NOCT cell temperature function #1177

cwhanse · 2021-02-25T22:32:29Z

~~[ ] Closes #xxxx~~
I am familiar with the contributing guidelines
Tests added
Updates entries to docs/sphinx/source/api.rst for API changes.
Adds description and name entries in the appropriate "what's new" file in docs/sphinx/source/whatsnew for all changes. Includes link to the GitHub Issue with :issue:`num` or this Pull Request with :pull:`num`. Includes contributor name and/or GitHub username (link with :ghuser:`user`).
New code is fully documented. Includes numpydoc compliant docstrings, examples, and comments where necessary.
Pull request is nearly complete and ready for detailed review.
Maintainer: Appropriate GitHub Labels and Milestone are assigned to the Pull Request and linked Issue.

Implement cell temperature function that is the default in SAM when the CEC module model is chosen.

cwhanse · 2021-03-01T20:47:07Z

Ready for review. check failure is codecov, but all of temperature.py is covered.

wholmgren

Mostly looks good. Are you planning a follow up issue/pr for integration into PVSystem and ModelChain?

wholmgren · 2021-03-01T20:50:48Z

pvlib/tests/test_temperature.py

+    expected = 54.41542289
+    result = temperature.noct(poa_global, temp_air, wind_speed, noct,
+                              eta_m_ref)
+    assert np.isclose(result, expected)


Suggested change

assert np.isclose(result, expected)

assert_allclose(result, expected)

wholmgren · 2021-03-01T20:50:57Z

pvlib/tests/test_temperature.py

+    result = temperature.noct(np.array(poa_global), np.array(temp_air),
+                              np.array(wind_speed), np.array(noct),
+                              np.array(eta_m_ref))
+    assert np.isclose(result, expected)


Suggested change

assert np.isclose(result, expected)

assert_allclose(result, expected)

wholmgren · 2021-03-01T20:51:17Z

pvlib/tests/test_temperature.py

+                              transmittance_absorbtance, array_height,
+                              mount_standoff)
+    expected = 58.36654459
+    assert np.isclose(result, expected)


Suggested change

assert np.isclose(result, expected)

assert_allclose(result, expected)

wholmgren · 2021-03-01T20:52:44Z

pvlib/temperature.py

+    '''
+    Cell temperature model from the System Advisor Model (SAM).
+
+    The model is described in [1], Section 10.6.


Suggested change

The model is described in [1], Section 10.6.

The model is described in [1]_, Section 10.6.

SAM document says "The NOCT photovoltaic cell temperature model is from Chapter 23.3 of Duffie and Beckman (2013), also de- scribed in De Soto (2004b)." Should we mention those references as well?

I agree with the comment. Let me see if I can locate a copy of Duffie and Beckman. There's an issue with the De Soto reference, I downloaded both (paper and thesis) and there's no discussion in either of the cell temperature model.

Maybe the reference to De Soto is for the reflection model? https://github.com/NREL/ssc/blob/055e2847ea1f4a81b4a0062beb972ec45e77d185/shared/lib_cec6par.cpp#L176-L182

wholmgren · 2021-03-01T20:56:57Z

pvlib/temperature.py

@@ -706,3 +706,94 @@ def fuentes(poa_global, temp_air, wind_speed, noct_installed, module_height=5,
        sun0 = sun

    return pd.Series(tmod_array - 273.15, index=poa_global.index, name='tmod')
+
+
+def _adj_noct(x):


something more descriptive like _adj_for_mounting_standoff

SAM document is not clear on if we should use >= or <=. Maybe check the ssc code for consistency? And probably add a comment on this detail, too.

