Class 1: Introduction and setting up a simple LCA calculation
Contents
Class 1: Introduction and setting up a simple LCA calculation#
Note: this notebook is based on Brightway2 tutorial 4 - meta-analysis.
Setup#
Import the necessary libraries
from brightway2 import *
import numpy as np
import pyprind
from scipy import stats
import random
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
Input In [1], in <cell line: 1>()
----> 1 from brightway2 import *
2 import numpy as np
3 import pyprind
ModuleNotFoundError: No module named 'brightway2'
Set a new project for this class session
projects.current = "Class 1"
Import the basic biosphere and LCIA methods (requires internet connection)
bw2setup()
Creating default biosphere
Applying strategy: normalize_units
Writing activities to SQLite3 database:
0% 100%
[##############################] | ETA[sec]: 0.000
Total time elapsed: 0.337 sec
Applying strategy: drop_unspecified_subcategories
Applied 2 strategies in 0.01 seconds
Title: Writing activities to SQLite3 database:
Started: 10/19/2015 09:46:40
Finished: 10/19/2015 09:46:40
Total time elapsed: 0.337 sec
CPU %: 96.100000
Memory %: 0.325644
Created database: biosphere3
Creating default LCIA methods
Applying strategy: normalize_units
Applying strategy: set_biosphere_type
Applying strategy: drop_unspecified_subcategories
Applying strategy: link_iterable_by_fields
Applied 4 strategies in 1.00 seconds
Wrote 692 LCIA methods with 170915 characterization factors
Creating core data migrations
Import ecoinvent 2.2#
We are using version 2.2 to make the calculations a bit quicker, as this is intended to be run during the class.
First, extract the data from the XML files.
ei = SingleOutputEcospold1Importer("/Users/cmutel/Documents/LCA Documents/Ecoinvent/2.2/processes", "ecoinvent 2.2")
Extracting ecospold1 files:
0% 100%
[##############################] | ETA[sec]: 0.000 | Item ID: /Users/cmutel/D
Total time elapsed: 8.460 sec
Title: Extracting ecospold1 files:
Started: 10/19/2015 09:48:02
Finished: 10/19/2015 09:48:10
Total time elapsed: 8.460 sec
CPU %: 78.700000
Memory %: 0.845587
Extracted 4087 datasets in 8.51 seconds
On windows, you will need to escape your backslashes by repeating them twice, e.g.
ei = SingleOutputEcospold1Importer("c:\\Users\\cmutel\\Process_infra_roh", "ecoinvent 2.2")
Next, we normalize some values, and link the different datasets to each other and the basic biosphere
ei.apply_strategies()
Applying strategy: normalize_units
Applying strategy: assign_only_product_as_production
Applying strategy: clean_integer_codes
Applying strategy: drop_unspecified_subcategories
Applying strategy: normalize_biosphere_categories
Applying strategy: normalize_biosphere_names
Applying strategy: strip_biosphere_exc_locations
Applying strategy: set_code_by_activity_hash
Applying strategy: link_iterable_by_fields
Applying strategy: link_technosphere_by_activity_hash
Applied 10 strategies in 2.35 seconds
Check to make sure everything is linked
ei.statistics()
4087 datasets
135892 exchanges
0 unlinked exchanges
(4087, 135892, 0)
Finally, write the database
ei.write_database()
Writing activities to SQLite3 database:
0% 100%
[##############################] | ETA[sec]: 0.000
Total time elapsed: 10.019 sec
Title: Writing activities to SQLite3 database:
Started: 10/19/2015 10:36:24
Finished: 10/19/2015 10:36:34
Total time elapsed: 10.019 sec
CPU %: 91.900000
Memory %: 0.883520
Created database: ecoinvent 2.2
Brightway2 SQLiteBackend: ecoinvent 2.2
list(databases)
['biosphere3', 'ecoinvent 2.2']
list(methods)[:10]
[('CML 2001', 'photochemical oxidation (summer smog)', 'MIR'),
('ReCiPe Endpoint (H,A)', 'human health', 'ionising radiation'),
('eco-indicator 99, (H,A) w/o LT',
'ecosystem quality w/o LT',
'total w/o LT'),
('EDIP', 'resource consumption', 'non-renewable resources, mercury'),
('IMPACT 2002+ (Endpoint)', 'human health', 'human toxicity'),
('ReCiPe Endpoint (H,A) w/o LT',
'resources w/o LT',
'fossil depletion w/o LT'),
('EDIP2003', 'global warming', 'GWP 20a'),
('EDIP', 'resource consumption', 'non-renewable resources, natural gas'),
('IMPACT 2002+ (Endpoint)',
'ecosystem quality',
'terrestrial acidification & nutrification'),
('EDIP2003', 'ecotoxicity', 'acute, in water')]
Do some calculations#
We select the new database we just created, and then use a bit of trickiness to get random ordering. All datasets have the type "process", so this is equivalent to random ordering.
