This package can be used to generate input files for the Temoa modelling tool. Transparency and repeatability are integral to the design motivations for Temoa. PyGenesys takes this concept even further by reducing the lead time and programming background required to run Temoa. PyGenesys offers technology models with cost and production time series data. The basic model is based on the state of Illinois and its flagship university, the University of Illinois at Urbana-Champaign.
Clone the repository and create a conda environment
git clone git@github.com:temoaproject/temoa.git
cd temoa
conda env create
conda activate temoa-py3
The installation is quite simple:
$ git clone git@github.com:arfc/pygenesys.git
$ cd pygenesys
There are two options for installation after cloning the repository. If you don't plan on editing the source code, use the first option.
$ pip install .
This option is for people that want to develop PyGenesys. By doing this method, any changes you make to the PyGenesys source code will take immediate effect on your local machine.
$ pip install -e .
and you're done!
pygenesys
is a package that is intended to be a simple and intuitive interface to run Temoa.
A pygenesys
input file is written in Python with clear and verbose variable names. The
beginning should look something like this:
import os
curr_dir = os.path.dirname(__file__)
folder = 'data_files'
database_filename = f'{folder}/my_temoadb.sqlite' # where the database will be written
scenario_name = 'test'
start_year = 2025
end_year = 2050
N_years = 5
N_seasons = 4 # the number of seasons in the model
N_hours = 24 # the number of hours in a day
The first two lines
import os
curr_dir = os.path.dirname(__file__)
are recommended, but not required. These lines tell pygenesys
to save the
database in the same directory as your input file. Otherwise, the database will
be saved in the pygenesys/pygenesys
folder.
The parameters
reserve_margin
(The fraction of excess capacity, for grid reliability)discount_rate
(The global discount rate)
are optional global parameters.
reserve_margin = {'region':0.15}
discount_rate = 0.05
Users should design their energy systems from right to left. Begin by
initializing the end-use demands. You can use either a preset commodity
located in pygenesys.commodity.demand
or build your own using the
DemandCommodity
class.
# Method 1:
from pygenesys.commodity.demand import ELC_DEMAND # initialized
# Method 2:
from pygenesys.commodity.commodity import DemandCommodity
H2 = DemandCommodity(comm_name='H2',
units='kg',
description='hydrogen')
# initialized
This step answers the following questions:
- How much of each commodity should Temoa produce?
- When should the commodity be produced (e.g. mostly at night, peak hours, etc.)?
- How does the demand change over time?
- How does the demand vary by region?
Users should call DemandCommodity.add_demand()
and DemandCommodity.set_distribution
.
ELC_DEMAND.add_demand(region='region', # specifies the region
init_demand=445.87, # the demand in first simulation year
start_year=start_year, # starting year
end_year=end_year, # ending year
N_years=N_years, # number of years
growth_rate=0.01, # modeled growth rate
growth_method='linear') # how does the growth change?
H2.add_demand(region='region',
init_demand=100,
start_year=start_year,
end_year=end_year,
N_years=N_years,
growth_rate=0.00) # growth rate of zero means demand is constant
Users should call DemandCommodity.add_demand
for each unique region and
demand commodity.
In the future, this function will accept a list for region
if all regions
have the same forecast.
The Demand distribution specifies load profiles. For example, electricity demand might peak in the summer at noon and natural gas demand might peak in the winter at night. This method specifies those differences.
