Skip to content

bl-young/olca-ipc.py

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

olca-ipc.py

Note

Not all features and bug-fixes are currently available on the version on PyPi.org. If you want to use the latest development version, just install it directly from the master branch, e.g. with pip:

pip install -U git+https://github.com/GreenDelta/olca-ipc.py.git/@master

openLCA provides an implementation of an JSON-RPC based protocol for inter-process communication (IPC). With this, it is possible to call functions in openLCA and processing their results outside of openLCA. The olca-ipc package provides a convenience API for using this IPC protocol from standard Python (Cpython v3.6+) so that it is possible to use openLCA as a data storage and calculation engine and combine it with the libraries from the Python ecosystem (numpy, pandas and friends).

The openLCA IPC protocol is based on the openLCA data exchange format which is specified in the olca-schema repository. The olca-ipc package provides a class based implementation of the openLCA data exchange format and an API for communicating with an openLCA IPC server via instances of these classes.

The current stable version of olca-ipc is available via the Python Package Index. Thus, in order to use it, you can just install (and uninstall) it with pip:

pip install -U olca-ipc

If you want to use the current development branch you can download it from Github and install it from the extracted folder:

# optionally, first uninstall it
# pip uninstall olca-ipc
cd folder/where/you/extracted/the/zip
pip install .

In order to communicate with openLCA, you first need to start an openLCA IPC server. You can do this via the user interface in openLCA under Window > Developer Tools > IPC Server. The IPC server runs on a specific port, e.g. 8080, to which you connect from an IPC client:

import olca
client = olca.Client(8080)

An instance of the olca.Client class is then a convenient entry point for calling functions of openLCA and processing their results. The following examples show some typical uses cases (note that these are just examples without input checks, error handling, code structuring, and all the things you would normally do).

Create and link data

The olca package contains a class model with type annotations for the olca-schema model that is used for exchanging data with openLCA. With the type annotations you should get good editor support (type checks and IntelliSense). You can create, update and link data models as defined in the openLCA schema (e.g. as for processes, flows, or product systems). (Note that we convert camelCase names like calculationType of attributes and functions to lower_case_names_with_underscores like calculation_type when generating the Python API).

The olca.Client class provides methods like get, find, insert, update, and delete to work with data. The following example shows how to create a new flow and link it to an existing flow property with the name Mass:

import olca
import uuid

client = olca.Client(8080)

# find the flow property 'Mass' from the database
mass = client.find(olca.FlowProperty, 'Mass')

# create a flow that has 'Mass' as reference flow property
steel = olca.Flow()
steel.id = str(uuid.uuid4())
steel.flow_type = olca.FlowType.PRODUCT_FLOW
steel.name = "Steel"
steel.description = "Added from the olca-ipc python API..."
# in openLCA, conversion factors between different
# properties/quantities of a flow are stored in
# FlowPropertyFactor objects. Every flow needs at
# least one flow property factor for its reference
# flow property.
mass_factor = olca.FlowPropertyFactor()
mass_factor.conversion_factor = 1.0
mass_factor.flow_property = mass
mass_factor.reference_flow_property = True
steel.flow_properties = [mass_factor]

# save it in openLCA, you may have to refresh
# (close & reopen the database to see the new flow)
client.insert(steel)

Running calculations

openLCA provides different types of calculations which can be selected via the calculation_type in a calculation setup. In the following example, a calculation setup with a product system and impact assessment method is created, calculated, and finally exported to Excel:

import olca

client = olca.Client(8080)

# create the calculation setup
setup = olca.CalculationSetup()

# define the calculation type here
# see http://greendelta.github.io/olca-schema/html/CalculationType.html
setup.calculation_type = olca.CalculationType.CONTRIBUTION_ANALYSIS

# select the product system and LCIA method
setup.impact_method = client.find(olca.ImpactMethod, 'TRACI 2.1')
setup.product_system = client.find(olca.ProductSystem, 'compost plant, open')

# amount is the amount of the functional unit (fu) of the system that
# should be used in the calculation; unit, flow property, etc. of the fu
# can be also defined; by default openLCA will take the settings of the
# reference flow of the product system
setup.amount = 1.0

# calculate the result and export it to an Excel file
result = client.calculate(setup)
client.excel_export(result, 'result.xlsx')

# the result remains accessible (for exports etc.) until
# you dispose it, which you should always do when you do
# not need it anymore
client.dispose(result)

Parameterized calculation setups

In order to calculate a product system with different parameter sets, you can pass a set of parameter redefinitions directly with a calculation setup into a calculation. With this, you do not need to modify a product system or the parameters in a database in order to calculate it with different parameter values:

# ... same steps as above
setup = olca.CalculationSetup()
# ...
for something in your.parameter_data:
    redef = olca.ParameterRedef()
    redef.name = the_parameter_name
    redef.value = the_parameter_value
    # redef.context = ... you can also redefine process and LCIA method
    #                     parameters by providing a parameter context which
    #                     is a Ref (reference) to the respective process or
    #                     LCIA method; with no context a global parameter is
    #                     redefined
    setup.parameter_redefs.append(redef)

As the name says, a parameter redefinition redefines the value of an existing global, process, or LCIA method parameter.

Monte-Carlo simulations

Running Monte-Carlo simulations is similar to normal calculations but instead of calculate you call the simulator method which will return a reference to a simulator which you then use to run calculations (where in each calculation the simulator generates new values for the uncertainty distributions in the system). You get the result for each iteration and can also export the result of all iterations later to Excel. As for the results of the normal calculation, the the simulator should be disposed when it is not used anymore:

import olca

client = olca.Client(8080)

# creating the calculation setup
setup = olca.CalculationSetup()
setup.calculation_type = olca.CalculationType.MONTE_CARLO_SIMULATION
setup.impact_method = client.find(olca.ImpactMethod, 'TRACI 2.1')
setup.product_system = client.find(olca.ProductSystem, 'compost plant')
setup.amount = 1.0

# create the simulator
simulator = client.simulator(setup)

for i in range(0, 10):
    result = client.next_simulation(simulator)
    first_impact = result.impact_results[0]
    print('iteration %i: result for %s = %4.4f' %
          (i, first_impact.impact_category.name, first_impact.value))
    # we do not have to dispose the result here (it is not cached
    # in openLCA); but we need to dispose the simulator later (see below)

# export the complete result of all simulations
client.excel_export(simulator, 'simulation_result.xlsx')

# the result remains accessible (for exports etc.) until
# you dispose it, which you should always do when you do
# not need it anymore
client.dispose(simulator)

For more information and examples see the package documentation

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 99.9%
  • Batchfile 0.1%