GHE g-function calculation enhancements #8708

j-c-cook · 2021-04-10T00:24:28Z

Pull request overview

A C++ library, named cpgfunctionEP, with the ability to compute
g-functions has been developed by Jack C. Cook, a student of Dr.
Jeffrey D. Spitler at Oklahoma State University.

The cpgfunction library is a low level implementation of Massimo
Cimmino's g-function methodology. The code was developed with the aim
of increasing performance (compared to Cimmino's open source Python
implementation, pygfunction), in both memory and time to be used in
the generation of g-function databases on high performance computing
clusters (HPCC). The resulting program makes it more practical to
compute g-functions for larger ground heat exchangers on a desktop
computer.

This proposal specifically aims to address issue #6651, to provide
EnergyPlus with an enhanced g-function calculation. The library will be
located in the third party folder.

Pull Request Author

Add to this list or remove from it as applicable. This is a simple templated set of guidelines.

Reviewer

This will not be exhaustively relevant to every PR.

A new feature proposal draft for a fast and accurate g-function calculation is included in this commit. A C++ library, named cpgfunction, with the ability to compute g-functions has been developed by Jack C. Cook, a student of Dr. Jeffrey D. Spitler at Oklahoma State University. The cpgfunction library is a low level implementation of Massimo Cimmino's g-function methodology. The code was developed with the aim of increasing performance (compared to Cimmino's open source Python implementation, pygfunction), in both memory and time to be used in the generation of g-function databases on high performance computing clusters (HPCC). The resulting program makes it more practical to compute g-functions for larger ground heat exchangers on a desktop computer. More information about the g-function calculation can be found in the NFP. This proposal specifically aims to address issue NREL#6651, to provide EnergyPlus with a 3rd party g-function calculation.

The NFP mentions proposals for new keys for the GroundHeatExchanger:System. This would result in new lines in IDF and IDD sections. Sample IDF and IDD are presented with proposed changes. It is helpful for the reader to quickly be able to distinguish the new lines. The code blocks in markdown do not allow for highlighting (e.g. bold, underline, etc.). Additionally, html font highlighting did not appear to render in the .md file on github. The changed lines were placed below, and were intended to be highlighted in red (rendering md in atom offline looked fine), but were not. Therefore, the proposed changed lines now begin with a ">" to show distinction.

mitchute · 2021-04-15T18:51:58Z

This NFP is in good shape.

One final comment regarding the NFP: it looks like the gains from the "adaptive integration scheme" are really going to be significant here. You have a plot showing the memory reductions in going from pygfunction to cpgfunction, and then to cpgfunction with the ADS (plot with red arrows). If you have it, it would be nice to show a similar plot for time reductions. This is obviously dependent on the GHE configuration, but an example would be useful. If you don't have it right now, we don't need to hold this up for it.

This is really going to be a nice addition to E+!

FYI: @j-c-cook @jeffreyspitler

jmarrec · 2021-04-20T13:34:28Z

Just a tiny note, note that the diff language marker can be used for your IDD snippets instead of adding a > to mark new lines.

eg:

! IDF example using GHE:Vertical:Array input
GroundHeatExchanger:System,
  Vertical GHE 1x4 Std,    !- Name
  GLHE Inlet,              !- Inlet Node Name
  GLHE Outlet,             !- Outlet Node Name
  0.004,                   !- Design Flow Rate {m3/s}
  Site:GroundTemperature:Undisturbed:KusudaAchenbach,  !- Undisturbed Ground Temperature Model Type
  KATemps,                 !- Undisturbed Ground Temperature Model Name
  2.5,                     !- Ground Thermal Conductivity {W/m-K}
  2.5E+06,                 !- Ground Thermal Heat Capacity {J/m3-K}
  ,                        !- GHE:Vertical:ResponseFactors Object Name
+ UHFcalc,                 !- g-function Calculation Model Name
  GHE-Array;               !- GHE:Vertical:Array Object Name

it also works for modifications

- something old
+ something new

nrel-bot-2c · 2021-05-19T01:27:37Z

@j-c-cook @lgentile it has been 28 days since this pull request was last updated.

…hirdparty

…cook/EnergyPlus into 6651_cpgfunction_thirdparty

…_thirdparty

related to NREL#6651

@jmarrec

As noted by @jmarrec, there is a way to highlight lines that conain - something old + something new NREL#8708 (comment)

For the diff highlighting to work on github, the !'s, +'s and -'s must be at the beginning of a new line.

relates to NREL#6651

On the EnergyPlus call Wednesday, June 30, 2021, a path was laid out for the future of cpgfunction. The project has been forked. There now exists a fork by the name of cpgfunctionEP. cpgfuncitonEP (a fork of cpgfunction) will contain neither Fortran nor Boost dependencies.

