[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [EnergyPlus_Support] Re: Model too big and too complicated? Part 1 - PC Performances





Monsieur Beaulieu,

I maintain a document for reminding myself of important aspects of modeling and also for use when I train others, so I have included portions of it for you.  See if any of these apply to you:

Windows and door layout

·         Group all windows on individual walls into a single window of equivalent size unless daylighting will be part of the simulation

o   Maintain windows within the correct zone

o   Position windows at a representative position on the wall

o   Position windows so that shading elements affect the window appropriately

·         Model doors only when they contain glass windows

o   This assumes that doors often have similar insulation value to the wall and are normally an insignificant portion of total wall area.

o   When doors represent a significant area and differ greatly from wall insulation, place them similarly to windows (in a group by zone)

Zone layout

·         Use as few zones as possible to speed simulation time and debugging.  DesignBuilder’s “Merge Zones” feature is very helpful for doing this.

·         Normally, a zone should not represent each room or VAV box.  Combine into one zone all areas which have:

Ø  Similar schedule

Ø  Similar densities (people, lighting and plug loads)

Ø  Use the same temperature and humidity setpoints

Ø  Have the same sun exposure

Ø  Are served by the same HVAC air system

o   Note that you may combine rooms that use different thermostats as long as their temperature setpoints are the same.

o   Separate HVAC systems (e.g., packaged rooftop units) may be combined if the efficiencies and operating controls are the same.

·         Areas on different floors may become part of a single zone if they meet the criteria above

·         Perimeter areas (within ~12-15 feet of the exterior wall) are normally treated separately from interior / core areas

o   If the perimeter is well insulated and has no windows, it may be combined with interior / core areas

Materials

·         Use ASHRAE 90.1 for wall and window materials and properties in baseline models

·         Use actual materials for Proposed and existing building models from Architectural drawings and specifications.  You need not model every small detail of construction.  Certain wall constructions are very similar to each other and can probably be combined as a single type.

Schedules and Loads

·         Do not attempt to define separate schedules for zones which are only slightly different.  An identical schedule is often “close enough” for the purpose of an energy model, especially a new building which does not have actual experience of operation.

AirLoop Systems

·         Multiple small systems of essentially identical operation and efficiency can be combined.  This might be small rooftop HVAC units or small exhaust fans..

Running and Debugging (with thanks to M.J. Witte, GARD Analytics, Inc.)

  • Set the warmup days to >= 25 (BUILDING object).

·         Run the simulation initially for DesignDays only. All normal models and modelers have many little problems that can be resolved without running the model for the entire weather year (and wasting a bunch of time!):

o   You may want to add a couple intermediate weather DesignDay to the normal Summer and Winter DesignDays after getting the simulation working well for Summer and Winter.  This is a day in which cooling and heating may occur on the same day and also one where your economizer will experience the trip condition (from full outdoor air to minimum)

o   It’s helpful to include full operation on DesignDays for all scheduled equipment items.  Energy Plus treats DesignDays as completely separate from “real” schedules.  In order to see if the systems are working properly during the DesignDay runs, you’ll want to make sure everything is ON.

o   You might also find that a large or complex model takes 20 minutes or longer to run just DesignDays!  If that’s the case, consider making a much smaller “dummy” file with enough zones for testing.  This smaller file will run a lot faster and help debugging progress more quickly.

§  The “dummy” file must follow all of the normal requirements, including correct zone names.

§  Certain multi-zone systems will need “doctoring “ to get the zone / system air balance in line

§  If you are autosizing anything, you may find that, for example, the autosized supply air flow is less than the specified exhaust air flow, causing a system to run at 100% OA continuously.  (Supply air flow is small because your “dummy” file represents a smaller building with less load.)

  • Use .rvi files to keep variables organized and use plenty of variables so that you can represent every system and its performance in detail.

·         Start debugging

o   Use the .err file which is created during each run to identify the source of errors.

o   Review the .svg diagram after getting the model to run without fatal errors using Visio or another .svg file viewer.

o   When debugging, focus on errors which are noted BEFORE a “Fatal” error.  Everything past that hasn’t been fully screened and is likely to be bogus. Although they may give a clue why the fatal error occurred, my experience is that I should NEVER pay any attention to errors listed after a fatal error; they’re meaningless.

o   Pay attention to the Warnings, not just “fatal” and “severe” errors.  These are occasionally indicative of a problem that, while not preventing completion of the simulation, should be corrected to assure accuracy.

o   In rare cases, you may want to turn off the HVAC system by setting availability manager schedules to always be OFF.  This may allow a simulation to complete, but with warnings for further diagnosis.

