Different results between Python and Javascript implementation of the PMV Elevated Speed? about comfort_tool HOT 9 CLOSED

Just an update on this issue..I was able to import through R's V8 library the comfort_tools javascript libraries as described on the wiki.

However, I got a third different PMV value as shown here.

I feel I am missing something..

from comfort_tool.

@carlosviansi ,

Thanks for the diligence with testing this.

I have found two key differences between this python function and @thoyt 's javacript function that are entirely my fault for not updating it to the 55-2013 standards. I have also found what appears to be one inconsistency between @thoyt 's javascript code on this github and the javascript that is currently in the CBE web tool, which I imagine Tyler should fix soon.

They two mistakes on my end are:

1. I had been using a still air threshold of 0.15 m/s to decide when to send values through the elevated air speed methods and Tyler was using 0.1 m/s. The new 55-2013 standard seems to have changed this value from 0.15 to 0.1

2. It seems that @thoyt had updated the method for calculating the cooling effect for elevated air speed a year ago. I had not updated to this new method.

Both of these issues have been addressed in the commit that I just made . So my stuff should now be in sync with @thoyt 's.

Finally, I noticed that @thoyt changed the core body neutral temperature from 36.49 to 36.8 recently . However, he has not yet updated the javascript in the CBE web tool as you can see in this screenshot:

Once @thoyt updates the web version of the CBE tool, you should get matching values between this python function and the online CBE tool. I'll leave the issue open until @thoyt updates the website.

Thanks, again, @carlosviansi , for pointing all of this out.

-Chris

from comfort_tool.

Thank you for the quick response and syncing the files @chriswmackey. I suppose the new python version and github javascript code should match results for time being until @thoyt moves the code to the CBE Tool server, correct?

I thought it wouldn't also hurt to point/ask:

1. I had to deal with the conversions between F to C, and fpm to m/s, and one thing I noticed was that for fpm to m/s since no function was created in utils.js at times 196.85 was used and elsewhere (and perhaps mostly 196.9) was used. Do you think this can influence the results?

2. Is there any plans to add the EN to the python implementation that is available in Javascript?

3. Out of the comfort models implemented, which one would you recommend for a room that has both a HVAC, ceiling fans, and windows people could open? Despite reading on the models from both your white paper and ASHRAE55-2013, it seem to me I needed some mix of the PMV/PPD Elevated Air Speed function and the Adaptative model given their limitations, so I ended up puzzled.

Thanks.

from comfort_tool.

@carlosviansi ,

Thanks for all of the feedback. Yes, the python version on this github should match the javascript version on this github and let me know if you find otherwise (things seemed to be matching with the online CBE tool when I changed the core body temp back to 36.49. To answer your questions:

1. Yes, it sounds like the extra decimal place in the conversion factor could cause some small differences, although the differences would probably not be significant enough to result in greatly erroneous PMV values (I would guess that the error would not be more than +/- 0.05 PMV). I imagine that, in the future, @thoyt might have everything reference just one conversion function like what currently with the CtoF conversion. I will leave this one to @thoyt as I don't know all of the places in @thoyt 's code where this conversion happens.

2. Yes, sorry for not including the EN model sooner. I had translated the EN function over for the python plugin that I regularly contribute to. . However, I forgot to send the function here after the translation. I have added this function with this commit . I also took out the claculate comfort range function as I have understood @thoyt 's much faster way of calculating this in the CBE tool and I will implement this shortly.

3. This is a very good question and is one that I have struggled with on a number of design projects. I can tell you what AHSHRAE says, which is that you are supposed to use the PMV model for any case that does not fit the Adaptive Ideal (ie. no AC installed in the bldg/ year-round natural ventilation, great freedom of dress code, etc.). However, I can also tell you that most experts find this explanation and course of action greatly unsatisfying and in need of more research to improve. Personally, I can understand using the PMV model in cases where operable windows are really small awning openings, through which you cannot get significant airflow (as is popular on a lot of recent US projects). However, for truly mixed mode natural ventilation / AC buildings, I have another means by which I deal with the present lack of a published standard. When the European comfort scientists were conducting the large comfort surveys that were used to build the EN adaptive standard, they used their same survey methods in both conditioned buildings and "free-running" naturally ventilated buildings. At the end of the survey, they derived two different adaptive correlations. One for free-running buildings, which is now published in the EN standard:
   

And another correlation for conditioned buildings, which is not published in either EN or ASHRAE standards:

The argument of the scientists is that a mixed-mode building would lie somewhere between these two correlations perhaps like so:

Although this claim was considered to be too speculative and reliant on a dubious "level of conditioning" value to be published in any standard, if you trust yourself to put a number on the level of mixed-mode operation of your building, it really seems to be the best representation of comfort that you will get given the present level or research. As such, this is what I tend to use while designing mixed-mode buildings. At the end of the design process, I will also usually do a final check of the values against a PMV model with very adaptive occupants (changing clothing, turning on fans, etc) just to show compliance with the standard. If you want to use this design method yourself, you will see that the python adaptive comfort functions include a levelOfConditioning input that is set to 0 by default. However, you can change this value to any value between 0 (full nat vent) and 1 (full conditioning) to use a correlation anywhere between these two as you see in the images above.

