RE: [EnergyPlus_Support] Re: Thoughts for the Energy Modeling Community

December 2, 2011

Multi-Use Facility Energy Modeling: Lessons Learned

James V. Dirkes II, PE, LEED AP, BEMP

Multi-use facilities are everywhere. I had never modeled one of them, but I had modeled all of the component occupancy types, including a swimming pool; how hard could it be? Harder than I thought; read on!

Lesson #1: Heat Pumps are not very appropriate for certain occupancies

ASHRAE 90.1’s Baseline system selection does not care too much about the specifics of an application. In this case, an all-electric building in Florida called for heat pumps (System #2 for guest rooms and System #4 for most of the other occupancies.) The project site in Florida meant that it is in Climate Zone 2 and did not require an economizer control.

When you combine a heat pump with no economizer and a zone that is dominated by internal loads, such as a ballroom or meeting room, two things happen:

· The DX coil tries to operate during cold weather (There are lots of internal loads that require cooling.)

· Millions of warnings arise (200 million in this case), telling you that your DX coil is producing sub-freezing outlet temperatures. This is guaranteed to win the favor and confidence of a LEED reviewer.

Without thinking this through completely, the first attempt at solving this was to turn off the darn DX coils when the outdoor temperature was below 55F using the EnergyPlus EMS controls. That got rid of most of the warnings … but the unmet load hours went up significantly. After looking at output variables, it became obvious: The DX coil is OFF and there’s no outdoor air cooling assist from the economizer. I solved 200 million warnings, but created another problem which would not win any favor or confidence from a reviewer.

The next attempt added an economizer for zones which were dominated by internal loads. Life was better, but not perfect. I was chafing at the fact that electricity was not being used by the DX coils to provide cooling during cold weather, because it would reduce “savings” compared to the Proposed system, but could not figure out a better alternative.

Lesson #2: EnergyPlus does not like to autosize systems with large outdoor air loads

While autosizing is neither a new idea nor a difficult one, a wrinkle arises for energy models that must comply with ASHRAE Standard 90.1’s Appendix G. Because ASHRAE 90.1 requires outdoor air amounts to be identical for the “Baseline” and “Proposed” energy models, I specified the minimum amount of outdoor air (to match the Proposed systems) instead of autosizing it.

This resulted in several systems with a high outdoor air fraction (20-50% of total flow). With more outdoor air, the coil entering air is hotter and more humid and requires a correspondingly large coil capacity compared to the air flow rate. EnergyPlus imposed its built-in flow / capacity limits and it reduced the coil size for several zones in order to meet the limits. Those down-sized DX coils failed to adequately cool the zones they served … and unmet loads resulted.

I tried an assortment of tactics to convince EnergyPlus to size the coils properly and failed rather miserably. Increasing the sizing factor in Sizing:Zone made some difference, but I was using 175% for some zones in order to get noticeable improvement in unmet loads. That meant much larger fans and a corresponding larger energy use for the Baseline, which seemed like cheating because it would be a false comparison to the Proposed system.

Experimentally, I told the Sizing:System object to consider those systems as “100% outdoor air” and saw some improvement in the coil capacity.

Lesson #3: Multiple Occupancy Start Times Breed Large Unmet Load Hours

I considered all of the occupancy schedules, HVAC system types, temperature and humidity criteria, compass orientation and glazing when defining zones. This facility is approximately 100,000 sq., ft. and one could have justified 50 zones; I used sixteen. The key aspect of those 16 zones was that their fan schedules, considering different days of the week, included nine different starting hours. In many buildings, it’s common and expected that you will see an unmet load hour during the first hour of operation. When the entire building starts at the same time, even if several zones have unmet loads, they all occur at the same time and result in 1 hour of unmet load. Nine different starting times, on the other hand, opens the possibility of an unmet load in each of those hours, for a total of 9 unmet load hours. This multiplication of unmet loads is, I suspect, not something considered by the 90.1 authors.

