model uncertain emissions together with known emissions over time using brightway - brightway

Question

I would like to understand how to model exchanges whose values are known to change sequentially (over time), together with exchanges that are uncertain. So for each period some will have known values, others uncertain (but independent). I think with the interfaces in brightway 2.5 this is possible. But I have not figured out how to do it.
A simple example, An activity emitting CO2 and CH4. Imagine we have measurements over time of the CO2 emissions but CH4 emissions follow some sort of random state that is independent of CH4. So for each period the CH4 emissions could take several values.
I think I know how to model the CO2 emissions using a dynamic_vector (see example below) but I don't know how I would model the CH4 emissions in this example. Perhaps with a dynamic_array ?
import brightway2 as bw
import numpy as np
import bw_processing as bwp
import bw2data as bd
import bw2calc
bw.projects.set_current('testing_things')
co2 = next(x for x in bd.Database("biosphere3")
if x['name'] == 'Carbon dioxide, fossil'
and x['categories'] == ('air',))
ch4 = next(x for x in bd.Database("biosphere3")
if x['name'] == 'Methane, fossil'
and x['categories'] == ('air',))
a_key = ("testdb", "a")
act_a_def = {
'name': 'a',
'unit': 'kilogram',
'exchanges': [{"input": co2.key, "type": "biosphere", "amount": 10},
{"input": a_key, "output":a_key,'type':'production','amount':1},
{"input": ch4.key, "type": "biosphere", "amount": 1},
],
}
db = bd.Database("testdb")
db.write(
{
a_key : act_a_def,
}
)
ipcc2013 = ('IPCC 2013', 'climate change', 'GWP 100a')
a = bw.get_activity(('testdb','a'))
lca = bw.LCA({a:1},ipcc2013)
lca.lci()
lca.lcia()
lca.score
lets assume I know the CO2 emissions because I have a sensor measuring them
co2_measurements = [10,11,9,10,8,11,12,13,8,12]
co2_interface = (np.array([c]) for c in co2_measurements)
indices_array=np.array([(co2.id, a.id),],
dtype=bwp.INDICES_DTYPE
)
flip_array=np.array([True])
hp = bwp.create_datapackage()
hp.add_dynamic_vector(
matrix="biosphere_matrix",
interface=co2_interface,
indices_array=indices_array,
flip_array=flip_array,
)
fu, data_objs, _ = bd.prepare_lca_inputs({a: 1}, method=ipcc2013)
lca = bw.LCA(fu, data_objs=data_objs + [hp])
lca.lci()
lca.lcia()
print(lca.score)
for _ in range(5):
next(lca)
print(lca.score)
18.700000762939453
20.700000762939453
19.700000762939453
21.700000762939453
18.700000762939453
17.700000762939453

Answer 1

I found a way to do it but it strikes me as awkward and probably not the way the dynamic calculations were thought to be used, as the datapackage is instantiated each time. In this solution the CH4 is the one with a time series and the CO2 the random one, but it does not really matter.
co2 = next(x for x in bwd.Database("biosphere3")
if x['name'] == 'Carbon dioxide, fossil'
and x['categories'] == ('air',))
ch4 = next(x for x in bwd.Database("biosphere3")
if x['name'] == 'Methane, fossil'
and x['categories'] == ('air',))
indices_co2 = np.array([(co2.id,act.id)],dtype=bwp.INDICES_DTYPE)
flip_array_co2 = np.array([False] * len(indices_co2))
indices_ch4 = np.array([(ch4.id,act.id)],dtype=bwp.INDICES_DTYPE)
flip_array_ch4 = np.array([False] * len(indices_ch4))
ch4_measurements = [np.array([r/10]) for r in range(5,100,1)]
class Interface_timeseries:
def __init__ (self, ts_data):
self.ts = ts_data
self.idx = 0
def __iter__ (self):
return self
def __next__ (self):
try:
item = self.ts[self.idx]
except IndexError:
raise StopIteration()
self.idx += 1
return item
def init_dynamic(act,lcia_method,co2_vector,ch4_vector):
dynamic_ghg = bwp.create_datapackage(combinatorial=False,sequential=False)
# CO2 random emissions
dynamic_ghg.add_dynamic_vector(
matrix = co2_vector['matrix'],
indices_array = co2_vector['indices_array'],
flip_array = co2_vector['flip_array'],
interface = co2_vector['interface'])
# CH4 measurements
dynamic_ghg.add_dynamic_vector(
matrix = ch4_vector['matrix'],
indices_array = ch4_vector['indices_array'],
flip_array = ch4_vector['flip_array'],
interface = ch4_vector['interface'],)
fu, data_objs , _ = bwd.prepare_lca_inputs({act:1},method=lcia_method)
lca_dynamic = bwc.LCA(fu, data_objs= data_objs +[dynamic_ghg])
lca_dynamic.lci()
lca_dynamic.lcia()
return(lca_dynamic)
iteration_results = []
for it in range(10):
co2_vector = {}
co2_vector['matrix'] = 'biosphere_matrix'
co2_vector['indices_array'] = indices_co2
co2_vector['flip_array'] = flip_array_co2
co2_vector['interface'] = (np.array([row]) for row in np.random.normal(size=100))
ch4_vector = {}
ch4_vector['matrix'] = 'biosphere_matrix'
ch4_vector['indices_array'] = indices_ch4
ch4_vector['flip_array'] = flip_array_ch4
ch4_vector['interface'] = Interface_timeseries(ch4_measurements)
dlca = init_dynamic(act,ipcc2013,co2_vector,ch4_vector)
scores = []
for _ in range(len(ch4_measurements)):
try:
next(dlca)
scores.append(dlca.score)
except StopIteration as si:
print('sequence completed')
break
iteration_results.append(scores)
Open to suggestions on how to do it better