Hope this helps,
-Chris

from comfort_tool.

@chriswmackey, I very much appreciate your insight in 3) since I am just getting started on this domain (I come from a Computer Science / Data Mining background). Are the papers you are referring to by Emmerich, Polidoro and Axley "Impact of adaptive thermal comfort on climatic suitability of natural ventilation in office buildings"?

I could not find the plots you just posted in your answer. Where can I find the published and unpublished work related to this? I believe I have the EN version EN15251-2006, but I also didn't see these plots there. What could give me a better understanding of the EN models levelOfConditioning? We have some survey data here, by default I would likely inspect different runs of the EN model with different levels of conditioning to see if the number generalize through different buildings, however I would like to at least get a bit more of a sense rather than go purely empirical :)

If you would happen to have any other basic references I would appreciate. I found more insight looking at published work and CBE White paper on this tool rather than ASHRAE55-2013 itself, sadly.

Thank you again.

Edit: Other two apparent related links I came across which I am unsure if you are referring to are this blog post and this tool.

from comfort_tool.

Hi @carlosviansi and thanks @chriswmackey for being so responsive.

You're correct in pointing out that the web tool hasn't been updated. I'm hoping to take care of that soon (damn full time job getting in the way!). Meanwhile, the best possible version of the code to use is here. As Chris mentioned, I've been making changes to the tool as addenda are adopted in ASHRAE 55. This includes the new elevated airspeed threshold, errata such as the neutral core temperature, and some other changes and features that don't affect the core models.

If you're interested and still have questions, I can put you in touch with the ASHRAE-55 community and/or my former colleagues at CBE. Unfortunately I don't know much about the problem of comfort modeling in mixed mode buildings, but I'm sure others at CBE may be familiar with the literature.

from comfort_tool.

@thoyt that would be great if wouldn't be an inconvenience for you. Most of the white papers are non-accessible and have multiple versions so it is really hard to make sense of this if you are just getting started. We have an undergoing project here for a dashboard in two new buildings which I think would safely be classified as mixed-mode, so I want to make the best out of it.

I am mainly trying to get a sense of the pros/cons of all models and their parameters (I believe I managed to get most of all the models implemented in CBE, but I do not even know which version the EN @chriswmackey mentioned the implementation is based of, so I can search for the levelOfConditioning experiments with AC and without AC).

from comfort_tool.

@thoyt ,

Glad to hear about the updates and thanks for keeping informed with the changing standards. This github is truly a great resource for helping keep the comfort standards sychronized across other software projects.

@carlosviansi ,

The best and most comprehensive source on Adaptive comfort research that I know is this book. I highly recommend purchasing it if you see yourself working with comfort models in the long run. Also, if your wallet is a bit tight, this shorter paperback book by the same authors contains most of the information that the more comprehensive one above does.

Both of them contain the correlations for free-running and conditioned spaces that were found by the EN+ASHRAE adaptive comfort research groups. The more comprehensive book also contains a really good historical account of how the adaptive models evolved from the initial research in the 60s/70s and discusses some of the shortcoming of the PMV model. Every chapter ends with a list of citations and, if I had not left the book in my office, I would tell you the exact original publication that contains the two correlations right now. Still, I can tell you that figure 5 in this pdf is the same as that which was published in the original study.

Also, to be clear, the levelOfConditioning is a variable that I created based on my design-driven needs and the argument made by the adaptive comfort scientists that mixed-mode buildings lie somewhere in between the two correlations. The actual method of selecting a dimensionless number to represent the amount of conditioning of a building is not something that the scientists endorsed. They only said that such a correlation is somewhere between the two and that more research would be necessary to uncover what exactly such a correlation could be. Also, I should admit that I have pretty much stuck to using levelOfConditioning of 0.5 (perfectly between the two) in most of my work and, in cases where the building is close to fitting the adaptive model (conditioning for only 2-3 weeks), I use the correlation for free-running buildings.

Hope this helps,
-Chris

from comfort_tool.

Thank you very much @chriswmackey. This is all the information I needed concerning the model inner works. It seems our group recently purchased one of these two books, so I am now looking even more forward reading them.

I finished re-reading the CBE Tool paper again, and I think I am more comfortable on what you are talking about :-) Thank you for sharing the numbers you have been using, I see how to go about them now.

Are you aware of any online forum where people discuss related work or current hot journal venues on the topic? While I was searching for material I landed on the Energy and Buildings and the Solar Energy journals. But I am unsure if there is anything else out there, especially conferences. We are based in Hawaii, and it seems there is only but a few studies related to here. I am a bit skeptical on the tropical influence when reading all results on general work done so far using all these models, hence the curiosity.

On your white paper, I have not found if the ranges panel are using PMV + Elevated Air Speed (and with the parameters you changed). Based on the chart used, I am assuming that is the case, and the Adaptative Model is not available to play with ranges. Is that correct?

I suppose these are all my concerns involving the tool and related work. I believe I deviated quite a bit on the topic trying to make sense, and again, thank you for your extensive detailed replies and advice! I hope this thread will serve as a getting started if other user of your tool ends up here 👍

from comfort_tool.