…cook/EnergyPlus into 6651_cpgfunction_thirdparty

mitchute · 2021-08-18T19:42:29Z

I've run this branch on two different Linux boxes, in release and debug modes. It looks like the segfaults are now gone. @Myoldmopar over to you for a final review and merge.

…rgyPlus into 6651_cpgfunction_thirdparty

…_thirdparty

Myoldmopar · 2021-08-18T19:49:37Z

Building now, thanks @mitchute !

Myoldmopar · 2021-08-18T20:34:24Z

Concur, everything is passing now. I'll be happy to resolve the conflict and do final review here.

Myoldmopar

This looks really close. I noted a few minor tweaks that I'm going to make in a follow-up commit. This also needs a minor transition rule which @mitchute is doing now. Otherwise this should be ready to go in.

Myoldmopar · 2021-08-19T13:39:04Z

src/EnergyPlus/CommandLineInterface.cc

+        if (state.dataGlobal->numThread == 0) {
+            DisplayString(state, "Invalid value for -j arg. Defaulting to 1.");
+            state.dataGlobal->numThread = 1;
+        } else if (state.dataGlobal->numThread > (int)std::thread::hardware_concurrency()) {


Interesting. I'm not sure we really need to take the burden of restricting this. And on some platforms where this cannot be determined, it is required to return 0, thus making EnergyPlus fail. I think I'll take this out.

Myoldmopar · 2021-08-19T13:39:30Z

src/EnergyPlus/CommandLineInterface.cc

@@ -149,6 +152,8 @@ namespace CommandLineInterface {

        opt.add("", false, 0, 0, "Display version information", "-v", "--version");

+        opt.add("1", false, 1, 0, "Multi-thread with N threads; 1 thread with no arg.", "-j", "--jobs");


I'm adding a little phrase in here to describe the GHE usage so it shows up in the usage() display.

…rgyPlus into 6651_cpgfunction_thirdparty

Myoldmopar · 2021-08-19T16:27:58Z

Thanks for the transition update @mitchute. I took a GSHP file that did not have the full list of fields and it left the object alone. I took GSHP-GLHE-CalcGFunctions.idf and transitioned it to find the new default field name put in there. It's working as expected. I have a few small tweaks and will do final testing, but I anticipate this is good to go in next/shortly.

…_thirdparty

Myoldmopar · 2021-08-19T17:54:33Z

Alright, this is looking great. I pulled in develop one more time and did a full build and test. Unit and integration tests are passing fine, and there are no unexpected regressions. This is good to go in. I'll push my small commit up and then get it merged. Thanks @j-c-cook and @mitchute !

Myoldmopar · 2021-08-19T18:04:34Z

Forgot about the Windows warnings, so addressed them real quick. Heads up @j-c-cook these changes were in the new third party code.

Myoldmopar · 2021-08-19T18:06:36Z

third_party/cpgfunctionEP-0.5/src/segments.cpp

@@ -49,7 +49,7 @@ namespace gt::segments {
            nq = jcc::interpolation::interp1d(drilling_depth,
                                              drilling_depths[0],
                                              ideal_segment_lengths[0]);
-        } else if(heights[0] < height < heights[n1]) {


In C++, x < y < z doesn't do what you think it does. C++ does not have chained operation. This will perform one binary comparison, which in this case returns a boolean, then try to compare that boolean to a floating point with the second binary operation. I'm assuming you just want to do x < y && y < z, so that's what I did here. This nifty syntax works in Python but not C++.

Myoldmopar · 2021-08-19T18:06:42Z

third_party/cpgfunctionEP-0.5/third_party/LU-Decomposition-0.1/src/lu.cpp

@@ -9,7 +9,7 @@ void jcc::decomposition(vector<vector<double> > &A, int &n, vector<int> &indx,
                   double &d) {
    const double TINY = 1.0e-20;  // a small number
    int i, imax, j, k;
-    double big, dum, sum, temp;
+    double big, dum, sum;


Unused variable removd.

Myoldmopar · 2021-08-19T18:07:36Z

third_party/cpgfunctionEP-0.5/third_party/qdt-master/methods/adaptive.h

@@ -376,7 +376,7 @@ class Adaptive<BaseMethod,Estimator,true,true,false,true> :
 template<typename BM>
 Adaptive<BM> adaptive(const BM& bm)
 {
-	return Adaptive<BM>(bm, 1.e-5, 1.e-10);
+	return Adaptive<BM>(bm, 1.e-5f, 1.e-10f);


Explicitly creating float literals instead of double to avoid the warning.

