is anyone interested in collaborating on building performance modeling for greenhouses?

I have been wanting to do some multi-criteria optimization for a passive solar greenhouse design.  I have experience with the same ML and AI techniques that Gagnon used in his work on multi-objective optimization[1] and his PhD dissertation[2].  In specific I am hoping to work with someone to develop an EnergyPlus model for one or more passive solar greenhouses initially based on UMN's Deep Winter Greenhouse (DWG)[3], and then add modular features like climate batteries, solar/ground heat-pumps, and active/passive venting.  The reason that I wanted to start with the DWG is that UMN not only provides detailed plans, dozens of them have been built, and many of them have been well instrumented.  In addition, one of my collaborators is willing to provide performance data from his first DWG greenhouse -- the data to be used for initial model calibration and validation.  In his build, he also installed a climate battery, which he has turned on and off as part of his as-built performance tests.  Once the models are "believable", I then want to apply the ML techniques to optimize the model for one or more fully instrumented greenhouses to be built on my small farm in Maryland.  Please PM me if you are interested, and we can discuss details.

[1] https://corpus.ulaval.ca/jspui/bitstream/20.500.11794/68283/1/Performance%20of%20a%20sequential%20versus%20holistic.pdf

[2] https://corpus.ulaval.ca/jspui/bitstream/20.500.11794/32726/1/34660.pdf

[3] https://extension.umn.edu/growing-systems/deep-winter-greenhouses

We are recently working on quite a few models, please check My YouTube Channel

I realize it may be awhile, but please remember to give us an update when your results are in.  

This is a great idea and I'm really excited to see your findings!

Thank you both for responding.  

@K Eilander,  I will be glad to post some of the results as they come along.  If I can get sufficient work forward, I have interest in presenting this as a use case for AI/ML optimization at a NASA conference.  Last year I taught a introductory workshop on what AI and ML are, and how to start started with some of the basic tools.  To get this really working not only do I need to automate data acquisition for the climatology, but also find a way to simulate weather dynamics given expected climate change scenarios.  Unless something keeps me from it, my intent is to make all of the models available open-source.  Anyway, I expect to present in several venues, and there is no reason I cannot also report them back here.  As a note, what I was hoping is that I found someone with a little experience with EnergyPlus, or other freely available or inexpensive building thermodynamics modeling software, and who is willing to collaborate or tutor.  I did have one person PM me, and while he does not have EnergyPlus experience he has other experience that is useful to us both.  I look forward to seeing how that collaboration move forward.

@Aubrey Zhang,  I looked at your video titles, and watched a couple of the videos.  I see that you have some movement simulations and CAD models for what appears to be bubble insulation pumped in between multiple glazing layers.  I see that you have some sensor data for a couple of your prototypes, but do you have any thermodynamic models for these greenhouses?   If so, what software do you use?  I will have to read more about your technology and think how it might integrate here, but I am still hoping to do some thermodynamic modeling and optomization.

I'm very interested in this information.  

I work in this space for MEP Building performance engineering.  

We worked with some clients doing CFD analysis for simulating heating / cooling for data centers, mining operations, and military installations.   https://www.autodesk.com/products/cfd/overview
I personally do not have the experience for CFD analysis but it is compelling work.  

By building models in Revit we can fast track the simulation aspects in CFD tools like Autodesk CFD.  This is not cheap software simulation but it is important for understanding stratification and where trouble areas might be.  

Our Energy engineers spend a lot of time on enhanced commissioning to see what the actuals are vs. what the predicted outcomes are.  Usually field adjustments are always needed.