One solution would be to make all of the schedules have uniform start times, but I had other reasons for wanting to retain many of them (See the next Lesson). My thanks go out to several Bldg-Sim members, who suggested relaxing the OutputControl:ReportingTolerances from 0.2C to something larger. The suggestion was based in the understanding that few HVAC systems are expected to maintain ± 0.2C. More commonly, ± 0.5C is perfectly acceptable and that’s what I used for both heating and cooling. Combined with some sizing factor increase, unmet loads became manageable!

Lesson #4: Hotels are not occupied 24/7

Hotels are 24/7/365 operations in many respects. Their HVAC systems, especially if they are PTAC or PTHP, are not operating in the occupied mode 24/7/365 if the hotel management has any clues about energy conservation. They’re ON when the room is occupied and OFF or in a setback mode when unoccupied. On top of this, more sophisticated owners use a guest room management system which controls HVAC based on room occupancy sensors and the rental status. My building has such a system and I wanted to represent this in the energy comparison.

ASHRAE’s 90.1 User Manual and EnergyPlus’s Schedule dataset for a Hotel show the fans as running 24 / 7 /365. Occupancy, lighting and plug loads are represented by typical sets of diversity fractions. This approach inherently:

· Maintains the entire guest room area at a constant occupied temperature (for the entire year).

· Runs all of the fans constantly.

· Might not account for slow seasons, during which a large fraction of rooms may be un-rented.

· Is very easy to model.

· Misrepresents the actual energy consumption by what seemed an unacceptable amount.

· Does not allow the model to accurately represent additional unoccupied periods made available by a guest room management system.

In order to capture something closer to the actual operation, you could define a lot of zones and create schedules and diversity which reflect the occupancy patterns. This was not an attractive option for reasons of the additional effort to create all of the zones and the additional computation time that I anticipated would also result.

With input from the Bldg-Sim and EnergyPlus listserve communities, I elected to continue using the reduced number of zones (16) and approach schedules in a way that created several discrete occupied periods throughout the day, similar to that shown in the table below.

ASHRAE’s 90.1 User Manual and its recommended schedules are equivalent to about 210 fully occupied days per year. The schedule above has the same number of equivalent days, but has discrete, multi-hour periods where “no one is home”. EnergyPlus treats these as unoccupied, begins cycling the fan instead of running it constantly, and changes the temperature setpoint accordingly. Cooling, heating and fan energy are all reduced with greater realism and life is good.

To model the guest room management system’s ability to do a better job of controlling rooms, this schedule can be easily modified to add more unoccupied hours. Until I have a better idea, I plan to add 10% unoccupied hours as a representation of this improved control, similar to the fraction allowed for lighting occupancy sensors.

Lesson #5: Outdoor swimming pools are different than indoor pools

Modeling indoor pools is a challenge, but at least there is reliable information about how to calculate the major energy factor, evaporation. An outdoor pool’s evaporation rate is subject to constantly changing outdoor temperature and humidity, as well as variable wind speed. None of these is true for an indoor pool and it does not appear that anyone has documented their process for modeling energy consumption of an outdoor pool. Little did I know….

Starting with a spreadsheet which calculates evaporation rates for indoor pools using the Shah method, the TMY weather data, and adding spreadsheet psychrometric functions from NREL, I calculated evaporation for every hour of the year.

Then I created a Schedule:File which shows the evaporation as a fraction of maximum. That Schedule is applied to the flow rate for a WaterHeater with a fixed temperature rise and results in energy for pool heating. At least it results in something defensible; more work is needed.

This approach does not account for the effect of wind, but pools are generally somewhat sheltered and I don’t yet have a method for adding wind’s impact. The pool also loses heat to the ground. I have not addressed that part yet.

Lesson #6: My fellow energy modelers contribute substantial wisdom to this growing industry

… and I am very grateful to them for sharing their wisdom and insights. 1+1 = 10 in this case! This is not a finished work, but if I waited until it was, you’d probably never see it. So, I am posting it in hopes that some of you benefit from my experience. You might also see a better way to approach these problems than I did. Admittedly, it’s an intertwined mess of factors and I may have lost sight of important aspects as I scrambled to figure out what was happening and how to fix it. Comments always welcome!