[Bldg-sim] Cooling Design Calc Methods

[Pegues, James F CCS]

Brendan,

A couple thoughts on your comparison.  Having done comparisons like this with other tools and having worked with teams doing these comparisons with multiple tools, my experience is there are many pitfalls that its easy to fall into which can lead to a comparison of results that are not apples-to-apples. That makes it difficult to draw meaningful conclusions from the comparison.  The pitfalls that may be in play for your comparison:


1.     Heat Gain vs Load – I suspect the component data from your EnergyPlus sim tool is “heat gain” data while the component data from HAP is “cooling load” data.  Therefore the comparison of results is heat gains versus loads and is not apples-to-apples.  Heat gain, of course, is heat from the individual sources like solar, occupants, lights.  The Load is heat transferred to air after going through the conversion process in the room involving conduction, radiation and convection.  I suspect the EnergyPlus results are heat gain for two reasons:


a.     In your “Component Comparison.PDF” your EnergyPlus plots look much more like heat gain profiles than load profiles.  For example, the solar profile only has values when the sun is shining.  Solar values are zero at night  If this is a load profile that means the building mass has zero heat storage capacity.  The lighting and occupancy profiles are completely flat during mid-day which also looks like a heat gain profile, or else a load profile in a zero mass building..


b.    As you noted EnergyPlus uses the “Heat Balance” load calculation method.  This method is a truer representation of building physics than other methods the industry has used like Transfer Functions and TA/TETD.  True building physics says there is no such thing as a component load like a “wall load” or a “people load” or a “solar load”.  This is because once a heat gain enters a room, it gets mixed together with all the other heat gains through radiation and convection processes.  All you get from Heat Balance is THE LOAD – total heat convected to the room air for the hour.  In that Load it is impossible to determine how much came from each individual heat gain source, so component loads cannot be defined.  Therefore an EnergyPlus calculation should not be able to produce a report of “component loads”.  But it could produce a report with “component heat gains”.


The Transfer Function Method (TFM) uses three key simplifying principles to speed up calculations.  One of these, the Principle of Superposition, results in loads being computed component by component.  That allows tools based on TFM to report component loads.  That’s part of the compromise of TFM – you get speed and component loads in exchange for some amount of accuracy and inability to see component loads.


2.     Apples to Apples Input Data – A common pitfall is not completely matching input data between the two programs.  Often this is caused by programs requiring different sets of inputs, using inputs in slightly different ways, or using similar terms with different meanings.  Often there is more than meets the eye to inputs and that can have adverse results in a comparison.  There isn’t enough information provided to say whether this happening in your comparison, but a couple things that would be worth considering:


a.     Weather Data – You mention you used 95/75 as the design condition.  More important is whether the complete 24-hour design day DB and WB profiles match.  And also what solar data was used.  HAP is using an ASHRAE clear sky profile for solar radiation for the 21st day of each month.  Is EnergyPlus using something similar?


b.    Steady-Periodic – Calculations using a single design day each month are aimed at reporting steady-periodic behavior.  Because heat transfer in the building is transient, you inevitably have to start with assumed initial conditions and then run the 24-hour cycle multiple times to damp out the effect of those initial conditions to get to steady periodic.  This is what HAP is doing.  If EnergyPlus does not automatically do this, it is important to force calculation of possibly 4-6 consecutive days of the same weather and heat gain conditions to get to an equivalent steady periodic set of results for comparison.


c.     Internal Loads – Not all internal heat gains like lighting, equipment and occupants are created equal.  1 W/sqft lighting can have different load results, for example, depending on the fixture type because that will cause a different split between the radiative and convective fractions of the heat gain.  For example, load behavior of a 100 W heat gain from free hanging lights will be different from that for recessed unvented fixtures.  Equipment and occupant heat gains also have radiative and convective splits depending on the type of equipment or the type of occupant activity.  Its important to match the assumptions about radiative / convective split between the two tools being compared to get apples-to-apples load results.


d.    Envelope – Wall and Roof assemblies need to have exactly the same sequence of material layers outside to inside, with each layer having a consistent set of thermal properties – thickness, conductivity, density, specific heat.


3.     Floor Conduction Heat Loss – Just based on intuition it looks to me like something strange is going on with your floor conduction heat loss. Its odd that the largest heat loss is occurring in the middle of the day when it is warmest outdoors and that heat loss magnitude is a significant percentage of the heat gain magnitude from other sources.  For this behavior to occur my guess is the surrounding soil temperature would have to be significantly colder than the indoor air and surface temperatures in the room.  Since this is a slab floor I wouldn’t think the near-surface temperature of the soil would be that cold on a 95 F day.  In addition, if detailed ground heat transfer calculations are being used in EnergyPlus, the heat transfer has very long transients on the order of weeks or months.  Could the huge heat losses represent the effect of initial conditions that assumed a cold ground temperature or a deep ground temperature?

Hope this helped in some way.  Would be interested to hear how your comparison turns out.

Best Regards,

Jim Pegues
Manager, Carrier E20-II Software Development
United Technologies / Climate | Controls | Security


From: bldg-sim-bounces at lists.onebuilding.org [mailto:bldg-sim-bounces at lists.onebuilding.org] On Behalf Of Hall, Brendan
Sent: Thursday, July 05, 2012 2:43 PM
To: bldg-sim at lists.onebuilding.org
Subject: [Bldg-sim] Cooling Design Calc Methods

Hey all,

I am looking at an energy plus based sim program and one of the major selling points would be doing the load calcs in the same program as the energy modeling. From what I have read, the heat balance method that E+ uses should be more accurate than the transfer function method that a widely used load calc program like HAP would use. However the results are different so I wanted to get people’s thoughts on the subject.

I did a test case, a 20 x 20 office space (384 ft conditioned area) with one southern facing window.

-          10 ft tall

-          4 People

-          1 W/ft2 Lighting

-          0.5 W/ft2 Equipment

-          ASHRAE Office default schedules

-          Basic Walls and Roof, ASHRAE minimum windows (U=.55,SHGC=.4)

-          Design Weather – 95 db / 75 wb

I’m attaching some of the results, but overall the E+ calc has a lower peak load. Occupant, lighting and plug loads seem to follow their schedules. HAP looks like it very heavily weights the delayed load effect (TFM) even though I used a medium weight (70lb/ft2) wall (fyi - changing it to a lightweight wall has some but not a huge effect). HAP also ignores the slab heat loss in its cooling calculations. This shows up most clearly in the unoccupied zone temperature. In HAP the residual loads drive the temperature up to 82 where in the E+ calc the slab loss dominates, driving the temperature into the 60s. I could see leaving it out of the peak to be conservative but then allowing the unocc temperature to jump up overnight and having to deal with that load seems a bit too unrealistic. HAP also seems to over predict the peak roof conduction gain, probably due to the use of the sol-air temperatures, which I have read can over predict gains.

I tend to want to believe the E+ analysis because I know that it’s calculation methods are in general more rigorous but I am very interested what others may think, sim engines are known for being poor design load predictors (I’m looking at you eQuest) and HAP is a very established and trusted program.

Thanks in advance for anyone that feels like diving into this with me.

Brendan Hall
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.onebuilding.org/pipermail/bldg-sim-onebuilding.org/attachments/20120705/2cf5e664/attachment-0002.htm>