Inaccurate Refrigerator Simulation

[Lucas McCarty]

I am using GridlabD version 5.1.0.
I need to accurately simulate the load profiles of various Household appliances (dishwasher, refrigerator, etc.) and have been trying to do so unsuccessfully via GridlabD . To my understanding, there are 3 ways to incorporate appliances into a GLM file (Implicit Loads, Zip Loads, and Unsupported Appliance Models). I attached two GLM files with my attempts for utilizing Implicit Loads and Zip Loads to do this. I got the Zip Load parameters and schedule from one of the GridlabD training courses. Here are my concerns/questions that I am requesting assistance with:

1. To my understanding, implicit Loads are based on profiles from a ELCAP load study. I was wondering where if I could gain access to the information from that study.
2. I have found load data from another study that monitored the load profile of a refrigerator over a week, and it has clear segments where the compressor was on or off. There were times when the device was idle and was consuming negligible amounts of power (only a few Watts). I do not see this or a similar pattern in either output from my GLM files. If there are any obvious issues with my simulation that may be leading me astray, I would greatly appreciate any troubleshooting help.
3. I recognize that the Single Appliance Models are unsupported and highly experimental, but if I were to attempt to use or modify them, it might be useful to see some examples or documentation. This far, I have not found any on the wiki. Is there another place I should be looking for example glm files?

Thanks for your help!

[Frank Tuffner]

Hello Lucas,

1. The ELCAP information should be based on this document, with many references here. I'm not sure if the raw data is available anywhere.
2. One item for your GLMs is you do have a ZIP refrigerator in both of them, so that may increase the likelihood of "always having a load". That said, if you look at the REFRIGERATOR schedule (and the implicit versions are similar), it never has a zero value. Since this is just effectively doing a power manipulation (and not cycling the compressor or anything), that's why it never has near-zero power. It may be an artifact of how the data set was created, or an assumption that the refrigerators of the time (ELCAP is really old) never actually cycled all the way off.
3. I'm not sure if there's any documentation on any of the individual appliance models - many of them, if you look at them closer, are just "shells of objects" (they don't actually do much). That said, the refrigerator model does appear to actually have something inside it, so it may meet your needs. There is an example file under the /residential/autotest/test_Refrigerator_opt.glm. It does fail the assert (which is for unit tests, though this is an optional one), but if you remove that, it does appear to run. The response doesn't look horrible, but I didn't look too closely at it or the GLM -- it might provide the basis you need to get started.

-Frank