** PyGenesys currently only handles demand distributions with 24 hour slices and four annual seasons. Distributions for time slices with fewer or more time slices should be generated manually and passed as a list or numpy array.**
PyGenesys has a small library of built in data in pygenesys.data.library
.
from pygenesys.data.library import campus_elc_demand
ELC_DEMAND.set_distribution(region='region',
data=campus_elc_demand,
n_seasons=N_seasons,
n_hours=N_hours)
# passing no distribution for H2 will default to a uniform distribution.
DemandCommodity
objects represent ultimate demands. They must be generated
somehow. They can be created directly from a production technology (e.g. a
power plant) or they can be "transported" using a transmission technology. The
latter is beneficial if the final demand and intermediate energy carriers are
similar. For example if you want to use electricity for electric cars and also
have an aggregated electricity demand.
from pygenesys.commodity.resource import electricity # an intermediate resource
from pygenesys.technology.transmission import TRANSMISSION
from pygenesys.commodity.commodity import Commodity
hydrogen = Commodity(comm_name='hydrogen',
units='kg',
description='hydrogen')
Users must also specify unique regional data for each technology by calling
Technology.add_regional_data()
.
TRANSMISSION.add_regional_data(region='region',
input_comm=electricity,
output_comm=ELC_DEMAND,
efficiency=1.0,
tech_lifetime=1000,)
Similar to DemandCommodity
, users can create their own technologies
using the Technology
class.
from pygenesys.technology.technology import Technology
H2_PIPES = Technology(tech_name='H2_PIPES',
units='kg-hours',
tech_sector='hydrogen',
tech_label='p',
description='hydrogen transport',
category='transmission',
capacity_to_activity=1.00,) # C2A is required
H2_PIPES.add_regional_data(region='region',
input_comm=hydrogen,
output_comm=H2,
efficiency=1.0,
tech_lifetime=1000,)
The technology lifetimes for "dummy" technologies are usually much longer than the simulation time because it can be built once to emphasize the importance of other technology choices. These are, of course, tunable parameters if one wishes to model transmission costs.
So far we only have dummy technologies that can supply ultimate demands and will pull from dummy commodities. Let's make those commodities real. The only difference is that these technologies must have the desired energy carrier as their output commodity.
If the region you're modeling is at the state or county level in the United States,
then you can access EIA data about those technologies using pygenesys.data.eia_data
.
from pygenesys.data.eia_data import get_eia_generators, get_existing_capacity
curr_data = get_eia_generators()
from pygenesys.technology.electric import NUCLEAR_ELC
NUCLEAR_ELC.add_regional_data(region='IL',
input_comm=ethos,
output_comm=electricity,
efficiency=1.0,
tech_lifetime = 40.0,
existing=get_existing_capacity(curr_data,
'IL',
'Nuclear'),
cost_invest=nuclear_invest,
cost_fixed=nuclear_fixed_annual,
cost_variable=nuclear_variable_annual,
capacity_factor_tech=0.935,
ramp_up=0.25,
ramp_down=0.25,
loan_lifetime=40,
emissions={co2eq:1.2e-5}
)
# Add more! As an exercise, try creating an electrolyzer to make hydrogen...
As a final step, users should collect all of the commodities they use by type.
# copy and paste these lines, then fill in...
demands_list = []
resources_list = []
emissions_list = []
This will be deprecated, eventually, and replaced with a function that automatically collects the commodities by reading the input file.
If you want to check your input file, plot any data, or make some scratch calculations, you can use
if __name__ == "__main__":
# your scratch work here!
import matplotlib.pyplot as plt
plt.style.use('ggplot')
plt.plot(ELC_DEMAND.demand['region'][0])
plt.show()
In fact, this feature is one of the advantages of using a python script as an input file!
Now that you have an input file, you can build the SQLite database required by Temoa via:
$ genesys --infile path/to/my/input/file.py
Note: This command can be run from any directory.
The tests can be run by executing the following command from the top level
directory of pygenesys
.
$ pytest
The documentation is generated using sphinx with numpy style docstrings, to render them locally run:
$ cd docs
$ make html
First, consult the CONTRIBUTING document (in progress). We recommend being passingly familiar with reStructuredText , though it is not a pre-requiste. Then you should be equipped to address issues from the GitHub repo, or add additional documentation to help users and developers alike! Most of the documenation is auto-generated from python docstrings in the numpy style.
Some of the code in driver.py
was borrowed and modified from another open
source project PyRK
.