sandialabs / matlab_pv_lib Goto Github PK

For equally-spaced data, the slope of GHI assumes 1 minute spacing.

Condition 5 is implemented as a difference between two maxima, where it should be the maximum of the absolute value of a difference.

See Issue at pvlib-python #505

Describe the bug:

Low sun angles with a tilted tracking axis cause an error in the backtracking calculation. Code attempts arcos() of a value outside [-1, 1]
pvlib-python #656

To Reproduce

SunZen=80;
SunAz=338;
Latitude=nan;
AxisTilt=30;
AxisAzimuth=180;
MaxAngle=60;
GCR=0.35;

Backtrack=0;
[~,~,mod_thetam, mod_azm]= pvl_singleaxis_v2(SunZen, SunAz, Latitude, AxisTilt, AxisAzimuth, MaxAngle, Backtrack, GCR)

Backtrack=1;
[~,~,mod_thetam, mod_azm]= pvl_singleaxis_v2(SunZen, SunAz, Latitude, AxisTilt, AxisAzimuth, MaxAngle, Backtrack, GCR)

This line should be

B = 2*pi*(doy - 1)/365;

As described in the original source, the day of year runs between 0 and 364.

In function “pvl_calcparams_CEC.m” the parameter dEgdT is set to 0.00002677 (i.e. 2.677e-5).
According to the documentation and the cited paper “[1] W. De Soto et al., "Improvement and validation of a model for photovoltaic array performance", Solar Energy, vol 80, pp. 78-88, 2006.”, the value should be one order of magnitude higher (2.677e-4). The Python version of the library also implements the larger value for dEgdT.

Thanks to Patrick for finding the bug.

pvl_singleaxis produces an incorrect AOI when the sun is above the earth horizon but behind the module plane.

Issue seems to be in this line of code.

Similar to issue 1221 for the python pvlib

Can we have a function to convert from PVSyst .pan files to SAPM coefficients?

For comparison with measured data, it would be great if pvl_singlediode accepted a list of voltages at which to return calculated current.

In the Klucher transposition model, direct irradiance is included in plane of array total when cos(AOI) < 0 (i.e., the AOI indicates that the sun is behind the array plane).

See pvlib/pvlib-python#526

This line

In order for PV_LIB to used as a commercial verification tool (which may be up for debate) or a verification tool (may also be up for debate), there should be a method to verify that the files within PV_LIB are the same as those distributed officially. A hash listing for each file could accomplish this.

@DanRiley I forgot to ask you to add to whatsnew.rst for the new function.

We currently use 3 iterations of a Newton method when the argument to the lambertw term is calculated in log-space. As the magnitude of the argument increases, the number of iterations should also increase to maintain the needed precision.

Some inverters derate across a range of temperature, sometimes these derates are included in manufacturer-provided information. Right now, the inverter models do not include ambient temperature derating. Would it be possible to add this?

pvl_singlediode returns a structure Result with the field Isc, and also can the field I containing a vector of current values. The first value of Result.I is not always equal to Result.Isc

The input parsing of pvl_soiling_hsu does not accept temperatures less than 0. This seems to be a bug, the correct check value for temperature should allow temperatures greater than -273.15.

On function : https://github.com/sandialabs/MATLAB_PV_LIB/blob/master/pvl_readtmy3.m
Move line 151 to 161 to after the "end" of "if isempty(strfind(char(DataLines{1,1}{1,1}), 'PresWth'))" in line 178.
Otherwise output TMYData won't have all that header information.

Sandia National Laboratories is conducting research on the change of snow albedo over time and temperature since snowfall. If a model is developed that is successful in determining this relationship, include it in MATLAB_PV_LIB.

There is a mistake in the calculation of the diffuse fraction in the Orgill Hollands function. There should be no jump discontinuities when plotting.

This should be changed from

if Kt(i)<=.35
    DF(i) = 1.0 - 0.249*Kt(i);
        
    % For Kt > 0.35 and Kt <= 0.75, set the diffuse fraction
elseif Kt(i)>.35 && Kt(i)<.75
    DF(i) = 1.577 - 1.84*Kt(i);
        
    % For Kt > 0.75, set the diffuse fraction
else
    DF(i) = 0.177;
end

To this

if Kt(i)<=.35
    DF(i) = 1.0 - 0.249*Kt(i);
        
    % For Kt > 0.35 and Kt <= 0.75, set the diffuse fraction
elseif Kt(i)>.35 && Kt(i)<=.75
    DF(i) = 1.557 - 1.84*Kt(i);
        
    % For Kt > 0.75, set the diffuse fraction
else
    DF(i) = 0.177;
end

To make it very clear DF(i) = 1.577 - 1.84*Kt(i); should be DF(i) = 1.557 - 1.84*Kt(i); and
elseif Kt(i)>.35 && Kt(i)<.75 should be elseif Kt(i)>.35 && Kt(i)<=.75

All according to Orgill, J. F. and Hollands K. G. T. "Correlation Equation for Hourly Diffuse Radiation on a Horizontal Surface" 1977.

A soiling model was presented by Merissa Coello and Liza Boyle from Humboldt University at PVSC/WCPEC 2018. Requires time-series data of particulate matter (PM) concentrations for 10 micron and 2.5 micron particle diameters, and rainfall. Other parameters needed for variable deposition models. I have a copy of the model paper from Merissa and Liza, but can't find a reference to the proceedings. I have a model function developed and am putting final touches on the model. I'll PR when complete.

Says 1.3 for 1.4 release

pvl_singleaxis can produce NaN results for the surface tilt angle when the tilt is 0. Error stems from the following line:
SurfTilt = 90 - acosd(dividend./divisor);

When divisor is 0, the division by 0 produces NaN.

Possible fixes could be setting divisor to a very small number (eps) when it is 0, or simply setting SurfTilt to 0 any time that divisor is 0.