Menu
▾
▴
2 Attachments
#460 Cooling Tower Loop Temperature Control
Originally created by: chitra.nbr (code.google.com)
Originally owned by: kbe... (code.google.com)@gmail.com
What steps reproduce the problem?
1. Run a large office model with CW Loop Supply temp of 71 F
2.compare Heat Rejection End use with version of same model with CW loop temp at 81F.
3.
What version of CBECC-NRes are you using? On what operating system?
v 1f(557)
Please provide any additional information below.
Unlike one would expect, the Heat Rejection End use energy is increasing for the greater CW Loop setpoint temperature. See attached .xml files
Related
Discussion
Originally posted by: chitra.nbr (code.google.com)
Changing this to high priority since the Large Office model updates for the Reference Test method relies on this.
Labels: -Priority-Medium Priority-High
Originally posted by: rhedr...@archenergy.com
Chitra, please upload your cibd file.
Originally posted by: chitra.nbr (code.google.com)
Roger, see attached.
Originally posted by: rhedr...@archenergy.com
Changing owner to Kyle.
I see no difference in the simulation xml files other than the change in <FluidSys>/<FixedSupTemp>. But I am seeing nearly constant operation of the cooling tower fan in the 81F supply temp case. I also don't see any change in the idf file other than the supply temp setpoint, although it is hard to tell with all the naming differences from OS. I've attached a screenshot from ResultsViewer of Cooling Tower Fan Electric Power, with the blue line from the lower temperature case, the black from the 81F case.
Owner: kbe...@gmail.com
Cc: rhedr...@archenergy.com
Originally posted by: rhedr...@archenergy.com
(No comment was entered for this change.)
Labels: -Priority-High Priority-Critical
Originally posted by: da...@360-analytics.com
Kyle,
I think this issue stems from the fact that the pump on the CT branch is VariableSpeed. We are testing some alternate combinations that make more sense to see if this is the root cause. If it is, we'll add some checks on our side to ensure that variable speed CW pumps are used appropriately. Will report back here with what we find.
Originally posted by: rhedr...@archenergy.com
I just ran cases with the CW pumps switched to constant speed with fixed CWST at 76F and 81F, and the results are as expected. I think we are OK, and adding checks in the rules will be adequate.
Originally posted by: da...@360-analytics.com
The little bit of testing I did with variable speed pumps on the loop (as opposed to branch) also don't control properly. The pump does operate between 50% and 100% flow with chillers staging, as expected, but the flow rate is 50% of the expected rate at each operating point. this leads to the CW temp floating upward, well beyond the control temp, and the CT fan runs nearly constantly trying to compensate.
I'd like to experiment with variable speed pumps on the loop more to see if we can figure out a way to make that work, but for now, I will add a CHECKSIM rule that checks for and disallows variable speed pumps on the CW loop.
Status: Resolved
Originally posted by: kbe... (code.google.com)@gmail.com
I'm going to post some images in a minute, but I also believe this is more of an defect (or at least limitation) of EnergyPlus.
Some take aways which I will try to support are:
Temperature might not be controlled correctly.
Changing ONLY the setpoint yields the strange outcome. That is I took the 81F model (idf) and adjusted only the schedule down to 71F and got (almost) the exact same results as from the original 71F SDD.
Mass flow through tower based on request from the chillers appears extremely close.
Some SDD inputs will result in a Design Inlet Air Wet-Bulb Temperature below 20C which is an E+ lower bound. In this case the value will stick at the OS default 25. Suggest a translator change to set the floor at 20 even if the SDD tries to set it lower. Keep in mind that this is a calculation based on supply temp - wet bulb approach.
Originally posted by: kbe... (code.google.com)@gmail.com
The following change was motivated by my investigation of this ticket, but has very little impact on what is being discussed here.
https://github.com/NREL/OpenStudio/commit/4d995832f54c1e2b52127febba63978ef57e5a01
I don't consider it a critical change and can simply be picked up in the next build, whoever that is.
Originally posted by: da...@360-analytics.com
Roger, I suggest we make the SDD property for Design Wetbulb Temp (HtRej:DsgnWBTemp property an explicit input in the input and sim XML model. This way we can check and warn users on the limit.
Originally posted by: kbe... (code.google.com)@gmail.com
Here are some images to try to explain what I see.
First of all the flow request moving through the towers seems almost identical between the two models. See the "flow comparison" attachment. This is a flood plot of the difference in flow between the models over the entire annual simulation. You can see it is awash in 0.