SAM code doesn't process a distance, only passes through an integer index to the values. Except for the "x > 3.5" case, the SAM interface text doesn't specify > or >=. So we're filling in a detail here. I'll adjust the default to 3.51 and put in some comments.

wholmgren · 2021-03-01T21:09:44Z

pvlib/temperature.py

+    cell_temp_init = ross(poa_global, temp_air, noct_adj)
+    heat_loss = 1 - eta_m_ref / tau_alpha
+    wind_loss = 9.5 / (5.7 + 3.8 * wind_adj)
+    return cell_temp_init * heat_loss * wind_loss


I don't think this is correct. ross returns air temperature + temperature correction. You want to multiply only temperature correction by heat_loss and wind_loss, but you've also multiplied the air temperature by these terms

You are correct. We can't re-use ross here.

cwhanse · 2021-03-01T23:14:12Z

Mostly looks good. Are you planning a follow up issue/pr for integration into PVSystem and ModelChain?

Yes, I wanted to get consensus on the function first.

This reverts commit 31240db.

kandersolar

I'm only partially able to reconcile the standoff adjustment values in SAM Tech Ref Eq. 10.36 with their supposed source, Fuentes Table 4. Not sure where some of these SAM/SSC values came from (e.g. 2inch -> +6C?); it almost seems like they interpolated/projected the Fuentes values. Maybe that makes sense here too -- to go from discrete to continuous, just interpolate instead of bin? Although that would be more of a departure from SAM/SSC.

Some helpful links for anyone following along:

SAM technical reference: https://www.nrel.gov/docs/fy18osti/67399.pdf
Fuentes: https://prod-ng.sandia.gov/techlib-noauth/access-control.cgi/1985/850330.pdf
SSC implementation: https://github.com/NREL/ssc/blob/055e2847ea1f4a81b4a0062beb972ec45e77d185/shared/lib_cec6par.cpp#L160
Additional SSC settings: https://github.com/NREL/ssc/blob/055e2847ea1f4a81b4a0062beb972ec45e77d185/shared/lib_pv_io_manager.cpp#L957-L963

pvlib/temperature.py

kandersolar · 2021-03-02T04:05:17Z

pvlib/temperature.py

+    tau_alpha = transmittance_absorbtance * irr_ratio
+
+    # [1] Eq. 10.37 isn't clear on exactly what "G" is. SAM SSC code uses
+    # poa_global where G appears


At the bottom of page 57 it says "The cell temperature is only calculated when the effective transmitted irradiance from Equation 10.25 is greater than zero, G>0.", and tracing back through to Eq 10.20 shows that G is after accounting for surface reflection and airmass.

kandersolar · 2021-03-02T04:10:38Z

pvlib/temperature.py

+    if effective_irradiance is None:
+        irr_ratio = 1.
+    else:
+        irr_ratio = effective_irradiance / poa_global


Not sure this is correct. Eq 10.22 has the ratio as G0/G, where G0 is irradiance after reflections and G is after reflections and airmass.

cwhanse · 2021-03-02T17:01:33Z

I'm only partially able to reconcile the standoff adjustment values in SAM Tech Ref Eq. 10.36 with their supposed source, Fuentes Table 4.

I think we should regard SAM as the source for standoff adjustment values.

About "G" vs. "G0" in the SAM document: the term "G" is redefined twice in Chapter 10 (Eq. 10.13 and 10.22 are POA broadband, Eq. 10.25 and 10.26 (and 7.8) are effective irradiance absorbed by the cells. I don't read C well, but it looked to me that G_total in the SAM function is POA irradiance, not effective irradiance.

It would be desirable if this temperature model could be used without requiring both poa_global and effective_irradiance, or a third quantity falling in between which isn't present in pvlib: absorbed irradiance, POA after reflection but before spectral adjustment, this quantity is 'S' in De Soto's papers.

The SAM implementation is more involved that what is described in Duffie and Beckman. One way forward would be to implement noct_duffie_beckman rather than noct_sam. In D&B:

'G' is POA irradiance
the transmittance/absorptance term isn't adjusted with effective_irradiance
wind speed isn't adjusted for standoff

cwhanse · 2021-03-03T22:57:35Z

I spent some time trying to get SAM to give me back cell temperatures, without success. That quantity isn't available in the time series output, and the relevant function (noct_celltemp_t) isn't directly accessible through PySAM. I was able to get a 'tcell' value using PySAM's Pv6parmod object and of course it's not what I expected to see from the inputs I provided.