Myoldmopar · 2021-08-19T18:07:58Z

Alright, I'm merging this now. Thanks all.

rraustad · 2021-08-20T20:17:37Z

@j-c-cook why do I have to add #include <algorithm> to boreholes.cpp and adaptive.h to get MS Visual Studio Professional 2019 Version 16.5.0 to compile? In boreholes.cpp was a max statement where MSVS said identifier max not found and in adaptive.h is a std::max and MSVS says max is not a member. When I type in std::m I don't see max in the pop up list as an available function. I also don't understand why adding #include <algorithm> in adaptive.h corrects a statement of std::max ?? since I did not change that line from std::max to max.

Do you have any insight into this?

rraustad · 2021-08-20T20:52:18Z

In develop boreholes.cpp I see this in MSVS.

j-c-cook · 2021-08-20T20:53:47Z

I cannot say anything definitive about how MS Visual Studio Professional 2019 Version 16.5.0 operates. I do nearly all development in Kubuntu 20.04.2, and compiled the code in Windows 10 only when the CI machine returned errors. I was under the impression that all errors in a Windows build or test would have been caught by the CI machines, and that the code would not have been merged before those errors were found. However, unforeseen bugs happen.

To eradicate this bug will I need to download a Visual Studio IDE specifically, or does compiling from the command prompt provide the same error?

rraustad · 2021-08-20T20:58:38Z

I can just commit this small change as it does not appear to change anything other than which headers are included. I just wanted to ask before I made that commit. It's only 2 lines of code, 1 in each file, and only a #include. Why I see this compiler warning and others who use MSVS do not is a mystery to me (other than they haven't pulled develop yet).

j-c-cook · 2021-08-20T21:01:53Z

Okay thank you! I'll make sure those changes propagate their way back into cpgfunctionEP's repo, along with some refactors Edwin had made as well.

j_c_cook added 2 commits April 9, 2021 18:40

j-c-cook force-pushed the 6651_cpgfunction_thirdparty branch from c6063ec to e042488 Compare April 10, 2021 01:34

mitchute added the NewFeature Includes code to add a new feature to EnergyPlus label Apr 10, 2021

mitchute added this to the EnergyPlus Future milestone Apr 10, 2021

Merge branch 'develop' into 6651_cpgfunction_thirdparty

a80388a

j_c_cook added 21 commits June 8, 2021 19:40

Merge remote-tracking branch 'origin/develop' into 6651_cpgfunction_t…

9ef0dc9

…hirdparty

Merge branch '6651_cpgfunction_thirdparty' of https://github.com/j-c-…

327bd8b

…cook/EnergyPlus into 6651_cpgfunction_thirdparty

Merge remote-tracking branch 'upstream/develop' into 6651_cpgfunction…

6ca1152

…_thirdparty

Append design documentation to NFP

49df1a0

related to NREL#6651

Document IDD snippets with the diff language marker

30160c1

As noted by @jmarrec, there is a way to highlight lines that conain - something old + something new NREL#8708 (comment)

Add diff as markdown language for IDD snippets

72b63cd

Refactor diff Documentation placements

69b797d

For the diff highlighting to work on github, the !'s, +'s and -'s must be at the beginning of a new line.

Document the design document in regards to dependencies

9b082d9

relates to NREL#6651

Update design document to reflect current outlook

40af248

Refactor design document to only include "bottom line up front"