Now look at "flow". This is flow for the 81 model but it doesn't really matter since I just showed that they have the same flow rates. It looks pretty reasonable to me, unless I missed something. Nice swing throughout the year with peak in the hot summer days.
Then look at tower temps coming out of the 71 model in "71 temps" Looks pretty decent to me. We seem to be pretty much under control during the times when the tower actually has water moving through it. Sure it gets crazy and floats around when the tower is not moving fluid but I don't think that is a concern.
In fact if you look at the "71 power" fan power used by the tower has a pretty normal looking profile too.
Now look at "81 temps" Crazy town here. I don't know what is going on but the tower seems to be completely out of control, even though we are asking less of it. I can't explain it except E+ is weird. I verified that we don't have some other inadvertant property besides the setpoint schedule messing with things. I did this by taking the perfectly good 71 idf and modifying the setpoint schedule value (1 field) and got the same crazy outcome.
Just for completeness a attached the 81 power so you can see the tower running like crazy while still not achieving setpoint. You might think, well, ambient conditions just aren't allowing for it, but the same tower seems to have no problem maintaining 71. ???
I didn't try this with a constant speed pump, but the above reports suggest this issue goes away. Seems weird since the flows are the same if you ask me. I'm interested in seeing this same plots with the constant pump, but patience is running low.
Originally posted by: da...@360-analytics.com
A few additional observations to add to last few comments:
As mentioned in #8, the CW loop flow was observed to be about 50% less than the capacity of the CW pump. Comment #9 indicates the tower mass flow is consistent with the chiller condenser request, which is a good clue.
This is now starting to make sense..,
In the IDF, the Chiller:Electric:EIR !- Reference Condenser Fluid Flow Rate {m3/s} is set to 0.05165 m/s, or ~820gpm/chiller, or 1640gpm total for the 2 chillers. The CW pump capacity is 0.21217 m/s, or 3363gpm, roughly double the specified Chlr condenser flow rate. My guess is that the chiller condenser flow rate is being calculated based on the design heat rejection of the chiller and the ChW delta T (20F), as opposed to the CW delta T (10F). This would explain why the CW loop flow is roughly 1/2 what I expect it would be.
I wasn't sure if this explains the strange CW supply temps. I first iterated the CW setpoint temp from 70F to 80F in 2F increments, and found that the control goes "crazy town" when the setpoint temp is somewhere between 76F (24.4) and 78F (25.5C) (see image 1). I ran a model @ 25C and CWS temps still are erratic in June and Sept (see image 2). Not sure if this is a climate specific issue (all runs using Blue Canyon epw), but reinforces what Kyle indicated.
The last thing I tested was revising the chiller condenser flow rate to be consistent with the flow needed to reject the heat @ design conditions. I ran the same model with both 70F and 80F setpoints. As shown in the images below, the fan energy and CWS temps are as expected (see images 3 and 4)
Keep in mind, for the model results described in this comment, I moved the pump from being a child of the CT (i.e upstream of CT on a branch) to being a child of the FluidSeg (i.e upstream of the splitter). I think this should actually be equivalent to having the pump as a child of a single CT. I think that Eplus is hard-programed to not allow CW flow to vary through the chiller condenser, so really, I don't think we need to have a check for VariableSpeed pumps, but rather, that the condenser pumps are consistent with both CT capacity and chiller condenser required flow.
Here is the proposed resolution:
Add a property to the SDD to describe the Chiller Condenser Flow Rate
Update reverse translator to set reference the new property.
Add CHECKSIM rules to make sure pump flow rates (within a certain tolerance) are consistent with primary component capacities.
This resolution will be reflected in a new GC issue so it is address separately after Add/Alt rule syntax updates are complete.
One other observation in this model: There are two chillers, each with constant speed primary ChW pumps. Chiller 2 operates much fewer hours than Chiller 1, yet the pumps for each chiller operate the same number of hours, as evidenced in image 5 and 6. Not sure what the cause of this control issue is, but I will create a new GC to investigate it.
Originally posted by: cnambiar@archenergy.com
(No comment was entered for this change.)
Cc: cnambiar@archenergy.com
Originally posted by: kbe... (code.google.com)@gmail.com
This is pretty old. Can it be closed?
Originally posted by: kbe... (code.google.com)@gmail.com
oh... I see now it is marked resolved. Should resolved = closed?
Originally posted by: da...@360-analytics.com
I think this can probably be closed, but Roger should probably review the resolution and see if it was implemented.
Originally posted by: rhedr...@archenergy.com
(No comment was entered for this change.)
Status: DoneVerified