Question-and-Answer Resource for the Building Energy Modeling Community
Get started with the Help page

# Cooling:Electricity meter usage exceeds coil capacities EnergyPlus

Hi,

I am working with DOE Commercial Reference Building models to do some simplified, high-level analysis of what might be expected to be 'typical' levels of the coincidence of the load on HVAC systems in commercial buildings and utility system (transmission grid) peak demand. For the case described here, I am running the post-1980 Full Service Restaurant model for the Houston weather zone as developed, using EnergyPlus 7.2 (latest version in which models are available on DOE website).

My issue is that the models appear to be telling me something impossible: that cooling electricity energy use in several hours (77 hours, when run using TMY3 data) exceeds the amount of electricity the HVAC systems should draw when fully loaded. To arrive at this conclusion I obtain hourly Cooling:Electricity from the Meter file, and I obtain the HVAC systems capacities as estimated from the sizing run from the Cooling Coils table in the Equipment Summary of the HTML Tabular output report.

According to the .mtd file, this is what is rolled up on the Cooling:Electricity meter:

For Meter=Cooling:Electricity [J], ResourceType=Electricity, EndUse=Cooling, contents are: PSZ-AC2:2COOLC DXCOIL:DX Cooling Coil Electric Consumption PSZ-AC1:1COOLC DXCOIL:DX Cooling Coil Electric Consumption PSZ-AC2:2COOLC DXCOIL:DX Cooling Coil Crankcase Heater Consumption

For capacity from the sizing run, I use the sum of the Nominal Total Capacity of the two units as reported in the Cooling Coils table of the Equipment Summary (Table report). To convert this (output) capacity to input demand, I divide by the COP.

My basic question is, how can this happen? Have I misunderstood/misused either the Nominal Total Capacity of the cooling coils from the Equipment Summary or the Cooling:Electricity meter 8,760 data as data sources for this comparison?

Further confusing matters, given use of a cooling sizing factor (Sizing:Parameters object) of 1.2, I would expect the maximum ratio of hourly cooling energy use (demand) to capacity to be closer to 1/1.2. Design day conditions used are standard - ASHRAE Fundamentals 0.4% design day, and I've run with both TMY2 and TMY3 data (FWIW, it's only with the TMY3 that I get demand in excess of capacity - with TMY2 I do get a coincidence factor less than 1, but not much less - 0.96).

Any insights would be greatly appreciated.

edit retag close merge delete

Sort by ยป oldest newest most voted

The peak coincidence ratio can be expected to be unity because at some point, say during morning pulldown, the unit is going to be operated full on. Whenever full-on operation occurs during times when conditions are more demanding than the rating point, the unit will draw more power than nominal capacity suggests.

Keep in mind that the sizing factor in that model is also being applied in the context of a load averaging window of one hour. That will smooth out the sizing impact of recovering from thermostat setback schedule changes. But once the modeling gets to the final simulation that averaging is not occurring and the actual loads when recovering from setback will usually end up large compared to those during sizing. You might see the 1/1.2 ratio you expect if the system was inadvertently sized so large as to be able to fully recover in one zone timestep. But that is considered poor practice and to be avoided by either using flat thermostat schedules during sizing or a long averaging window.

more

Thanks @Archmage! For my definition of peak coincidence I am looking at the hours between 4 and 6 PM on hot summer days. While that is a period of extreme loading on cooling coils, i've confirmed that there's no interaction with a thermostat setpoint change in my model. More fundamentally, your answer suggests that the ASHRAE Fundamentals 0.4% design day sizing method significantly undersizes systems for the loads they are likely to encounter (e.g. by 20%) before application of a sizing factor. i do expect some loads to not be met, but I did not expect by that much!

( 2015-01-30 11:08:15 -0500 )edit

I believe you may be getting tangled up in the meaning of nominal capacity. As discussed in this related question nominal capacity is what happens at standard rating conditions. It is not the capacity of the coil at the conditions you are running at. Secondly, the COP that you specify as an input to the cooling coil, is at rated conditions, it is not a constant value.

more

This answer is based on the v8.2.0 example file RefBldgFullServiceRestaurantNew2004_Chicago. EnergyPlus sizes supply airflow rates to meet the peak zone sensible load (for this model cooling will set the airflow) and the sizing factor is applied to this flow rate. Then the coil is sized to meet the specified cooling supply air temperature at that flow rate at the design mixed air conditions. All of this sizing is based on the SizingPeriod:DesignDay conditions which in this case is 31.1C (88.2F) outdoor dry-bulb.

The reported rated (nominal) total cooling capacities for the two DX coils are 68.82kW and 18.19kW. With COPs of 3.23 and 3.67, this results in a rated total power input of 26.26kW. The power input for a DX coil at any point during the simulation is (RatedCapacity/RatedCOP)CapfTCapfFlowFracEIRfTEIRfFlowFrac*(PartLoadRatio/PLFCurve). All of the curve values can be reported as output:variable "Performance Curve Output Value". For an annual simulation with Chicago O'Hare TMY3, the peak hour cooling electricity is 26.65kW which is about 1.4% higher than rated.

The peak cooling electriicity occurs on July 19 16:00, with an outdoor dry-bulb temp of 34.3C (93.7F) which is well above the design day condition, but still below the rating conditions of 35C (95F). So, if you look at the curves, they are showing CapfT >1.0, EIRfT <1.0, which multiplied together give a total power draw for coil1 of 0.99, and coil2 1.03. Because this day in the TMY3 file is hotter than the cooling design day, the extra capacity is being put to use, so there is no discount for the 1.2 oversizing.

more

Thanks, Michael. In short, then, it seems that the answer to my follow-up question to Archmage is "yes." At least in your example and in mine, a system that is not oversized by 20 percent according to the summer design day method will fall far short of meeting its load at conditions under which it is very likely to have to operate. Thanks to all for the helpful insights!!

( 2015-01-30 15:13:32 -0500 )edit

Kyle, thank you for responding to my post. Your comments are correct - the nominal capacity and the cited COP are indeed at rated conditions, which makes my calculation crude at best. You are also correct that my issue is similar to the referenced post. Specifically, like the creator of that post, I only see sensible loads in the zone sizing and system sizing reports, which are lower than the nominal total capacity by about the same 20% as my sizing factor (1.2). However, as in the cited section from the I/O Reference Manual, I take that difference to be due to the inclusion of latent loads (and, yes, also incorporate the differences between design operating conditions and rated operating conditions).

I am going to work on a thing or two and revise my original post shortly - I still have a core issue, which is that - even using not exactly right capacities and COP values to estimate full load input electric demand on the cooling systems - I would expect the peak coincidence ratio to be much closer to 1/1.2 than to unity given that the sizing run applied a 120% sizing factor.

more