Brussels. Step 3.a Defining simple transition scenarios¶
In [1]:
import datetime; print(datetime.datetime.now())
2018-04-09 10:49:38.925958
Notebook Abstract:
The following notebook describes the process to construct simple transition scenarios.
The transition scenarios are define as efficiency rates induced by technology development or behavioral changes. These rates can be used as proxies for all types of efficiency improvements.
In order to define transition scenarios the model need the following information:
- A technology penetration rate. This defines the share of the
population adopting the technology. The model uses
sampling rulesfor the selection of the population adopting this technology. - Development of efficiency rates. This define the actual technology development rate.
Import libraries¶
In [2]:
from smum.microsim.run import transition_rate
from smum.microsim.util_plot import plot_transition_rate
from smum.microsim.run import reduce_consumption
/usr/lib/python3.6/site-packages/h5py-2.7.1-py3.6-linux-x86_64.egg/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
from ._conv import register_converters as _register_converters
In order to compute the transition scenarios we make use of three
modules of the urbanmetabolism library:
growth_rate. This module will return a vector with linear growth rates given a starting and end rate.plot_growth_rate. This a simple function to visualize the defined growth rates.reduce_consumption. This function creates new samples with reduced consumption levels for the selected selection of the population.
Define simple population selection rules¶
In [3]:
sampling_rules = {
"w_ConstructionType == 'ConstructionType_appt'": 10,
"e_sqm < 70": 10,
"w_Income > 13650": 10,
}
Part of the scenario development is to identified which section of the
population will adopt the new technology. The model defined this by a
sampling probability. This probability is initially define as a
uniform distribution (i.e. each individual on the sample has equal
probability of being selected). A scenario is defined by allocating new
sampling probabilities to a section of the population, by defining
sampling rules. The sampling rules are passes to the
query
function of a pandas DataFrame.
On the example above, all individuals with a Income larger of equal
to 180 000 Philippine Pesos are 30 times more likely to adopt the
technology that the rest of the population. The sampling probabilities
will sum up. This means that an individual with Income >= 180000 and
living on an urban area e_Urban == 'Urbanity_Urban' is 60 times more
likely to be selected (i.e. adopt a new technology) that other
individuals.
Initial sample data¶
In [4]:
import pandas as pd
file_name = "data/survey_Brussels_Electricity_Water_projected_dynamic_resampled_bias_1000_{}.csv"
sample_survey = pd.read_csv(file_name.format(2016), index_col=0)
sample_survey.head()
Out[4]:
| index | w_ConstructionType | w_Age | w_ConstructionYear | w_HHSize | e_sqm | w_Income | Water | Electricity | w | wf | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | ConstructionType_hous | Age_54 | ConstructionYear_1961 | HHSize_3 | 73.029173 | 13563.144795 | 35.477657 | 1424.293271 | 1512.187817 | 1512.194014 |
| 1 | 1 | ConstructionType_hous | Age_79 | ConstructionYear_1961 | HHSize_1 | 73.020808 | 13651.074297 | 26.794691 | 1403.605964 | 1512.187817 | 1512.188162 |
| 2 | 2 | ConstructionType_hous | Age_29 | ConstructionYear_1961 | HHSize_3 | 73.291210 | 13789.696751 | 37.695896 | 1054.339670 | 1512.187817 | 1512.178937 |
| 3 | 3 | ConstructionType_appt | Age_39 | ConstructionYear_1945 | HHSize_6 | 73.366936 | 13721.455861 | 48.304569 | 1335.381382 | 1512.187817 | 1512.183478 |
| 4 | 4 | ConstructionType_hous | Age_120 | ConstructionYear_1981 | HHSize_1 | 72.718209 | 13649.570622 | 26.428328 | 950.361211 | 1512.187817 | 1512.188263 |
The reduce_consumption module will use as input the samples created
by the MCMC algorithm and select specific sections of the population to
reduce their consumption levels.
The input data is the constructed proxy sample data. Depending on the simulation type (reweight/resample) the input data is a single file containing the weights for each simulation year (reweight) or individual samples for each simulation year (resample).
Most of the variables of the sample data is formatted as categorical
data. The sample data shown above presents individual records with the
predefine simulation variables as well as the computed Electricity
and Water consumption levels. The sample also contains two sets of
weights: w and wf. The w weights correspond to the weights
assign by the MCMC algorithm (uniform distributed) while the wf
(final weights) are the weights computed by the GREGWT algorithm. The
reduce_consumption function will use the wf weights for the
selection of individuals.
Define growth rates¶
In [5]:
Elec = transition_rate(0, 0.4, default_start=2016, default_end=2025)
Water = transition_rate(0, 0.2, default_start=2016, default_end=2025)
pr = transition_rate(0, 0.3, default_start=2016, default_end=2025)
With help of the growth_rate() function we define the efficiency
growth rate and the technology penetration rate. For the technology
penetration rate we define the start year to be equal to the benchmark
year (2016). The function will automatically include the necessary zeros
at the beginning of the growth rate vector. The function
plot_growth_rate() allow us to visualize the predefined
efficiency growth rates and the technology growth rates.
