Why is autosizing leading to greatly oversized equipment?

asked 2017-10-17 10:44:09 -0600

LiamOBrien
81 ●3

updated 2017-12-28 12:50:46 -0600

__AmirRoth__

4436 ●5 ●17 http://bleedinggreenna...

Dear group: (I submitted a similar question to the Yahoo group and got a few helpful ideas (thanks Julien Marrek!) that didn't resolve my problem. After days of troubleshooting, I am turning to this group). Here goes...

I am trying to autosize a boiler and chiller in a medium sized office building using EnergyPlus directly. My goal is to look at the comfort impact of undersizing equipment. Therefore, I need to find the minimum possible size to meet loads on the design days. I have set the sizing factor to 1 everywhere, autosized most fields that allow it, and run a simulation for just two design days. I have also smoothed the setpoint schedule to avoid abrupt changes and corresponding peak loads. To my bewilderment, the part load ratios for the chiller and boiler barely exceed 0.5 (see graph: https://www.dropbox.com/s/rw5nke3lwcf...). I'd have thought they'd hit nearly exactly 1.0, given the iterative approach that EnergyPlus takes (I allowed 10 passes).

My only guesses are: 1) the design days are somehow using schedules or design conditions other than those I specified or 2) there are some complex interactions occurring between all of the autosized equipment. But a systematic exploration of the issue has not yielded anything.

So I thought I'd be smart and pick the peak load out of the design day in an autosized run and call that the minimum equipment size. Perhaps not surprisingly, this didn't work as the corresponding size seems to be ~10% too low or so. But notably, when I used this size and allowed the maximum part load to be 2, the part load exactly reached 1.0.

Any insights would be greatly appreciated. Thanks in advance, Liam O'Brien

answered 2017-10-17 12:18:24 -0600

rraustad

13804 ●79 ●12 http://www.fsec.ucf.edu/

updated 2017-10-17 13:57:51 -0600

The trouble with autosizing equipment is that you don't know how the equipment will respond to actual conditions encountered during the simulation. For this reason, users enter a supply air temperature (and humidity ratio) in the Sizing:Zone object. The supply air temperature and zone set point temperatures are used to "size" the zone air flow rate needed to meet the zone load. Using this air flow rate and the corresponding coil inlet air (mixed) condition (which the simulation tracks during sizing) along with the SAT conditions entered in the Sizing:Zone (or Sizing:System depending on the equipment type), the coil capacity can be calculated.

From the simulation, look at the zone entering air temperature(s) and compare that temperature to values entered in Sizing:Zone and/or Sizing:System. Adjust the sizing SAT values as necessary. This can be an iterative process since changing these values will change the zone air flow rate, and changing the zone air flow rate can change the mixed air temperature if OA is included in the simulation.

When the chiller and boiler are sized, all water coil equipment is sized (similar to description above) and then the required coil water flow rate is reported to the plant. These values are summed and used to size the chiller/boiler using the plant delta T specified in the Sizing:Plant object. Since all coils will likely not be at the peak load at the same time, the boiler will be oversized. You could add a sizing factor in the chiller and boiler as, for example, 0.75 for a diversity factor since you know the capacity is oversized since all coils do not peak at the same time.

edit flag offensive delete link

Comments

Thank you Richard; this is useful insight. I'm still left wondering: is it/should it be this complicated? I ran one of the example files (LargeOffice in Chicago) and it behaved exactly as I'd expect with the part load hitting 0.999 on the design day, so I'm inclined to think some parameter(s) in my model are at fault. I will close the loop on my original question if and when I get to the bottom of this. Thank you.

LiamOBrien ( 2017-10-18 01:29:33 -0600 )edit

@LiamOBrien did you ever find a solution to this? One other issue that may cause oversizing is the load averaging window, see Sizing:Parameters Timesteps in Averaging Window.