Energy simulation for models with non-convex surfaces


old thread

Sorry to re-ask but it was unclear to me from the thread above whether or not Pollination now accepts non-convex surfaces. My Pollination runs are failing and I’m trying to determine if this is the issue.

I’m running LB1.4.0, OS3.3.0, EP9.6.0 and just a few days ago installed Pollination0.3.8 for GH.

For a 7 story office building, I’m able to run a full HB Model to OSM in 2 hours (close enough results to HB Annual Loads to verify that the model is probably set up correctly), but throwing an error saying…
“1. ** Severe ** DetermineShadowingCombinations: There are 10 surfaces which are casting surfaces and are non-convex.”
When I try to run the same model through Pollination for GH, it fails in 13 minutes.


Mostapha before you waste any time on this, just to make sure it it isn’t garbage in on my side…I shared my code with a friend at Buro Happold; he thought my results looked weird, he ran it with different results. He suspects that something is wrong with one of my underlying applications so I’m going to clean house today and reinstall everything. My friend is on the west coast so we won’t be comparing notes until the end of this east coast workday.

@mostapha @chriswmackey

Hey guys, If you read this and want me to share my model and code with you to see if there’s an issue, please let me know.

Firstly, Thanks for your support and encouragement. I’m hesitant to subscribe to Pollination until I have a good grip on basic HB workflow. FWIW, I can now run fast and full simulations in GH/HB but all of my recent attempts to run my model through GH/HB/Pollination have failed. More context below.

I was struggling with a seven story office building model for a month (learning HB on the fly, watching everything Chris has posted since 2014). After I streamlined my code to be current and getting it to work, last week I connected with an engineer friend at Buro Happold to compare notes. When we both ran my code unchanged, we got dramatically different results from the HB Annual Loads component (his professional judgement was that my EUI results were wrong.) I was running LB1.4.0 with all of the recommended software versions from your compatibility matrix. He did not have OpenStudio installed and was running E+22.1.0. Tried many permutations of reinstalls but finally, this weekend I uninstalled LB, E+, OS, Rad and Python, word searched my hard drive to wipe away all remaining related directories, washed my system registry 10 times and then did a reinstall. I installed LB1.4.0, Python 3.7.9 Rad 5.4a, OS 3.4.0, E+9.6.0 installed under the folder tree of ladybug_tools, and I installed E+22.1.0 out in the root of C: where it wants to be. I re-ran HB Annual Loads and did a full simulation through HB Model to OSM and I got results similar to my friend at Buro Happold. So I don’t know what was broken about my prior installation, but something was.

Back to Pollination, I followed Mostaphas instructions (link below) using the “custom-energy-sim” component. Before my LB reinstall, the Pollination run would last about 12 minutes and fail. Now it lasts about 45 minutes and fails. In all cases I could see that the model uploaded but in each case, the simulation failed. So I’m a bit flummoxed.

Hey @rnarracci ,

That is a warning and not an error. EnergyPlus likes to throw a lot of them when it comes to its solar distribution calculations (since E+'s solar engine isn’t the greatest compared to things like Radiance) but, most of the time, it is actually pretty safe to ignore them. If the warning really bugs you, you can make it go away by using the PixelCounting calculation method on the HB Shadow Calculation component.

For your 7-story office building model, it should simulate correctly in EnergyPlus as long as the model passes the PO_ValidateModel command in the Rhino plugin like so:

If something is not correct with your model, then the validation command will tell you what exactly is wrong and the latest Rhino plugin even helps you zoom in on the invalid geometries of your model so you can fix them.

Lastly, if you model is passing the PO_ValidateModel command but it does not simulate successfully in EnergyPlus, then this is a bug on our end and we will fix it. If you find a case like this and you share a model that lets us recreate the issue, I will fix it.

FYI, if you are using and of the Pollination single-click installers (either the Pollination Rhino plugin installer or the Pollination Grasshopper one), then you do not need to worry about managing installations for any of the simulation engines. The pollination installers automatically come with a compatible version of EnergyPlus, OpenStudio, and Radiance for the version of Ladybug Tools that they ship with. Furthermore, these engines that install with the Pollination installers are wholly separate from any installations of these engines that you might have elsewhere (eg. in the root of your machine). So I would recommend using the single-click Pollination installers over what you are currently doing with what sounds like the Food4Rhino installer.

Thank you Chris. I used the executable installer for the Pollination Grasshopper plugin directly from your Pollination site.
I do not have a paid subscription for the Rhino plugin so I cannot use PO_ValidateModel.
Its just curious that I’ve now got seemingly reasonable results coming out of HB Annual Loads and HB Model to OSM but not Pollination.
Was going to DM a file link to you but I don’t think I have permission to do so (no “message” button on your avatar when I click on it.)

I would encourage you to get the Rhino plugin (there’s a 14-day free trial) but there is another way to validate your model with a free Grasshopper component. It just isn’t nearly as nice as the Rhino plugin since it’s only going to give you the IDs of the invalid geometry and not allow you to actually zoom and visualize the invalid parts.

There’s a component in the latest release called “HB Validate Model” that will give you a report of everything that is wrong with your model before you go to simulate it.

If that component tells you that your model is valid (like the screenshot below) and you are still getting simulation errors, then that’s a bug on our end and we will fix it if you can give us a sample file to recreate it.

Hey Chris,
Sorry so long in getting back to you but I got dragged into another design competition. I swung back around and used HB Validate Model to find garbage-in; lo an behold…I found some overlapping solids that HB doesn’t like. Will be sweeping through the model to repair and will get back to you about the results. Have a happy and safe 4th.

1 Like

Hi @rnarracci! Thank you so much for the update.

Have you considered using the Pollination Rhino plugin? The combination of PO_ValidateModel UI in Rhino to find the problematic cases with PO_AlignToGrid to fix them will cut your fixing time drastically. Here is a video that shows how to use the PO_AlignToGrid command.

Here is UI for the PO_ValidateModel command.

1 Like

Mostapha, I would consider it but only after I’ve gotten reasonable results from GH that I can compare with other tools to validate. So far I’ve had lots of problems, including having to completely reinstall the entire suite of LB/HB and backbone apps so that my results match colleagues running the exact same model and script. Just today I did the recent Pollination upgrade and I’m getting wacky EUI results again, so I may have to once again do an entire suite uninstall and reinstall. I removed the overlapping masses and got a valid model and now Pollination won’t even accept the model onto the website (it got that far before and failed). I know you guys are trying really hard to monetize the project but I can’t take the plunge until I’m more confident.

That’s how it should be! And we are here to support you through this process. :grinning:

I feel you might be looking at the wrong place to solve this problem. The core Ladybug Tools libraries are very stable at this point and most of the changes are for edge cases or adding some new features. Also, with the new single-click installer, we ensure that the version of OpenStudio/EnergyPlus is in sync with the LBT core libraries.

If the EUI values are changing it is most likely the result of a change in the input model or the Grasshopper definition. If you can share the IDF files that have been generated or the model we should be able to help you better. I would start with looking into an energy balance chart or a breakdown to see what has changed in the results before thinking about uninstalling the whole package. Let me know if there is anything that I can help you with! :slightly_smiling_face:

Could I privately transmit the model and code to you? See if you get the same results a me?

Yes! You can click on my user profile and continue the conversation in a private message. I just sent you a message doing the same with your user profile. :point_down:

DM’d a WeTransfer link to you in case you didn’t see it.