In [6]:
plot_transition_rate(
{"Technology penetration rate": pr,
"Electricity efficiency increase rate": Elec,
"Water efficiency increase rate": Water},
"scenario 1", start_year=2016, end_year=2025)
Reduce consumption¶
The actual modifications on the sample is performed by the
reduce_consumption() function. The function requires as input the
following parameters:
- A base file name for the samples (in case of implementing the resample method).
- The sample year (in this case generated within the loop via
range(2010, 2031). - The penetration rate for the sample year (iterated from vector
pr). - The predefined
sampling_rules. - A dictionary containing the efficiency rates for specific variables. (in this case for Electricity and Water).
- A name for the scenario.
In [7]:
for y, p, elec, water in zip(range(2016, 2026), pr, Elec, Water):
_ = reduce_consumption(
file_name,
y, p, sampling_rules,
{'Electricity':elec, 'Water':water},
scenario_name = "scenario 1")
00.00% both reduction; efficiency rate 00.00%; year 2016 and penetration rate 00.00
00.15% Electricity reduction; efficiency rate 04.44%; year 2017 and penetration rate 00.03
00.07% Water reduction; efficiency rate 02.22%; year 2017 and penetration rate 00.03
00.59% Electricity reduction; efficiency rate 08.89%; year 2018 and penetration rate 00.07
00.30% Water reduction; efficiency rate 04.44%; year 2018 and penetration rate 00.07
01.31% Electricity reduction; efficiency rate 13.33%; year 2019 and penetration rate 00.10
00.67% Water reduction; efficiency rate 06.67%; year 2019 and penetration rate 00.10
02.33% Electricity reduction; efficiency rate 17.78%; year 2020 and penetration rate 00.13
01.18% Water reduction; efficiency rate 08.89%; year 2020 and penetration rate 00.13
03.66% Electricity reduction; efficiency rate 22.22%; year 2021 and penetration rate 00.17
01.83% Water reduction; efficiency rate 11.11%; year 2021 and penetration rate 00.17
05.28% Electricity reduction; efficiency rate 26.67%; year 2022 and penetration rate 00.20
02.65% Water reduction; efficiency rate 13.33%; year 2022 and penetration rate 00.20
07.12% Electricity reduction; efficiency rate 31.11%; year 2023 and penetration rate 00.23
03.59% Water reduction; efficiency rate 15.56%; year 2023 and penetration rate 00.23
09.34% Electricity reduction; efficiency rate 35.56%; year 2024 and penetration rate 00.27
04.77% Water reduction; efficiency rate 17.78%; year 2024 and penetration rate 00.27
11.81% Electricity reduction; efficiency rate 40.00%; year 2025 and penetration rate 00.30
05.96% Water reduction; efficiency rate 20.00%; year 2025 and penetration rate 00.30
In [8]:
Elec = transition_rate(0.0, 0.5, default_start=2016, default_end=2025)
Water = transition_rate(0.0, 0.3, default_start=2016, default_end=2025)
pr = transition_rate(0.0, 0.6, default_start=2016, default_end=2025)
By modifying the growht rates we can create different scenarios.
In [9]:
plot_transition_rate(
{"Technology penetration rate": pr,
"Electricity efficiency increase rate": Elec,
"Water efficiency increase rate": Water},
"scenario 2", start_year=2016, end_year=2025)
In [10]:
for y, p, elec, water in zip(range(2016, 2026), pr, Elec, Water):
_ = reduce_consumption(
file_name,
y, p, sampling_rules,
{'Electricity':elec, 'Water':water},
scenario_name = "scenario 2")
00.00% both reduction; efficiency rate 00.00%; year 2016 and penetration rate 00.00
00.36% Electricity reduction; efficiency rate 05.56%; year 2017 and penetration rate 00.07
00.22% Water reduction; efficiency rate 03.33%; year 2017 and penetration rate 00.07
01.47% Electricity reduction; efficiency rate 11.11%; year 2018 and penetration rate 00.13
00.89% Water reduction; efficiency rate 06.67%; year 2018 and penetration rate 00.13
03.29% Electricity reduction; efficiency rate 16.67%; year 2019 and penetration rate 00.20
02.01% Water reduction; efficiency rate 10.00%; year 2019 and penetration rate 00.20
05.84% Electricity reduction; efficiency rate 22.22%; year 2020 and penetration rate 00.27
03.56% Water reduction; efficiency rate 13.33%; year 2020 and penetration rate 00.27
09.16% Electricity reduction; efficiency rate 27.78%; year 2021 and penetration rate 00.33
05.51% Water reduction; efficiency rate 16.67%; year 2021 and penetration rate 00.33
13.22% Electricity reduction; efficiency rate 33.33%; year 2022 and penetration rate 00.40
07.97% Water reduction; efficiency rate 20.00%; year 2022 and penetration rate 00.40
17.88% Electricity reduction; efficiency rate 38.89%; year 2023 and penetration rate 00.47
10.79% Water reduction; efficiency rate 23.33%; year 2023 and penetration rate 00.47
23.45% Electricity reduction; efficiency rate 44.44%; year 2024 and penetration rate 00.53
14.29% Water reduction; efficiency rate 26.67%; year 2024 and penetration rate 00.53
29.69% Electricity reduction; efficiency rate 50.00%; year 2025 and penetration rate 00.60
17.92% Water reduction; efficiency rate 30.00%; year 2025 and penetration rate 00.60