Hi all-
Sort of technical question here that requires some knowledge of the inner workings of either EnergyPlus or DOE2/eQuest...hoping there is someone out there who knows the answers to this. What does either an EnergyPlus or DOE2 user have to do when dealing with different units, particularly when it comes to the curve fits? Obviously, the same coefficients do not apply for both units. Do the default curve coefficients for the chiller, for example, have both IP and metric equivalents?
Does anyone know how this works? Has anyone bumped into a regression model that can take raw data in metric units, and create a set of coefficients in one set of units (let's say IP for starters), and then use those newly-created coefficients, generate an new set of raw data, and then create a new regression to produce a new set of coefficients in converted units (Metric, in this case?)
Has either DOE2 or EnergyPlus already solved this problem, and I don't need to even think about itanymore?
Thanks,
It is my understanding that the chiller curve data are dimensionless ratios, so the curve-fits should work regardless of units-system. The raw data would be different, but the curve coefficients should be identical.
Can someone confirm this?
Thanks,
Dan Russell, EIT
In eQUEST a curve that references temperatures or other dimensioned
variables must be fit using English units. We plan to address this at some
point in the future.
Not really. Some curves are the function of temperature and/or other variables with dimension. Although the results of the performance curves are usually dimensionless, the coefficients of the curves depends on the unit of the variables.
Dan,
I think I misspoke in my reply below.
I believe the standard version of eQUEST/DOE-2 allows you to input curve fit
DATA using either English or metric; if metric independent temperatures are
used, the Building Design Language processor should convert the metric
temperatures to English before doing the curve fit. The coefficients
generated would then expect hourly temperature in English. However, if you
do a curve fit independently of BDL using metric temperatures, and simply
give the program the COEFFICIENTS rather than the DATA, then the curve would
expect metric temperatures, and the hourly curve algorithm would feed it
English; resulting in a severe error.
So, as long as you generate the curve using BDL, I believe you can use
either English or metric units. It should be easy enough to test to
confirm. Just input the same curve with both English and metric temperatures
(specifying METRIC input), and see if you get the same coefficients. If this
is not the case, please let me know.
The refrigeration version of the program has a curve-fit routine that allows
either English or metric curve coefficients to be used; there is a keyword
that allows you to specify which values you are using. So the library could
have curves of both types in it. On an hourly basis, if the curve is
metric, the program converts the input temperatures from English to metric,
feeds them into the curve, and converts the output from metric back to
English (if the output has dimensional units). Hopefully, we will port that
algorithm over to the standard version in the future.
Steve
Chien Si, others,
The problem is more complicated than what you're asking, because EnergyPlus and DOE-2 do not necessarily express their curves using the same equation formulation. When I worked a few years back in translating DOE-2 inputs to EnergyPlus, I (or others in the then LBNL team!) had to reformulate many of the equations and coefficients, and in some cases, redo the curve fit from scratch. As several people have already pointed out, even if the equation is the same, the coefficients are unit-specific and hence differ between IP and SI.
Joe Huang
Hi Joe,
Could you please shed more light on this topic? --
Assuming the performance equations are the same in DOE2 and EnergyPlus. If I just wanted to convert the curve coefficients from IP units to SI units (for example, the chiller capacity performance curve (function of chilled water supply temperature and entering condenser water temperature, of type Bi-quadratic-T in DOE2), do I have to derive the unit conversion formula for the curve coefficients manually, based on the curve type and the unit conversion rule for temperature? As the curve coefficients in this case can not be converted independently.
I tried to derive the units conversion formula for one particular curve type before, and got the same results as those in the EnergyPlus library. But I am wondering if there is any more systemmatic way of converting the coefficients between SI and IP, so that the conversion work could be done in an automatic way. As you are one of the developers of EnergyPlus, I believe you are the right person to ask this question.Thanks,Dongyi
From: joe at drawbdl.comTo: chien.harriman at iesve.com; bldg-sim at lists.onebuilding.orgDate: Mon, 8 Dec 2008 17:05:14 -0800Subject: Re: [Bldg-sim] chiller performance curves?
Chien Si, others,
The problem is more complicated than what you're asking, because EnergyPlus and DOE-2 do not necessarily express their curves using the same equation formulation. When I worked a few years back in translating DOE-2 inputs to EnergyPlus, I (or others in the then LBNL team!) had to reformulate many of the equations and coefficients, and in some cases, redo the curve fit from scratch. As several people have already pointed out, even if the equation is the same, the coefficients are unit-specific and hence differ between IP and SI.
Joe Huang