At this point, my recommendation is to fall back to implementing the model in Duffie and Beckman, and adding the SAM model in the future (as a different function) when we're clear exactly what the SAM model does. @kanderso-nrel? @wholmgren?

The driver for this PR is the Solar Performance Insight project, where it would be nice to have a pre-packaged "SAM" model chain.

kandersolar · 2021-03-03T23:08:56Z

No preference from me on the way best way forward. For getting cell temps from SAM: it is available in the UI, but maybe not under a name you expect (Subarray 1 Cell Temperature), see screenshots below. You can grab numerical values from the Data tables tab. I think I've pulled it directly from SSC before, but that was before the days of PySAM so I don't know how to do it there.

Click to expand!

cwhanse · 2021-03-05T22:07:32Z

@kanderso-nrel thanks for those pointers to SAM output. I am able to reproduce the Subarray 1 Cell Temperature values. To reproduce, I equated the following terms used in the SAM code with pvlib variables:

Geff_total is equivalent to pvlib's effective_irradiance,
G_total is equivalent to pvlib's poa_global (plane of array before reflections or spectrum adjustment).

I will put comments in the pvlib code to clarify why we assign these variables and write one test that reproduces the SAM value.

cwhanse · 2021-03-07T19:57:28Z

Failed check is codecov. This is ready to go, I think.

wholmgren

👍 for test_noct_sam_against_sam. I don't know what's going on with codecov. Maybe we need to start specifying a target value for that configuration.

wholmgren · 2021-03-07T20:04:52Z

pvlib/temperature.py

+    temp_air : numeric
+        Ambient dry bulb temperature. [C]
+
+    wind_speed : numeric, default 1.0


no default value in the function signature

wholmgren · 2021-03-07T20:10:08Z

pvlib/temperature.py

+        The irradiance that is converted to photocurrent. If None,
+        assumed equal to poa_global. [W/m^2]
+
+    eta_m_ref : float


comes before effective_irradiance in function signature.

Can we call this module_efficiency so that more people can easily understand it?

I'm ok with module_efficiency but I'd rather have a common name with pvsyst_cell, where eta_m is the same quantity. I used eta_m_ref here because for SAM it's clear that it's at STC, which is not clear for Pvsyst.

wholmgren · 2021-03-07T20:10:55Z

pvlib/temperature.py

+        20C. Calculate as
+        .. math::
+
+            \eta_{m} = \frac{V_{mp} I_{mp}}{A \times 1000 W/m^2}  # noQA: W605


I'm surprised you need noqa: W605 since the whole docstring is preceded by r

Suggest changing this to inline with :math: as right now it's rendering as a code block (not a math block) on RTD.

I had the noQA before I put the r in the docstring. Should be OK without the noQA now.

kandersolar

After taking a closer look and correcting my understanding of the SSC implementation, LGTM besides @wholmgren's comments and a formatting note below.

kandersolar · 2021-03-07T20:54:16Z

pvlib/temperature.py

+        20C. Calculate as
+        .. math::
+
+            \eta_{m} = \frac{V_{mp} I_{mp}}{A \times 1000 W/m^2}  # noQA: W605


Suggest changing this to inline with :math: as right now it's rendering as a code block (not a math block) on RTD.

cwhanse added 2 commits February 25, 2021 14:09

initial code for noct cell temperature function

ad30ffe

error message, test, docs

7ab38cd

cwhanse added this to the 0.9.0 milestone Mar 1, 2021

cwhanse added the enhancement label Mar 1, 2021

complete the docstring

5bf2eef

kandersolar self-requested a review March 1, 2021 20:54

wholmgren reviewed Mar 1, 2021

View reviewed changes

cwhanse added 4 commits March 1, 2021 17:46

correct noct function, adjust tests, add comments

31240db

Revert "correct noct function, adjust tests, add comments"