db = Database("ecoinvent 2.2")
db.order_by = "type"
Many of the built-in LCIA methods are too narrowly focused - we pre-select a set of candidates to use for our meta-calculations.
CANDIDATES = sorted([
(u'CML 2001', u'acidification potential', u'average European'),
(u'CML 2001', u'climate change', u'GWP 100a'),
(u'CML 2001', u'eutrophication potential', u'average European'),
(u'CML 2001', u'freshwater aquatic ecotoxicity', u'FAETP 100a'),
(u'CML 2001', u'human toxicity', u'HTP 100a'),
(u'CML 2001', u'land use', u'competition'),
(u'CML 2001', u'marine aquatic ecotoxicity', u'MAETP infinite'),
(u'CML 2001', u'resources', u'depletion of abiotic resources'),
(u'CML 2001', u'stratospheric ozone depletion', u'ODP 25a'),
(u'EDIP2003', u'ecotoxicity', u'in sewage treatment plants'),
(u'EDIP2003', u'eutrophication', u'terrestrial eutrophication'),
(u'EDIP2003', u'renewable resources', u'wood'),
(u'EDIP2003', u'stratospheric ozone depletion', u'ODP total'),
(u'EPS 2000', u'total', u'abiotic stock resources'),
(u'EPS 2000', u'total', u'emissions into soil'),
(u'EPS 2000', u'total', u'emissions into water'),
(u'EPS 2000', u'total', u'land occupation'),
(u'IMPACT 2002+ (Endpoint)', u'ecosystem quality', u'land occupation'),
(u'IMPACT 2002+ (Endpoint)', u'human health', u'ozone layer depletion'),
(u'IMPACT 2002+ (Endpoint)', u'resources', u'mineral extraction'),
(u'IMPACT 2002+ (Endpoint)', u'resources', u'non-renewable energy'),
(u'IMPACT 2002+ (Midpoint)', u'ecosystem quality', u'aquatic acidification'),
(u'IPCC 2001', u'climate change', u'GWP 100a'),
(u'ReCiPe Endpoint (H,A)',
u'ecosystem quality',
u'agricultural land occupation'),
(u'ReCiPe Endpoint (H,A)',
u'ecosystem quality',
u'freshwater eutrophication'),
(u'ReCiPe Endpoint (H,A)',
u'ecosystem quality',
u'natural land transformation'),
(u'ReCiPe Endpoint (H,A)',
u'ecosystem quality',
u'terrestrial acidification'),
(u'ReCiPe Endpoint (H,A)', u'ecosystem quality', u'urban land occupation'),
(u'ReCiPe Endpoint (H,A)', u'human health', u'particulate matter formation'),
(u'ReCiPe Endpoint (H,A)', u'resources', u'fossil depletion'),
(u'TRACI', u'environmental impact', u'acidification'),
(u'TRACI', u'environmental impact', u'eutrophication'),
(u'TRACI', u'environmental impact', u'global warming'),
(u'TRACI', u'environmental impact', u'ozone depletion'),
(u'TRACI', u'human health', u'respiratory effects, average'),
(u'eco-indicator 99, (H,A)',
u'ecosystem quality',
u'acidification & eutrophication'),
(u'eco-indicator 99, (H,A)', u'ecosystem quality', u'ecotoxicity'),
(u'eco-indicator 99, (H,A)', u'ecosystem quality', u'land occupation'),
(u'eco-indicator 99, (H,A)', u'human health', u'carcinogenics'),
(u'eco-indicator 99, (H,A)', u'human health', u'climate change'),
(u'eco-indicator 99, (H,A)', u'human health', u'ozone layer depletion'),
(u'eco-indicator 99, (H,A)', u'resources', u'fossil fuels'),
(u'eco-indicator 99, (H,A)', u'resources', u'mineral extraction'),
(u'ecological footprint', u'total', u'CO2'),
(u'ecological footprint', u'total', u'land occupation'),
