Issue33 commong function interface #84

MassimoCimmino · 2021-04-08T21:39:14Z

This closes #33, closes #54, closes #85.

I am opening this PR to get some feedback and review on the new interface (#33). It is mostly complete. The remaining tasks are to

complete/modify the documentation.
use the find_duplicates function from Functions to check duplicates and remove duplicate boreholes #81 to check the validy of a bore field layout.
update CHANGELOG.md.

Some highlights:

I removed functions from pygfunction.gfunction, namely load_history_reconstruction, _borehole_segments, _temporal_superposition as they are now part of the new classes.
I modified the existing examples to use the new interface.
There are new visualization features in the gFunction class. These are demonstrated in the modified examples, e.g. uniform_temperature.py.

MassimoCimmino · 2021-04-17T01:33:10Z

The PR is ready. I squashed commits and rebased the history (trying out for #83).

…terface

This introduces nominal conditions (m_flow ans c_p) to the Network class. These are used at initialization to initialize the values of the Network coefficients. This is needed to simplify the interface to the new gFunction class, so that m_flow and c_p are not needed to evaluate the g-function.

The g-function template now calls the existing functions. Future changes will now be able to be tested.

The new Solver classes will be called by the gFunction class to calculate the gFunction. This will facilitate the implementation of new calculation methods as they are developed in the literature. The _BaseSolver class implements the `solve` function, which is aimed to be applicable independent of the solution method. Method-specific tasks can be implemented in their own classes.

The gFunction class can now plot the borehole temperatures and heat extraction rates (both as a function of time and the profiles over the length). These are often shown in publications and are useful to interpret the result of g-function calculations. In the past, modified versions of pygfunction were used to output these for publications. The gFunction class and the new visualization methods add this functionality natively while limiting the memory use through the `profiles` options.

Make sure the provided inputs are valid and automatically choose the boundary condition depending on the provided `boreholes_or_network`, if not specified by the user (this is simply for conveninence). This also makes use of the `find_duplicates()` function introduced in #81 to make sure the bore field is valid and the g-function can be properly calculated.

MassimoCimmino · 2021-04-22T23:51:39Z

@j-c-cook Could you take a look at this PR? It is pretty much complete from my side.

j-c-cook · 2021-04-27T16:54:23Z

@MassimoCimmino I haven't forgot about this. We are finishing up the semester down here and its a wild time. I will get around to this as soon as practical.

If at some point you feel the need to merge it and I haven't gotten around to looking closely, please feel free to merge.

j-c-cook · 2021-05-13T02:33:57Z

I was studying your equation 18 from Cimmino (2018).

In master you pass dh_ij to the _temporal_superposition. However, in this branch you are passing h_ij and computing a dQ to find the dimensionless borehole wall temperature.

Did you decide to do this because you thought it was less work on the CPU? Does this provide both a speed increase and a reduction in memory? I remember you saying somewhere that you're getting the right results with this new code, so that leads me to believe this formulation works. Why is it that equation (18) has Q_reconstructed(Tau_k^prime), but the code makes use of Q[:, nt-it-1]?

Edit: It’s been a while since I looked at this, and now I see equation 37, which explains what is in master.

Massimo Cimmino (2018) Fast calculation of the g-functions of geothermal borehole fields using similarities in the evaluation of the finite line source solution, Journal of Building Performance Simulation, 11:6, 655-668.

MassimoCimmino · 2021-05-13T12:34:53Z

The change is motivated by #85. In the PR, dh_ij[:nSources, :nSource, :nt] is no longer computed and instead we only have h_ij[:nSources, :nSources, :(nt+1)] and h_dt[:nSources, :nSources]. We previously had all 3 plus the stored array in the interp1d object.

The two formulations of the temporal superposition process are equivalent : dT = convolution(dh, Q) = convolution(h, dQ). I don't see it affecting the CPU time in any significant way but made no attempt to quantify this. The implementation of _temporal_superposition() is probably not optimal (it creates a new array of size nSources x nt every time step). This is something that could be revisited after #78. The two remaining computational bottlenecks in that branch (for large fields) are the construction of h_ij from the already computed FLS and the solution of the system of equations (which the optimization of _temporal_superposition() may help).

On another note, I noticed there are now unused functions in the heat_transfer module : thermal_response_factors, similarities, _similarities_group_by_distance, and _similarities_one_distance. These were all transferred to the Solver classes. Unless you have any use for them, I propose they be removed.

