Question-and-Answer Resource for the Building Energy Modeling Community
Get started with the Help page
Ask Your Question

Part Load Performance for ASHRAE 90.1

asked 2017-02-01 15:55:34 -0600

Jim Dirkes gravatar image

updated 2017-05-03 19:51:54 -0600

Dear Forum, When modeling for compliance with ASHRAE 90.1, DX and chiller systems must comply with specific requirements for EER / COP and IEER (or IPLV for older versions). That seems to be a "two-edged sword" - you're trying to meet two different criteria, each of which affects the other. For actual / specified equipment you can (in theory) obtain the performance data and create curves which represent the performance. That's somewhat painful, but do-able.

For "baseline" systems, there is no manufacturer data. Hmmmm.... What's a modeler to do?

I have some ideas about how to solve this dilemna, but I'm wondering what others have done. The most obvious solutions are:

  1. Ignore the IEER and hope for the best
  2. Plug in default or example file values and hope for the best
  3. Tinker endlessly with curve coefficients till the result are close to what is needed
  4. Wave a magic wand :)

I'd love to hear what approach you take.

edit retag flag offensive close merge delete


FWIW, this is something that I haven't dealt with in the OpenStudio baseline automation code, and that PNNL has not dealt with in the DOE Prototype Buildings. In both cases, we use "typical" curves and then modify the COP. If anyone has a good idea for a scheme to do this repeatably I can code it up and see how it works.

aparker gravatar imageaparker ( 2017-02-01 16:00:51 -0600 )edit

3 Answers

Sort by ยป oldest newest most voted

answered 2017-02-02 09:54:08 -0600

Jim Dirkes gravatar image

The comments and answers are helpful. So far, though, it doesn't sound like there's a better solution than the ones I mentioned initially.

What if... we start with a "legitimate" curve set like: image description

... and "drag" the Rating condition (red) curve downwards for better performance like the dotted line: image description

That can be done using a formula which modifies the original curve formula in a way that maintains the original general shape while also maintaining the Rating Condition value (35C) so EER remains the same.

It's a completely artificial manipulation of the original curve set, but seems reasonable for a "baseline" coil. When I mentioned in the original post that I "had some ideas" about how to do this, this curve manipulation is what I had in mind.

Any further thoughts from the UMH community?

edit flag offensive delete link more


Agree with this approach. I implemented IPLV formulas in Excel so that I could reach the desired IPLV minimizing the variation in a default cuadratic curve coefficients (using Solver). I'd have to polish it to be shareable though.

ecoeficiente gravatar imageecoeficiente ( 2017-02-03 07:11:14 -0600 )edit

Would it be prudent to find an old manufacturers catalog with the baseline performance and plot the EIR curve to see what it looks like?

rraustad gravatar imagerraustad ( 2017-02-03 07:29:56 -0600 )edit

That is encouraging to hear. My solution uses Excel's GoalSeek. So far, it sounds as though two people on this forum have attempted something "rigorous" for part load performance of Baseline systems. Tell me if I'm wrong!

Note that it took me about 10 years of detailed modeling to get to the point of knowing enough to make the calculation and having enough time to do it, so I'm not pointing my finger at anyone!

I could argue that since few are doing this and USGBC hasn't complained, we shouldn't bother with it. What do you think? (or should I share my tool widely for others to use?)

Jim Dirkes gravatar imageJim Dirkes ( 2017-02-03 07:36:12 -0600 )edit

Richard, For a baseline curve, I can't find enough enthusiasm within me to work very hard at getting something more "historically correct". It's a fictitious coil; how good does it need to be?

Jim Dirkes gravatar imageJim Dirkes ( 2017-02-03 07:38:57 -0600 )edit

My impression is that detailed technical aspects like this are totally ignored by many modelers and also by most USGBC reviewers. Quite often I wonder why I take so much effort doing things as correct as possible when I am the only one aware of it... At least, deepest knowledge allows me to take advantage of the lack of specification to get better results without breaking any Ashrae's modelling rule.

ecoeficiente gravatar imageecoeficiente ( 2017-02-03 10:14:44 -0600 )edit

answered 2017-02-01 17:42:20 -0600

This is a very common dilemma for modeling against ASHRAE 90.1 as they don't provide default curve coefficients. Full load efficiency at rated conditions is easy to define but it's true there are an infinite combination of part-load curve possibilities that may equate to the same IEER/IPLV/NPLV.

I have leveraged the California Title 24 Nonresidential ACM Reference Manual (link HERE) which does provide default curve coefficients for cooling equipment. Many of the efficiency requirements for Title 24-2013 align with the requirements of ASHRAE 90.1-2010.

edit flag offensive delete link more


Just for connection - the coefficients in the CA T24 ACM are from / align with COMnet.

EDIT: from this other source, for CA T24, the curves are different - from the CA T24 ACM and COMnet....SUPPORTING CONTENT:

dradair gravatar imagedradair ( 2017-02-01 18:02:29 -0600 )edit

answered 2017-02-01 17:56:38 -0600

updated 2017-02-01 18:04:31 -0600

This is an outstanding question we have run into time & again. Especially since 90.1 does not address.

AN approach we have take is to work with a rep to get minimally compliant Screw and Centrifugal curves and generate 'custom' curves using an in-house tool. A major issue, as @Jim Dirkes, alluded to is that the EER and IEER and contradictory goals in some/many/all cases. Finding chillers that even come close to both values has been...challenging. So for example; 'savings' are left on the table when a chiller with a better IEER is picked b/c is the 'worst' performing chiller EER.

This approach was initially taken because we found the 'default' curves that come in eQuest to be unrealistic when compared to actual selections. An example would be the performance of the a Centrifugal chiller true at low Part Load Ratio (PLR) load conditions. The EIR/PLR ratio for a the default shows a drastic drop-off at part-load, whereas a 'real' chiller will not exhibit this behavior. Translation: The default chiller unloads unrealistically good; placing the Proposed/comparison at a significant disadvantage at part load.

Note: I thought the eQuest curves (mostly?) align with COMnet chiller curves; though my quick spot checks are leaving me second guessing...... It is also not clear from my (quick) re-review of COMnet, and I don't recall, seeing a source for these curves sited in the text. I personally find these curves hard to believe - both COMnet and eQuest- , but I also like data. (Engineer here). A key discrepancy is the use of a single quadratic for EIR-FPLR in COMcheck vs a Bi-quadratic in eQuest. Though, both aren't terribly 'believable' when we've compared them to actual chiller selections. I'd love to hear/review any sources anyone can share or point to.

I have asked the question as to why ASHRAE does not supply/specify this information. The answer I have gotten has been something along the lines of "...its not ASHRAE's position to specify equipment curves - its up to AHRI/ARI since its their domain to rate equipment...."...I'm paraphrasing, but same idea. These leave modelers in many cases in an odd spot. I don't think it is unreasonable to have a standard set of curves which define the performance of a Baseline piece of equipment. Maybe there's a committee for that, unknown.

This is an area that is lack solid guidance on and would also like to hear others' take on the approach; even if it is 'default and run'.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer


Question Tools



Asked: 2017-02-01 15:55:34 -0600

Seen: 734 times

Last updated: Feb 02 '17