Myoldmopar
commented
10 years ago OK, so this isn't fixed yet. Gah.
Problem
A file from @kbenne, with common pipe and constant speed supply branch pumps was showing the second supply branch running even when it wasn't needed. It looks like the bookkeeping goes well within PlantLoopSolver::SetupLoopFlowRequest.
- The demand side has a flow request of only 4 kg/s.
- Each chiller branch is capable of like 50 kg/s. The supply side needy-if-on request is then about 100 kg/s.
- Then, after the loopsides are swept to calculate their flow requests, a final flow request is calculated. Because there is a non-zero need-and-turn-on flow request on at least one side (the demand side), the flow request is calculated as the max of the two sides.
- This inevitably turns on both pumps.
Even though this is a common pipe file, this makes sense because if we didn't, the demand side would keep running getting hotter and hotter, and the flow would all go down the common pipe. So the issue isn't that the pumps should only respond to supply requests.
Possible Solutions
- It is possible with the current code and inputs that we could be more precise about what we turn on. In this case, we only need 4 kg/s, so we really should only turn on one chiller. However, we don't know that for sure, because we aren't talking load distribution yet. We could hack it in so that it only turns on 1-n branches until the n'th branch meets the flow request. But even though the flow request is low, we may actually need both chillers to meet the load (heat transfer) needs. This could be investigated within the scope of a bug fix.
- It is possible we could go one step farther and add an input for determining flow distribution logic. Perhaps a set of keys such as "AllOn" or "Sequential". This would only be used on loop sides with branch pumps, probably only constant speed, and would be used only when the demand side flow (needy-and-turn-on) request is less than the supply side (needy-and-turn-on) request. That is the situation of this example file. This is on the border of bug fix / new feature...probably more toward new feature.
- The next possible solution would probably require major surgery within the plant. Restructuring the plant so that it doesn't fixate on this overall loop flow request so early, instead using it as a predicted value, and then in the flow-locked (corrector) step, re-do the loop flow request calculation with information about load dispatch being included. This would be much heavier lift than a bug fix.
jasondegraw
mjwitte
kbenne
EnergyArchmage
xwanguk
Branch pumps on when they shouldn't be
Added on 2012-02-14 13:10 by @mjwitte
Description
MJW 14 Feb 2012 User took 5ZoneBranchSupplyPumps example file, changed the chillers and their branch pumps to be constantflow, changed the loading to be sequential, and also changed the equipment operation scheme to make single chiller available at low loads.
Both chiller branch pumps run whenever there is a chilled water load, even when the load is small and only one chiller is operating. Would expect, just one pump to operate until the second chiller is needed.
This is resulting in poor ChW supply temp control - setpoint is 7.2C, but supplying 7.8C, and the operating chiller is putting out 5C chilled water - adversely impacting its efficiency.
MJW 16 Feb 2012 It's not just constant flow branch pumps. Even the straight 5ZoneBranchSupplyPumps example file does not run as it used to. The two pumps always have the same mass flow rate. Wasn't that way in v6. Added new test files, 8749-5ZoneBranchSupplyPumps-Annual, and 8749-5ZoneBranchSupplyPumps-v6-Annual. In the xlsx files, look as September 19 as an example case where in v6, only chiller 1 and its pump are operating, but in v7, both chillers are operating, chille 2 with a tiny load.
Inputs: 8749-* Weather: Chicago
MJW 10 Apr 2012 v7.1 files checked in. look at 5ZoneBranchSupplyPumps-v7.1Annual.xlsx which has results for v7.1.0.006. On 4/10 at 11:00 and 12:00 the Chiller2 evap haet trans heat is 0.0 but the Chiller2 Pump shows flow and interesting enough its exactly what the Chiller1 Pump flow is. Both chillers always have the same flow even though the evap heat transfer is different.
[BG 19 april 2012] Cannot reproduce in current starteam code. I believe this was fixed by changes made for CR 8732 that came in after 7.1.0.006, specifically the branch flow request determination for series active branches.
MJW 19 Apr 2012 Congratulations - it's working! Well, almost.
I've looked at 8749-5ZoneBranchSupplyPumps-v7.1Annual.idf and the dual chillers and companion branch pumps are working as expected - pump is on only when chiller is on, loads are distributed as specified (even tested and compared optimal, sequential, and uniform loading.) Chiller min supply temp is honored. This has variable flow Chiller:Electric with variable flow branch pumps and looks great.
Next is 8749-5ZoneBranchSupplyPumps_constant_scheme-morereporting which has constant flow chillers and branch pumps, also using Chiller:Electric, and some changes in the operation scheme. **Still have a problem here - the chiller 2 pump is running whenever the chiller 1 pump is on, even if chiller 2 is off.
Took the first file (sequential version) and changed to constant flow chillers and pumps (no other change) - same problem.
Noticed that the pumps have a flow schedule on them - deleted that - same behavior.
Additional idfs checked in.
Do we need to test other chiller types, such as Chiller:Electric:EIR just to be sure?
[BG 4-24-2012] checked in source to fix this. marking fixed
MJW 04 May 2012 Note to self - Brent checked in more related to this on May 1, plus lots of other plant changes happening now, so hold on verifying this until that's all checked in.
Fix
2012.March.21 ESL Improved pump and solver logic in two ways: 1) variable speed branch pumps now don't request flow 2) loop solver considers all components on variable speed pump parallel branches, not just the inlet node CR file now runs properly.
External Ref: Ticket 5351 Addressed in
12.04.25 V7.1.0.009by Brent Griffith Last build tested:12.04.25 V7.1.0.009