Testing

  • Once a file has executed without crashing, severe errors or problematic Warnings, grab a friend and give him / her a big hug and “high five”.
  • Now double check that report variables (including node temps and node setpoint values) are working properly and review the data to see if each loop is performing correctly.

·         Visually review the CSV (Report Variable) data for believability and conformance with your (intended) input.

o   Check temperatures, setpoints, flow, outdoor air and humidity as appropriate.

o   Check peak summer, peak winter and intermediate periods. 

o   Pay attention to proper operation in occupied and unoccupied periods.

o   Make sure equipment is OFF when it’s supposed to be OFF during unoccupied periods, Summer or Winter as appropriate.

Past Problems and their solution

  • I failed to include the zone exhaust node in CONTROLLED ZONE EQUIP CONFIGURATION for four exhaust fans. 169,000 warnings about “maximum  iterations exceeded” and 2.5 hours later, the simulation concluded.
  • The room temperature for a heat pump system did not meet the setpoint, even though the DX coil capacity was much greater than was being used.  This was due also using the new CoolReheat feature in v7 and the code was not dialing capacity in as closely as it could have.  I expanded the reporting tolerance for unmet loads to 0.5C instead of 0.2C.
  • Currently (2/08), not all equipment types register plant flow requirements for autosizing. In general if the water flow rate is not autosizeable (as is the case for COIL:WaterToAirHP:) then the plant loop probably does not see that flow rate when autosizing the loop flow. Check the autosized plant loop flow rate in the eio output file to see if it seems reasonable. I suspect you will need to use hard sizes for the plant loop flow rates.
  • In general, it's easiest if the individual components are always available and the thermostat decides whether to cool or heat.
  • For night cycle control, it might be easier to use the fan mode schedule to switch between continuous and cycling fan and let the system be "available" 24/7 instead of using the night cycle availability manager.
  • Minimum flows for some controllers don’t always work; better to set the min = 0.
  • “node did not find an appropriate matching "outlet" node”  Every node that is used as an inlet node must be connected to the outlet node of some other object, or to an outside air node. There can be no "dangling" inlet nodes. On the air side, this means that any node name used as an inlet in one component must appear again in the .idf file as an outlet or as an outside air node. On the plant side, because mixers and splitter are used to connect branches, this kind of error usually means an entire branch is dangling and is not connected to a mixer or splitter. These errors usually occur because of a typing error in the node name in one component or another, or because a connection has been left out.

·         'Flow per Exterior Surface Area' infiltration option includes surface areas for the roof and floors if they are exposed to "external environment" or "othersideconditionsmodel".  Ground contact surfaces are not included.

·         For “Maximum iterations exceeded” warnings, keep in mind:

o   Most problems occur at the interface between the plant side and demand side of loops.  They can also occur at startup times (for Night Cycle control as well as normal Schedule change from OFF to ON.)

o   Use the “Diagnostics, DisplayExtraWarnings” command to get more detail and specifically to see which variables fail to converge, and at what time period.

o   Also use Report Variables “HVACManage Iterations” and “SimAir Iterations”.  These identify the number of iterations at each timestep, which allows you to see what else is going on at that time.  If all setpoints are met, everything should be OK.

o   If you are reporting at the “timestep” frequency, then it is expected that the number of iterations can be larger than the specified maximum.  EnergyPlus has two different modeling timesteps, zone and system.  The HVACManage Iterations report variable and the Maximum Iterations setting are at the system timestep level.  If you are reporting HVACManage Iterations at the timestep frequency, then that is the zone timestep and the reported value is the sum of all the system iterations for all the system timesteps within that zone timestep.  Report out at “detailed” frequency if you really need to see the results at the system timestep level.

o   The tolerance flags are not necessarily intended for users, they are for developers and haven’t been documented for use by users.

·         “Interzone surfaces are occurring in the same zone(s)”    This is a warning to tell you that you have interzone (connecting)  surfaces in the same zone. Normally you would have different zones  on the side of each surface but you might use these (such as linking  a ceiling to a floor) for a "middle zone" in a large building. It's  a warning to make sure you know what you are doing as you use these.  If you put in the Diagnostics, DisplayExtraWarnings; statement you will see exactly what surfaces it is complaining about.