077c684

This reverts commit 31240db.

redo function fixes and tests

f4d6622

finish redo of edits

1bd108b

kandersolar reviewed Mar 2, 2021

View reviewed changes

fix tests, change name to noct_sam

abdfccb

cwhanse added 5 commits March 7, 2021 10:09

add test to reproduce SAM output

442dd26

format, fix text asserts

2c0b224

more assert fixes, add rtol

88ebe92

fix tes

74494f3

docstring format

9e3c6f3

wholmgren reviewed Mar 7, 2021

View reviewed changes

kandersolar reviewed Mar 7, 2021

View reviewed changes

fixes from review

979734b

wholmgren approved these changes Mar 9, 2021

View reviewed changes

wholmgren mentioned this pull request Mar 9, 2021

change eta_m to module_efficiency #1188

Closed

wholmgren merged commit 684b247 into pvlib:master Mar 10, 2021

cwhanse mentioned this pull request Mar 12, 2021

Add noct_sam cell temperature model to PVSystem, ModelChain #1195

Merged

8 tasks

wholmgren added the SPI DOE SETO Solar Performance Insight project label Mar 15, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

NOCT cell temperature function #1177

NOCT cell temperature function #1177

cwhanse commented Feb 25, 2021 •

edited

Loading

cwhanse commented Mar 1, 2021

wholmgren left a comment

wholmgren Mar 1, 2021

wholmgren Mar 1, 2021

wholmgren Mar 1, 2021

wholmgren Mar 1, 2021

wholmgren Mar 1, 2021

cwhanse Mar 2, 2021

kandersolar Mar 2, 2021

wholmgren Mar 1, 2021

wholmgren Mar 1, 2021

cwhanse Mar 2, 2021

wholmgren Mar 1, 2021 •

edited

Loading

cwhanse Mar 2, 2021

cwhanse commented Mar 1, 2021

kandersolar left a comment

kandersolar Mar 2, 2021

kandersolar Mar 2, 2021

cwhanse commented Mar 2, 2021 •

edited

Loading

cwhanse commented Mar 3, 2021

kandersolar commented Mar 3, 2021

cwhanse commented Mar 5, 2021 •

edited

Loading

cwhanse commented Mar 7, 2021

wholmgren left a comment

wholmgren Mar 7, 2021

wholmgren Mar 7, 2021

cwhanse Mar 8, 2021 •

edited

Loading

wholmgren Mar 7, 2021

kandersolar Mar 7, 2021

cwhanse Mar 8, 2021

kandersolar left a comment

kandersolar Mar 7, 2021

	assert np.isclose(result, expected)
	assert_allclose(result, expected)

	The model is described in [1], Section 10.6.
	The model is described in [1]_, Section 10.6.

NOCT cell temperature function #1177

NOCT cell temperature function #1177

Conversation

cwhanse commented Feb 25, 2021 • edited Loading

cwhanse commented Mar 1, 2021

wholmgren left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

wholmgren Mar 1, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cwhanse commented Mar 1, 2021

kandersolar left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cwhanse commented Mar 2, 2021 • edited Loading

cwhanse commented Mar 3, 2021

kandersolar commented Mar 3, 2021

cwhanse commented Mar 5, 2021 • edited Loading

cwhanse commented Mar 7, 2021

wholmgren left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cwhanse Mar 8, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kandersolar left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cwhanse commented Feb 25, 2021 •

edited

Loading

wholmgren Mar 1, 2021 •

edited

Loading

cwhanse commented Mar 2, 2021 •

edited

Loading

cwhanse commented Mar 5, 2021 •

edited

Loading

cwhanse Mar 8, 2021 •

edited

Loading