Caner Hamarat, Jan Kwakkel, Erik Pruyt
Accepted with minor revisions in Technological Forecasting & Social Change
Adaptive policymaking is an approach for developing policies that can adapt over time to surprises caused by deep uncertainty. There is a strong need for novel methods and techniques to support adaptive policymaking. Exploratory Modeling and Analysis (EMA) is a methodology for analyzing dynamic and complex systems and supporting long-term decision making under deep uncertainty. EMA is a good candidate to be employed for supporting adaptive policymaking under uncertainty. In this paper, we propose an iterative approach using EMA for developing adaptive policies. The proposed approach is demonstrated through a case about a structural and systemic societal transformation, namely energy transitions towards a sustainable future. The case illustrates how the performance of a policy can be explored across an ensemble of plausible futures and be improved in light of the identification of troublesome and promising futures, until a satisfying dynamic adaptive policy emerges. The proposed approach can be efficiently used for developing dynamic adaptive policies for grand societal challenges. test
In this paper, the proposed approach, Computer Aided Dynamic Adaptive Policy Design (CADAPD), has been illustrated through a case about the structural and systemic transformation of the energy generation systems toward a more sustainable future. For this, we used a System Dynamics models of the energy transition, implemented in Vensim. The essential point of this approach is that it helps design and test performance of adaptive policies. Our analysis shows that CADAPD can be used to develop long-term, adaptive and robust policies for energy transitions toward sustainability.
The initial analysis starts without introducing any policy. As a next step, we have analyzed the initial results and designed a basic policy that is focused on promoting the decommissioning of the dominant non-renewable technology. This initial policy is called basic policy. However, the basic policy was not enough to promote transition toward sustainable future. According to the analysis on new results, there appeared two opportunities that can be benefited. One was related with dominant renewable technology and the other was related with economic development and costs. Thus, two adaptive strategies have been designed and added on top of the basic policy. This new resulting policy was called adaptive policy.
For a sustainable future, the renewable energies should be promoted and there should be a transition toward a more sustainable energy system. For this reason, we have used the total fraction of the renewable energies over the whole energy system as a performance indicator. For each policy, a wide range of uncertainties were explored and more than 9,000 runs were generated. The left side of the figure shows the envelope of the fraction of sustainable technologies over a time horizon of 100 years for each policy with a different color. Envelope can be understood as the upper and lower bounds for the fraction over all the runs (>9000). On the right side of the figure, the density estimates for the end states of the renewable fraction is shown. Each policy is represented with a different color. This method helps us understand where the runs are mainly concentrated.
Legend: color codes for different policies
| No Policy | Basic Policy | Adaptive Policy |
|---|---|---|
| Red line | Blue line | Green line |
As shown on the right side (gaussian kernel density estimates), there is a considerable improvement by introducing the basic policy. The red line stays below the other two policies. It can be concluded that the basic policy is successful for promoting the renewable transition. By looking at density estimates, adaptive policy seem not to be that better than the basic policy. However, it can be seen from the left side that adaptive policy performs better than basic policy in the early years. Although the effect of the adaptive policy seems to be neutralized over years, the reasons behind this behaviour is a possible research topic.
'''
Created on 3 Jan 2012
@author: chamarat, jhkwakkel
'''
import matplotlib.pyplot as plt
from expWorkbench import load_results
from analysis.graphs import envelopes
import analysis.graphs as graphs
# force matplotlib to use tight layout
# see http://matplotlib.sourceforge.net/users/tight_layout_guide.html
# for details
graphs.tight = True
# get the data
results = load_results(r'.\data\TFSC_policies.cPickle')
# make an envelope
fig = envelopes(results, column='policy', fill=True, legend=False)
# set the size of the figure to look reasonable nice
fig.set_size_inches(9,3)
# save figure
plt.savefig("policycomparison.png", dpi=75)