Existing functions now call the gFunc class instead of having their separate implementation. Existing examples are changed to use the new gFunction class. Examples are also extended to demonstrate the new visualization features.

This is for #85

Solver classes are not meant to be directly accessible to the user. They also made the documentation very hard to read. Moved to end of file.

The heat extraction rates and borehole wall tempratures only need to be stored after the g-function is evaluated.

MassimoCimmino marked this pull request as draft April 8, 2021 21:40

MassimoCimmino marked this pull request as ready for review April 8, 2021 21:41

MassimoCimmino added this to the v2.0.0 milestone Apr 8, 2021

MassimoCimmino marked this pull request as draft April 8, 2021 21:42

j-c-cook mentioned this pull request Apr 10, 2021

Reduce segment responses held in memory #85

Closed

MassimoCimmino force-pushed the issue33_CommongFunctionInterface branch from 845d7cf to 7894902 Compare April 17, 2021 01:23

MassimoCimmino marked this pull request as ready for review April 17, 2021 01:33

j_c_cook and others added 9 commits April 16, 2021 21:48

JCC initial recomendation for g-function interface

890def6

Remove example, the existing examples were modified to use the new in…

ee338c4

…terface

Fix gFunction template

07ff805

The g-function template now calls the existing functions. Future changes will now be able to be tested.

Display total calculation time for g-function

5b65f7d

Partial formatting of gFunction class

cd59974

MassimoCimmino force-pushed the issue33_CommongFunctionInterface branch from 502f64a to aad279c Compare April 17, 2021 01:49

MassimoCimmino mentioned this pull request Apr 26, 2021

Issue78 optimize integrals #90

Merged

MassimoCimmino added 7 commits May 15, 2021 17:01

Update existing functions and examples

40c5294

Existing functions now call the gFunc class instead of having their separate implementation. Existing examples are changed to use the new gFunction class. Examples are also extended to demonstrate the new visualization features.

Add docstrings

a2f232a

Change default value of options to an empty dict

b224299

Check for duplicate boreholes at initialization of gFunction

f47a710

Add dtype option to gFunction solvers

40d61cf

This is for #85

Fix docstrings for references, formatting

4e025ab

Format messages for disp True

40029d7

MassimoCimmino added 10 commits May 15, 2021 17:01

Remove Solver classes from public interface

5d98bd0

Solver classes are not meant to be directly accessible to the user. They also made the documentation very hard to read. Moved to end of file.

Reformat

00775cd

Update CHANGELOG.md

29099a6

Update README.md example

8d2a59d

Reformat examples

d5b165c

Reformat docstrings

746250d

Update CHANGELOD.md

569ff18

Add missing time assignement in solve

95b9bdc

Fix indentation in _BaseSolver.solve

9e16524

The heat extraction rates and borehole wall tempratures only need to be stored after the g-function is evaluated.

Deprecate unused functions in heat_transfer

916c658

MassimoCimmino force-pushed the issue33_CommongFunctionInterface branch from 08975c0 to 916c658 Compare May 15, 2021 21:02

MassimoCimmino added 2 commits May 17, 2021 16:36

Allow float input for time

2851f8d

Update unit tests

6d46830

MassimoCimmino merged commit 558e367 into master May 17, 2021

MassimoCimmino deleted the issue33_CommongFunctionInterface branch November 12, 2021 21:03

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Issue33 commong function interface #84

Issue33 commong function interface #84

MassimoCimmino commented Apr 8, 2021 •

edited

Loading

MassimoCimmino commented Apr 17, 2021

MassimoCimmino commented Apr 22, 2021

j-c-cook commented Apr 27, 2021

j-c-cook commented May 13, 2021 •

edited

Loading

MassimoCimmino commented May 13, 2021 •

edited

Loading

Issue33 commong function interface #84

Issue33 commong function interface #84

Conversation

MassimoCimmino commented Apr 8, 2021 • edited Loading

MassimoCimmino commented Apr 17, 2021

MassimoCimmino commented Apr 22, 2021

j-c-cook commented Apr 27, 2021

j-c-cook commented May 13, 2021 • edited Loading

MassimoCimmino commented May 13, 2021 • edited Loading

MassimoCimmino commented Apr 8, 2021 •

edited

Loading

j-c-cook commented May 13, 2021 •

edited

Loading

MassimoCimmino commented May 13, 2021 •

edited

Loading