Hi All,

I would like to compare two screw chillers with same full load kW/ton and

different IPLV kW/ton. I am using the full load number for the EIR. Where

can the IPLV number be used? Is there some way to scale the performance

curves to reflect different IPLV's?

-Rohini

Rohini:

Won't happen. IPLV is a number which has one equation and 4 unknowns. Actually,

three unknowns as you know the 100% point. You need a manufacturer's curve run

at 85?F constant condenser water temperature. Let eQuest do the curve bending

after that. ARI curve won't work right.

John R. Aulbach, PE, CEM

John,

If the only parameter Rohini enters into the program is the full-load EIR ? which he said is the same for both chillers ? how will eQuest know to bend the curves differently for the two chiller options. It is my understanding that program would use the same chiller efficiency curves for both machines even though their IPLVs are different.

In order to overcome this issue, are you saying Rohini should get chiller performance data from each of the manufacturers and enter his own custom curves? I recall there was a discussion on that topic a few months back.

Regards,

James Waechter Jr., P.E., CEM, LEED A.P.

James et al,

Rohini will need to create performance curves to make this particular comparison. Otherwise using the default curves will result in identical part load performances.

The following is excerpted from the following (short) recommended reading concerning what IPLV means: http://ashrae-cfl.org/2010/03/understanding-iplvnplv/

IPLV (or NPLV) = 0.01A+0.42B+0.45C+0.12D

Where:

A = COP or EER @ 100% Load

B = COP or EER @ 75% Load

C = COP or EER @ 50% Load

D = COP or EER @ 25% Load

John is saying knowing only the IPLV and the 100% load condition efficiency (variable ?A,? above) is not enough information to extrapolate the IPLV curves to compare the two chillers.

For the exercise of comparing the two non-centrifugal chillers of the same technology/type, I would just focus on making a custom EIR-FPLR (not EIR-fPLR&dT) for each chiller, and use the library CAP-FT AND EIR-FT curves. My understanding is the effects of temperature on capacity and EIR outside of centrifugal VSD chillers is negligible. NOTE: Whenever using any of the default library chiller performance curves, that means you MUST normalize to and specify the chiller at ARI conditions ? those curves aren?t normalized to anything else.

So! If you can find the IPLV A, B, C and D values for both chillers? curves, you could come up with your EIR-FPLR curve coefficients (curve type = quadratic or cubic) using a curve-fit in excel? or alternatively make eQuest figure the coefficients by entering those points as raw data.

You may more easily just make your own curves, following John?s advice and getting part load unloading curves (100%, 90%, 80%... etc) held at a constant chilled & condenser temperature to match the ARI* conditions (85CWT if this is water cooled) at which you?re specifying the chiller capacity/EIR. Again, you could either go into excel, normalize the data (review DOE2 help entry for EIR-FPLR), make a scatter chart, and get the coefficients using a curve fit? or enter the data as raw points into eQuest and the coefficients will be figured by eQuest? whatever makes more sense to you.

For others and personal future reference?. If you are looking to make an EIR-fPLR&dT curve (for centrifugal chillers or otherwise): Ask your rep instead for multiple (minimum 3) part load unloading runs, holding the delivered chilled water temp constant, and vary the condenser water temperature incrementally for each run (i.e. 85, 75, 65). Choose a range of CWT?s that cover the anticipated range to be encountered in the actual design. This will get you enough data to have the minimum 3 delta-T?s represented in your part load data points to build this curve correctly. You?ll be using eQuest ?raw-data? entry method to make it generate the coefficients.

~Nick

* Rather than ARI conditions (85F CWT), you could be normalizing to Design condition EIR/CAP, provided you?re making a full set of custom curves in the same fashion.

NICK CATON, E.I.T.

Thanks all. As I thought all along, seems like there is no easy way out.

Any ideas on how HAP/TRACE treat chiller simulation? Do they use IPLV's or

do they have many more options for performance curves or is it same as

eQuest?

