pv-system-profiler

Estimating PV array location and orientation from real-world power datasets.

Latest Release
License
Build Status
Code Quality
Publications
PyPI Downloads
Conda Downloads

Install & Setup

1) Recommended: Set up conda environment with provided .yml file

We recommend setting up a fresh Python virtual environment in which to use pv-system-profiler. We recommend using the Conda package management system, and creating an environment with the environment configuration file named pvi-user.yml, provided in the top level of this repository. This will install the statistical-clear-sky and solar-data-tools packages as well.

Creating the env:

bash $ conda env create -f pvi-user.yml

Starting the env:

bash $ conda activate pvi_user

Stopping the env

bash $ conda deactivate

Additional documentation on setting up the Conda environment is available here.

2) PIP Package

sh $ pip install pv-system-profiler

Alternative: Clone repo from GitHub

Mimic the pip package by setting up locally.

bash $ pip install -e path/to/root/folder

3) Anaconda Package

sh $ conda install -c slacgismo pv-system-profiler

Solver Dependencies

Refer to solar-data-tools documentation to get more info about solvers being used.

Usage / Run Scripts

Serial run

The parameter_estimation_script.py script creates a report of all systems based on the csv files with the system signals located in a given folder. The script takes all input parameters as kwargs. The example below illustrates the use of reportscript: ```shell python 'repository location of run script'/parameterestimationscript.py report None all s3://s3bucketwithsignals/ 'repeatingpartof label' /home/results.csv True False False False s3://'s3pathtofilecontainingmetadata/metadata.csv' None s3 `` In the example above the full path toparameterestimationscript.pyis specified to run a report. The script allows to provide acsvfile with list of sites to be analyzed. In this case no list is provided and therefore thekwargNoneis entered. The script also allows to run an analysis on the firstnfilescontaining input signals in thes3repository. In this, case theallkwargspecifies that all input signals are to be analyzed. In this example, allcsvfiles containing the input signals are located in thes3bucket with the name s3://s3bucketwithsignals/. Usually thesecsvfiles are of the formIDrepeatingpartoflabel.csv, for example: 1composite10.csv,2composite10.csv, wherecomposite10is the repeating part of the label. The repeating part of the label is either None or a string as in the example above. Next, an absolute path to the desired location of the results file is provided, in this case/home/results.csv. The two followingkwargsare type Boolean and are used to set the values of thecorrecttzandfixshiftspipelinekwargs. The nextkwarg,checkjsonis also Boolean. It is used to indicate if there is ajsonfile present ins3://s3bucketwithsignals/with additional site information that is to be analyzed. The next Booleankwargis used to set theconverttotskwargwhen instantiating the data handler. The nextkawrgcontains the full path to thecsvfile containing site metadata, here calledmetadata.csv. The information that this file should contain varies depending on theestimationto be performed. This file is optional and thekwargcan be set toNone. For the case of areport, acsvfile with columns labeledsite, systemandgmtoffsetand their respective values need to be provided. Alternatively, if thegmtoffsetkwarg, the nextkwarg(in the example above set toNone), has a numeric value different toNone, all sites will use that single value when running the report. For the case of thereport estimation, the metadata file should containsite,systemandgmtoffsetcolumns with the respective values for each system. For the case of thelongitudeestimation, the metadata file should containsite,system andlatitudecolumns with the respective values for each system. For the case of thetiltazimuthestimation, the metadata file should containsite,system,gmtoffset,estimatedlongitudeandestimatedlatitude,tilt, azimuthcolumns and with the respective values for each system. Additionally, if a manual inspection for time shifts was performed, another column labeledtimeshiftmanualhaving a zero for systems with no time shift and ones for systems with time shift may be included. If atimeshiftmanualcolumn is included, it will be used to determine whether thefixdst() method is run after instantiating the data handler. The nextkarg isgmtoffsetand in this case it is set to None. The last kwarg corresponds to thedatasource. In this case the value iss3since files with the input signals are located in ans3bucket. ## Partitioned run A script that runs the site report, the longitude, latitude and tilt and azimuth scripts using a number of prescribed Amazon Web Services (AWS), instances is provided. The script reads the folder containing the system signals and partitions these signals to run in anuser prescribed AWS instances in parallel. Here is an example shell command for a partitioned run: ``shell python 'repository location of run script'/runpartitionscript.py parameterestimationscript.py report None all s3://s3bucketwithsignals/ 'repeatingpartof label' /home/results.csv True False False False s3://'s3pathtofilecontainingmetadata/metadata.csv' None s3 'repository location of run script'/parameterestimationscript.py pvi-dev myinstance `` where the individual value of each kwarg are defined in run_partition_script.py. This script takes the same inputs as theparameterestimationscript.pyplus three additional parameters. Note that the first kwarg is the partitioning script repository location of run script/runpartitionscript.py parameterestimationscript.py. The estimation run script/parameterestimationscript.pis specified as the third to last kwarg. The second to last kwarg is the conda enviroment to be used to run the estimation, in this casepvi-dev. The last kwarg is the name of the AWS instances to be used to runrunpartitionscript.py, in this casemyinstance. Previous to running this command it is necessary to createnidentical AWS instances that correspond to the number of desired partitions. These instances need to have the same Name='instance name'AWS tag. The simplest way to accomplish this is by parting from an AWS image of a previously configured instance. This image needs to have all the repositories and conda environments that would be needed in a serial run. Once each partitioned run is finished, results will be automatically collected in the local folder whererunpartition_script.py` was run.

Unit tests

In order to run unit tests: python -m unittest -v

Test Coverage

In order to view the current test coverage metrics: coverage run --source pvsystemprofiler -m unittest discover && coverage html open htmlcov/index.html

Versioning

We use Semantic Versioning for versioning. For the versions available, see the tags on this repository.

License

This project is licensed under the BSD 2-Clause License - see the LICENSE file for details