Remarks on HB natural ventilation - height difference

Hi @chris , hi all,

I have some thoughts about the component setEPNatVent and its relationship with EnergyPlus.
For stack driven ventilation, EP takes into account the input called “height difference”” which reffers to the difference between the midpoint of the lower opening and the Neutral Pressure Level. On top of that, according to 2009 ASHRAE Handbook of fundamentals chapt. 16, equation 38 (from which the EP formula is based on), when one deals with only one opening, the NPL is the midpoint of that opening and the height difference should be half of that opening height.

The problem is that when I run a simulation with one window on one side using the setEPNatVent component, and when I check the IDF file, the input for height Difference is the total window height. And this is not corrected by EP because the result airflow volume from my simulation corresponds to the theoritical calculation using the EP formula (

In this situation, shouldn’t honeybee write in the IDF file half of the total window height for the height difference input ?

Also, is there any particular reason why honeybee sets the discharge coefficient with fixed values (stackDischargeCoeff_ input) instead of the formula used in EP when user sets it to autocalculate ?
Again, according to ASHRAE, this coefficient should be set to 0.65 when the openings away from each other and the air flow is unidirectional, and should be C = 0.40 +0.0045(Tint - Tout) when there is one window and mixing can occur.

My guess is that all what I mentioned above is actually taken into account in the stack discharge coefficient of 0.25 that you suggest to use when there is only one window… but I’m not sure

Cheers !

@Vincent ,

Glad to find someone who has dug into the formulas on this one and, yes, your last sentence hit it on the nail exactly. There are a couple of different ways of arriving at a decent estimate of the buoyancy-driven airflow. One of them is to use half of the total window height as you mentioned and the 0.65 coefficient. The other (what honeybee uses by default) is to take the total window height but use a smaller coefficient of 0.25. In honesty, the main reason why I preferred the method of adjusting the coefficient is that I knew recommended values for the coefficient that help account for insect screens and, at least in my experience, the vast majority of operable windows have insect screens. You can see in the discharge coefficient description that this is actually why 0.17 is the default (and not 0.25):

I am open to changing the default method if someone makes a case for why an alternative method is better (you are one of the first people I have met on the forum to dig this deep into it). However, I should also mention that part of the reason why there’s an option for CUSTOM STACK/WIND VENTILATION is so that you can customize the buoyancy-driven function exactly as you like if you are not happy with the Honeybee default. Now that you bring it up, though, I realized that I never exposed a way to access the ‘autocalculate’ feature for the stack discharge coefficient. So I just pushed a small change to the component, which will use the ‘autocalculate’ method is you select CUSTOM STACK/WIND VENTILATION and you don’t input anything for the discharge coefficient:

Let me know your thoughts!

Hi @chris , thanks for your answer.

Indeed, the assumption of taking 0.25 for a sliding window is pretty good. I made a test, with a box shaped room with windows on one door, with NatVent always on, and I compared the air flow obtained with the 0.25 stack coefficient with the theoritical one calculated from the formula of Ashrae taking the Tint and Tout into account (which is not the best thing to do since Tint actually depends on the NatVent rate but this a gross calculation) and the result was that the formula from Ashrae overestimated the airflow by just 16% over the year, so I would say it’s pretty good since we are dealing with fluid dynamics and their will never be ONE correct answer.

One thing though, I would stress that these coefficient make sens only for sliding windows, and should be used with care. Since sliding windows nor insect screens are not common at all in Europe for example (or at least in France), I would encourage users to take these stackcoefficient with care or to put their own… Because If one has a “Italian style opening” (horizontal rotation axis) or a “French style opening” (Vertical rotation axis), the stack discharge coefficient may be much lower, if we open a winfow with a small angle.This paper right here
takes a look at how discharge coefficient change with non sliding windows.
Basically what I’m gonna do for my simulation tool project is make my own discharge coefficients to reflect french typical windows.



@Vincent ,

Thank you for the thoughtful for response and for widening my world-view (I didn’t realize that you usually don’t need insect screens in France). I apologize for any US-centric biases that may be in the components like this one and I wholeheartedly support your recommendations regarding the adjustment of the stack coefficient for different window types.

Just to give another reason why I decided to go with the insect screens by default, I figured that the design consequences of someone leaving unnecessary insect screens in the energy model is that they end up sizing the operable portion of the window a little larger than it actually needs to be, which seems better than the alternative of under-sizing it and ultimately not being able to make full use of natural ventilation. Or I can at least say that I sleep better at night thinking that less-experienced users are accidentally over-sizing their operable windows rather than under-sizing them :slight_smile:

But, while we are giving out notes of caution to the users of the component, I just want to add that I still loose sleep at night over one thing: people assuming that the honeybee default is suitable for awning windows (aka. the “windows with a small angle” that you link to in that paper). Especially since I see so many buildings in the US right now going up with awning windows, I would particularly caution users that they really need to decrease the discharge coefficient for this case.

Thanks again, @Vincent!


we are dealing with fluid dynamics and their will never be ONE correct answer.

You don’t know how much I wish more people understood this. I remember the first time that I took a course in fluid dynamics and expected all of the formulas to be deterministic, ideal-gas-law-type equations. Man, I had a lot to learn then and I get the sense that a lot of energy modelers still have a lot to learn now.