Delta-Delta-Delta

7 posts / 0 new
Last post

Melissa:
This may come as a shock, but I have had the privilage of runniing DOE2 since the cave man days (1984). Every version (DOE 2.1 A, B, C, D, E, and eQuest 3.55, 3.60 thru 3.64,have ALL had different results !!!!
This isnt one of God's 10 Commandments, its an energy estimation program. I am sure if you compare delta energy usage between windows, the Deltas between the two programs will be similar.
And in the end, who will know you don't have the 100% best glass, as you will have
have built the best buildng, and not the inefficient one.
And I agree on your 3.64 decision.

John Aulbach's picture
Offline
Joined: 2011-09-30
Reputation: 1

John, old buddy,

I also have fond memories of the cave man days about 1982-1984 running Trace uploaded to the mainframe in La Crosse, WI at $150/run. You could sure run up a big bill in a hurry back then. A couple of years later we moved to Ross Merriwether's PC version of ESAS. I cut my teeth with DOE-2 in 1996 with COMPLY-24, which had the first DOS front end to DOE-2. The next year we moved to VisualDOE 2.5. When that reached v3.0, it was a pretty solid and stable program and I used it for many years.

If differing versions of a program produce differing results, perhaps, we should do it like the Olympics.......throw out the high & low scores and average everything else in between. :-)

BTW, I haven't trusted going to 3.65 yet.

Michael R. Busman, CEM

Busman, Michael R's picture
Joined: 2012-07-12
Reputation: 0

John

I do know that 5% error is well within all the other assumptions I have made, and in the end it is a comparative analysis, but I assumed that since both versions of eQuest are driving the same DOE2 engine I expected the same results.

Thanks for historic reference... at least it doesn't take as long to run as it would have on my old "trash"80!
Melissa

Melissa
Melissa Page Crowe's picture
Joined: 2011-09-30
Reputation: 1

Mellisa:
?
What do you think changed in each Version? The DOE2 engine.!!!! Slightly each time. eQuest is just the compiler to make it look pretty in windows. You should have been around when all we had was the .INP file.

John Aulbach's picture
Offline
Joined: 2011-09-30
Reputation: 1

Can't claim I was energy modeling when I was in diapers back then, but I would also expect discrepancies in absolute figures (annual kWh, maximum demand) as you move the same* model between various simulation programs and versions therein. Relatively small changes like a 1.05 factor peak difference if anything would be more of a consoling symptom than a worrisome one - much worse can go awry when trying to move models around like that.

I would focus instead whether the conclusions drawn from your study are different - i.e. what options were you comparing against and does 3.64 suggest different results?

All the best,

~Nick

*of course, claiming "all else being equal" when moving a model between programs/versions always comes with qualifiers - it's a difficult thing to be absolutely certain you haven't missed something.

NICK CATON, P.E.

Nick-Caton's picture
Offline
Joined: 2011-09-30
Reputation: 805

AH! I did miss the DOE2 changes. It now makes sense. Thank you Jeff.

And John, just by-the-by, I have been there done that, last time about 4 years ago I spent 12 months running DOE2 with no interface, text editors and bdl view. So I do have some perspective and am deeply grateful for graphic interface.

Thank you all for the discussion.

Melissa
Melissa Page Crowe's picture
Joined: 2011-09-30
Reputation: 1

Results change for all sorts of reasons, even if there has been no change in the fundamental algorithms. For example, there was a sizable shift from heating to cooling loads going from DOE-2.1B to 2.1C, which I traced to a change in the defaults for neighborhood-terrain-factors that changed the local wind speed, hence higher air-film-coefficients, more radiative heat gain going into the space, etc. If you're puzzled why numbers have changed, I would recommend looking through the documentation on what bug fixes and changes have been noted, and thinking through how they might affect the numbers. I know that could be tedious and frustrating, but I know of no other way, because even the developers might not be aware of all the knock-on effects. I'm not sure averaging results from various versions of the same program gets you the most reliable results. I assume that overall bug fixes are intended to improve the program!

Sent from my IPad

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406