Actually what everyone is calling dt is more accurately referred to as lift. It

is the difference between the saturated evaporating temperature and the

saturated condensing temperature, which is different than the condenser water

temperature and the evaporator water temperature.

A more accurate curve can be built if you have the chiller manufacturer model

both these variables for you, rather than leaving the evaporator water

temperature constant and just varying the condenser water temperature. Any

change in the evaporator pressure will effect the condensing pressure and any

change in the condensing pressure will effect the evaporator pressure.

Paul

Paul, I may not have been crystal-clear, but if someone is merely making a comparative analysis between two screw chillers, aren?t the relative effects of varying lift between the two chillers negligible? My general understanding is that lift is a more critical variable when comparing VSD centrifugal chillers? This is why I suggested a simpler EIR-fPLR curve would be sufficient in lieu of an EIR-fPLR&dT ? I was trying to simplify Rohini?s comparative analysis.

I?d agree that an EIR-fPLR&dT curve would be more precise and more appropriate if the goal is to better match the chiller behavior (and creating custom CAP-FT and EIR-FT would be even better), but I was thinking this would require an unnecessary amount of extra work for Rohini?s comparative purpose.

My ?suggested information to request? below for constructing EIR-fPLR&dT curves is based on my past experience with limitations of my local manf. rep?s software ? they need to set certain items constant to get the numbers to crunch? Have you had luck collecting PLR runs where the evaporator and/or condenser temp was allowed to float? I?ve picked up through the lists that a better way to skin the cat may be to approach the chiller controls reps where they may exist, as they may have more flexible software?

I throw this disclaimer out sometimes: I certainly haven?t been doing this for decades! If I?m misunderstanding something, I very much welcome corrections ;).

Thanks,

~Nick

NICK CATON, E.I.T.

Rohini,

Good question... the capacity for non-eQuest programs to handle an IPLV

inputs may be better addressed separately to the [bldg-sim] list.

I'm not sure whether HAP or Trace can explicitly use IPLV figures to

help determine part load curves, but John makes a fair point: Math

fundamentals are getting in the way... you can't solve for a curve

(quadratic - 3 unknowns or cubic - 4 unknowns) with only one weighted

efficiency. No program should be able to determine part load behavior

knowing only the IPLV figure - at best a program might assume a part

load curve shape and "re-factor" it to match the IPLV - you'd probably

be better off assuming the IPLV weightings are appropriate to your

project and just apply a straight factor (IPLV1/IPLV2) to the chiller

energy consumed.

The ability to generate a curve by inputting the IPLV values (at

25/50/75/100) could be a UI-improvement/timesaver for any modeling

program (hint-hint, developers!), but note this is really already built

into eQuest in the form of the raw-data entry method of defining a

curve... My suggestion to work out the normalized data and/or

coefficients separately in excel is simply a way to

document/check/visualize your work to avoid any major hiccups.

~Nick

NICK CATON, E.I.T.

Nick:

I come from a mechanical contracting and service background. Condenser water

temperature is a hotly debated item. Some manufacturers define a very narrow

range of acceptable condenser water temperature and others define a very narrow

range of the temperature difference between entering and leaving condenser

water temperature.

During one investment-grade audit I performed recently, Trane told me that the

condenser water dt can be no less than 5 degrees for maximum efficiency on the

650 ton CVHE chillers I was working with. The water temperature could be

anywhere to 50-80 degrees. This machine had hot-gas bypass to keep the head

pressure up.

Other chillers that I have overhauled require a very narrow band of condenser

water temperature, such as 75-78 degrees supply temperature. So modeling a

wide condenser water range does not make much sense to me since it is not a

real world application. The chillers I have worked on either have a tower

bypass loop or a hot gas bypass to keep the condensing pressure up where it

belongs. Energy is wasted when too cold a condenser water temperature is

specified.

Other manufacturers say the colder the better.

Specifying variable primary chilled water flow and variable condenser water

flow has a large impact on chiller efficiency. I ask the manufacturers to

provide me the kW/Ton for minimum flow, standard flow and maximum flow for the

evaporator and condenser bundles. A chiller that has an kW/Ton of .56 at AHRI

conditions can often have a 1.3 kW/Ton at minimum condenser flow (3 gpm/ton) at

30% load.