 

So far, my longest-running model is about 2 hours for a complete weather year.  That is borderline intolerable, but I have not (yet) attempted other methods to find make E+ run faster during the preparation and debugging phase.  I recall receiving some tips from one of the Help Desk staff a few months ago, but could not find them. From memory (and not mentioned above):

·         Set Building, Solar Distribution to “MinimalShadowing”

·         Set ShadowCalculation, Calculation Frequency to a larger value than default (say, 30 days)

·         Set Timestep to 2  (maybe even 1.  Accuracy is reduced, but this should help for speed until you are confident that all problems are resolved.)

·         Set ConvergenceLimits, Minimum System Timestep to a larger value (say, 15 minutes.  Same caution as for timestep above)

Caution: Since I have not used these much, I may remember incorrectly!  Let me know how they work.

 

Some of the ideas above just make the model preparation faster for you.  Some help it run faster.  I don’t have a good idea of the magnitude of savings for each of these and there are probably other methods, known by others in this forum.

 

The Building Performance Team
James V. Dirkes II, P.E., BEMP , LEED AP
1631 Acacia Drive NW
Grand Rapids, MI 49504
616 450 8653

 

From: EnergyPlus_Support@xxxxxxxxxxxxxxx [mailto:EnergyPlus_Support@xxxxxxxxxxxxxxx] On Behalf Of jipp06
Sent: Wednesday, December 07, 2011 2:05 PM
To: EnergyPlus_Support@xxxxxxxxxxxxxxx
Subject: [EnergyPlus_Support] Re: Model too big and too complicated? Part 1 - PC Performances

 

 

For the record I figured I'd give some feedback on the outcome of this problem I had then. For those who read this in the future.

I have had the model CRASH on me, because it was lacking hard disk space. My output files were quite big for many reasons and they filed the laptop hdd. :-)

read this post for more useful info:
"Increasing the performance of the computer"
http://tech.groups.yahoo.com/group/EnergyPlus_Support/message/22939

cheers
J-Philippe Beaulieu, ing.

--- In EnergyPlus_Support@xxxxxxxxxxxxxxx, "jipp06" <jippnojunk@...> wrote:
>
> Hello,
>
> In the beginning, it's easy to add details that will make your model more complex later in the modeling process. I think I realised that. And it's kind of different for every software.
>
> I put too many zones in my building... it's easy to do so in Openstudio. Don't you think?
>
> Down the road, I think I am hitting the maximum performances of my PC.
>
> 1st set of questions:
> #1 - If a model only occasionnaly CRASH, is it a good indication that the PC is not powerfull enough? last night a model crashed, I rebooted the PC to free RAM and it went through this time.
> MAYBE that the output file are too big: 1Gb .SQL file / 1,6Gb .ESO file... my PC has 2Gb RAM. I have a feeling PC may crash if I don't have at least that space in RAM, no?
>
> #2 - Can I trust the results of that successfull run (if it has crashed in the past). I suspect that if a run goes through without crashing, it probably means the results are good, right?
>
> AND WHAT ABOUT YOU?
> #3 - What kind of PCs are you running. I consider upgrading. I have an Intel Core2 Duo CPU 2,2 GHz with 2Go RAM running on WIN Vista Ultimate.
>
> #4 - Should I go for more CPUs or more RAM or Both... I understand from the old msgs on this forum that E+ is single core but there are plans to make it multi-core, right? I should upgrade to 4CPU's then. I think I'll go for a Intel Core i7-2600 Processor (3.40GHz, 1066MHz, 8MB) with 4Gb RAM
>
> I have a second set of questions regarding my "complexe" model but I think I'll post them seperately in another msg. coming next!
>



__._,_.___


Primary EnergyPlus support is found at:
http://energyplus.helpserve.com or send a message to energyplus-support@xxxxxxxx

The primary EnergyPlus web site is found at:
http://www.energyplus.gov

The group web site is:
http://groups.yahoo.com/group/EnergyPlus_Support/

Attachments are currently allowed but be mindful that not everyone has a high speed connection.  Limit attachments to small files.

EnergyPlus Documentation is searchable.  Open EPlusMainMenu.pdf under the Documentation link and press the "search" button.




Your email settings: Individual Email|Traditional
Change settings via the Web (Yahoo! ID required)
Change settings via email: Switch delivery to Daily Digest | Switch to Fully Featured
Visit Your Group | Yahoo! Groups Terms of Use | Unsubscribe

__,_._,___