(u'ecological footprint', u'total', u'nuclear'),
(u'ecological scarcity 2006', u'total', u'deposited waste'),
(u'ecological scarcity 2006', u'total', u'emission into groundwater'),
(u'ecological scarcity 2006', u'total', u'energy resources'),
(u'ecological scarcity 2006', u'total', u'natural resources'),
(u'ecosystem damage potential', u'total', u'linear, land occupation'),
(u'ecosystem damage potential', u'total', u'linear, land transformation'),
])
assert all(x in methods for x in CANDIDATES)
print("There are %s methods to test" % len(CANDIDATES))
There are 52 methods to test
Choose ten LCIA methods and 500 datasets from ecoinvent 2.2 at random
chosen_methods = random.sample(CANDIDATES, 10)
chosen_processes = []
for index, obj in enumerate(db):
if index >= 500:
break
else:
chosen_processes.append(obj)
Set up the LCA object, optimized to do many calculations.
See making LCA calculations faster blog post for more details on factorization.
lca = LCA({chosen_processes[0]: 1}, method=chosen_methods[0])
lca.lci(factorize=True)
lca.lcia()
Create an array to store our LCA results - processes on rows, methods on columns
results = np.zeros((500, 10))
Do 5000 LCA calculations in a single thread. Store the results in results.
bar = pyprind.ProgBar(5000, monitor=True)
for col, method in enumerate(chosen_methods):
lca.method = method
lca.load_lcia_data()
for row, process in enumerate(chosen_processes):
lca.redo_lcia({process: 1})
results[row, col] = lca.score
bar.update()
print(bar)
0% 100%
[##############################] | ETA[sec]: 0.000
Total time elapsed: 22.971 sec
Title:
Started: 09/29/2015 23:22:36
Finished: 09/29/2015 23:22:59
Total time elapsed: 22.971 sec
CPU %: 98.000000
Memory %: 5.591297
We only care about processes which have non-zero LCA scores - there are a few processes in ecoinvent 2.2 which we want to filter automatically (if they are selected in our earlier random sample).
mask = (results.sum(axis=1) != 0)
print("Ignoring {} processes".format((~mask).sum()))
Ignoring 8 processes
Calculate the rank-order correlation for all processes
def create_correlation_matrix(scores_array):
num_methods = scores_array.shape[1]
correlations = np.zeros((num_methods, num_methods))
for row in range(num_methods):
for col in range(num_methods):
if col <= row:
continue # Only need to compute correlation once
dataset_1 = scores_array[:, row]
dataset_2 = scores_array[:, col]
mask = (dataset_1 != 0) * (dataset_2 != 0) # Ignore activities that have zero score
correlations[row, col] = stats.kendalltau( # Get tau value, drop p-statistic
dataset_1[mask],
dataset_2[mask]
)[0]
return correlations
correlation_matrix = create_correlation_matrix(results[mask, :])
Visualize the results
%matplotlib inline
import matplotlib.pyplot as plt
fig = plt.gcf()
fig.set_size_inches(12, 12)
masked_correlation = np.ma.array(correlation_matrix, mask=correlation_matrix == 0).T
plt.pcolor(masked_correlation, cmap=plt.cm.cubehelix_r)
plt.colorbar(label=r"Kendall $\tau$ rank-order correlation coefficient")
plt.ylim(None, 10)
plt.xlim(None, 10)
plt.axis('off')
(0.0, 10.0, 0.0, 10.0)