The IPLV is really a useless rating for a real world application since it

assumes a certain percentage load a certain percentage of the time. It all

depends on the design of the system and the load profile. It is a good rating

to compare various chillers if they conform to the load profile. I see more

chillers that run in the 40-60% range 90% of the time than I see chillers that

match the IPLV conditions.

During my investment-grade reviews with our local utility, the lift of the

chiller is always an important consideration.

There was an e-mail for someone, I believe York, that offered to model any

manufacturers chiller on this forum a few months ago. I have had good luck

getting the rep to provide the information that I requested as long as I find

the right person.

Paul

Thanks very much for the informed response, Paul!

With all respect, I want to be sure I?m picking up the right lessons here =):

Accepting the arbitrary nature of IPLV ratings, what I?ve proposed for Rohini?s case is to define a curve that will utilize his model?s unique annual part-load profile (tossing IPLV?s weightings out the window), and would isolate the effects of PLR to make his equipment comparison.

Am I correctly understanding that any comparative analysis that isolates only the effects of part-load in comparing two chillers is pretty much pointless?

I am not trying to make a case that lift, variable flows, and their effects on capacity and efficiency are not important (and I have also experienced responses all over the map from different manufacturers). I agree that at least considering all of these in many cases is necessary when one wants to model accurate behavior, particularly from a commissioning/servicing context. With all other things being equal, I?m however proposing these factors shouldn?t be critical if the specific goal is to determine whether chiller A or chiller B fits a model?s part load profile better.

So to clarify and make our bridges meet: Is it critical that Rohini creates an EIR-fPLR&dT curve for his analysis (between two screw chillers with everything else presumed identical), or is it fair to say this may be overkill considering what he?s trying to achieve?

I for one will continue the good fight to obtain more solid input data for my curves, sometimes I have to settle for ?the best I can get? with the people I?m supposed to be talking to, but your experiences are further motivation to try to find the right people =)!

Thanks again,

~Nick

NICK CATON, E.I.T.

Nick:

I agree with your analysis. A curve built using different kW/Ton for the four

different load ratios that are used in the IPLV calculation would be sufficient.

I would just be careful when creating a curve that uses different condenser

water supply temperatures. It is more of a design issue. There has to be a

reason why the condenser water temperature varies. It will effect the lift and

kW/Ton. For example, current chillers will tolerate a much lower condenser

water temperature. Does this mean that it is more efficient to provide the

lower condenser water temperature? Not necessarily, since energy will be wasted

running the cooling tower fans to achieve the lower temperature and energy will

be wasted for the hot-gas bypass to bring the condenser pressure up.

In my experience, the lower condenser water temperature is most often an issue

on chiller start-up in cold weather. Some chillers will surge until the chiller

warms the condenser water loop. Surging sounds like someone with advanced COPD

breathing.

So, if Rohini's chillers are supplied with a nearly constant condenser water

temperature, the EIR f(PLR&Lift) would not be critical.

Please see attached modeling data from Trane for a CVHE 450 ton chiller. All

the data to build the various curves is shown. Note the difference in condenser

bundle pressure drop for standard flow v. minimum flow. Much energy can be

saved with a variable condenser water pumping strategy.

This data was used to calculate the loss of chiller efficiency due to reduced

condenser water flow. In this particular project, I had three existing

chillers. Total savings realized by the variable condenser water scenario was

250,000 kWh/Year. The reduced chiller efficiency consumed 50,000 kWk per year

so the net energy savings was 200,000 kWh/Year. Notice that the condenser water

temperature varies in the data. That was based on the maximum heat rejection of

the towers during the cooling season with a minimum of 65 degrees condenser

water supply temperature. If I assumed a constant 78 degrees condenser water

supply temperature the kW/Ton values would be much different.

It takes a little time to find the right person to provide this chiller data,

but it is well worth it.

Paul