8c9f71a

Add in links to design document

063c86c

Merge branch 'develop' into 6651_cpgfunction_thirdparty

afa9878

Add cpgfunctionEP-0.1 to the third_party folder

a795a7e

Add cpgfunctionEP subdirectory to E+ CMakeLists for third party

a6779b7

Comment out cpgfunctionEP tests temporarily

772f055

Modify GHE.unit.cc GHE:System object for new field

58e74a3

Add copy of GSHPSimgple-GLHE.idf file

28257f8

Modify GLHE idf file (copy) to call UBHWT calc

f5e1534

Add GHE:CalculationModel name paragraph to IO tex

0537d0d

Add gFunctionCalcModelName field to IDF listings

d958866

j-c-cook and others added 3 commits August 17, 2021 18:00

Make EP's cmake point to cpgfunction-v0.5

6440e9d

Merge branch '6651_cpgfunction_thirdparty' of https://github.com/j-c-…

f5bfa7f

…cook/EnergyPlus into 6651_cpgfunction_thirdparty

Merge branch 'develop' into 6651_cpgfunction_thirdparty

376c7f9

Myoldmopar added 2 commits August 18, 2021 14:49

Merge branch '6651_cpgfunction_thirdparty' of github.com:j-c-cook/Ene…

aa6b665

…rgyPlus into 6651_cpgfunction_thirdparty

Merge remote-tracking branch 'upstream/develop' into 6651_cpgfunction…

05668db

…_thirdparty

Myoldmopar and others added 2 commits August 18, 2021 15:52

Add clarification in cli usage

22f3445

Merge branch 'develop' into 6651_cpgfunction_thirdparty

4d0e9a2

Myoldmopar reviewed Aug 19, 2021

View reviewed changes

mitchute and others added 2 commits August 19, 2021 09:31

add ghe transition rules

b7ff75b

Merge branch '6651_cpgfunction_thirdparty' of github.com:j-c-cook/Ene…

fed355b

…rgyPlus into 6651_cpgfunction_thirdparty

Merge remote-tracking branch 'upstream/develop' into 6651_cpgfunction…

f8caacd

…_thirdparty

Address Windows build warnings

4c9f832

Myoldmopar reviewed Aug 19, 2021

View reviewed changes

Myoldmopar merged commit 2632627 into NREL:develop Aug 19, 2021

Myoldmopar deleted the 6651_cpgfunction_thirdparty branch August 19, 2021 18:08

rraustad mentioned this pull request Aug 20, 2021

Address MSVS compile issue with recent commit. #8991

Merged

20 tasks

j-c-cook mentioned this pull request Dec 1, 2021

Enhancement Increase performance of LU decomposition j-c-cook/cpgfunctionEP#57

Open

j-c-cook mentioned this pull request May 13, 2022

cpgfunction cleanup #9307

Closed

20 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

GHE g-function calculation enhancements #8708

GHE g-function calculation enhancements #8708

j-c-cook commented Apr 10, 2021 •

edited

Loading

mitchute commented Apr 15, 2021

jmarrec commented Apr 20, 2021 •

edited

Loading

nrel-bot-2c commented May 19, 2021

mitchute commented Aug 18, 2021

Myoldmopar commented Aug 18, 2021

Myoldmopar commented Aug 18, 2021

Myoldmopar left a comment

Myoldmopar Aug 19, 2021

Myoldmopar Aug 19, 2021

Myoldmopar commented Aug 19, 2021

Myoldmopar commented Aug 19, 2021

Myoldmopar commented Aug 19, 2021

Myoldmopar Aug 19, 2021

Myoldmopar Aug 19, 2021

Myoldmopar Aug 19, 2021

Myoldmopar commented Aug 19, 2021

rraustad commented Aug 20, 2021 •

edited

Loading

rraustad commented Aug 20, 2021 •

edited

Loading

j-c-cook commented Aug 20, 2021 •

edited

Loading

rraustad commented Aug 20, 2021

j-c-cook commented Aug 20, 2021

		@@ -149,6 +152,8 @@ namespace CommandLineInterface {

		opt.add("", false, 0, 0, "Display version information", "-v", "--version");

		opt.add("1", false, 1, 0, "Multi-thread with N threads; 1 thread with no arg.", "-j", "--jobs");

GHE g-function calculation enhancements #8708

GHE g-function calculation enhancements #8708

Conversation

j-c-cook commented Apr 10, 2021 • edited Loading

Pull request overview

Pull Request Author

Reviewer

mitchute commented Apr 15, 2021

jmarrec commented Apr 20, 2021 • edited Loading

nrel-bot-2c commented May 19, 2021

mitchute commented Aug 18, 2021

Myoldmopar commented Aug 18, 2021

Myoldmopar commented Aug 18, 2021

Myoldmopar left a comment

Choose a reason for hiding this comment

Myoldmopar Aug 19, 2021

Choose a reason for hiding this comment

Myoldmopar Aug 19, 2021

Choose a reason for hiding this comment

Myoldmopar commented Aug 19, 2021

Myoldmopar commented Aug 19, 2021

Myoldmopar commented Aug 19, 2021

Myoldmopar Aug 19, 2021

Choose a reason for hiding this comment

Myoldmopar Aug 19, 2021

Choose a reason for hiding this comment

Myoldmopar Aug 19, 2021

Choose a reason for hiding this comment

Myoldmopar commented Aug 19, 2021

rraustad commented Aug 20, 2021 • edited Loading

rraustad commented Aug 20, 2021 • edited Loading

j-c-cook commented Aug 20, 2021 • edited Loading

rraustad commented Aug 20, 2021

j-c-cook commented Aug 20, 2021

j-c-cook commented Apr 10, 2021 •

edited

Loading

jmarrec commented Apr 20, 2021 •

edited

Loading

rraustad commented Aug 20, 2021 •

edited

Loading

rraustad commented Aug 20, 2021 •

edited

Loading

j-c-cook commented Aug 20, 2021 •

edited

Loading