably recent computer. There may be a problem in your inputs that is causing the simulation to take so long.

Would you mind attaching a copy of the file that is taking so long to simulate so we can troubleshoot?

As for the progress bar, this is on our To Do list to fix, just not as high priority as some other issues for the moment: github.com/NREL/SAM/issues/522

Best regards,
Paul.