1
MODELLING UNCERTAINTY OF HOUSEHOLD DECISION-MAKING IN SMART GRID APPLIANCES ADOPTIONINCLUDING HOUSEHOLD BEHAVIOURAL UNCERTAINTY IN THE IDENTIFICATION OF POLICIES TO SUPPORT SMART GRID APPLIANCES ADOPTION BY USING AGENT-BASED MODELLING AND THE SCENARIO DISCOVERY TECHNIQUE
Tristan de [email protected]
m
Yvonne BoerakkerYvonne.Boerakker@dnvgl
.com
2
CONTENT
1. Project overview
2. Uncertainty
3. Types of uncertainties
4. Scenario-discovery
5. Added-value of including uncertainties
6. Conclusions
3
1. PROJECT OVERVIEW
Research question Which directions for policy can stimulate the adoption of smart grid
appliances to increase the capacity for demand response in city districts?
Research output Set of directions for policy to support the adoption of smart grid appliances
How Creation of a simulation model and identification of directions for policy
to increase adoption
4
1. PROJECT OVERVIEW
Agent-based modelling
A simulation method where the study of the choices and behaviors of ‘agents’ are central: An agent is “a thing which does things to other things”
(Rohilla Shalizi, 2006)
Why? Easier to include various types of heterogeneities (differences
in behaviour and characteristics between agents
Various types of interaction networks between households can be easily included
5
1. PROJECT OVERVIEW
The model
6
2. UNCERTAINTY
Uncertainty: any type of aberration from utter certainty About the future state of a system
But also about the modeller’s lack of knowledge about a system’s structure and properties
Deep uncertainty is the lack of consensus among stakeholders about : the proper conceptualisation of a system
the probability distributions used to represent uncertainty in the model
the valuation of the desirability of various outcomes
(Lempert, et al., 2003)
7
2. UNCERTAINTY
Why should they be included? Most policies fail because of the omission of uncertainties in policy-making
a policy that was effective under a particular scenario appears to fail in reality due the erroneous understanding of a system and/or due to its evolution in an unexpected direction
Why are they especially relevant in the case of the adoption of smart grid appliances? Mainly since the decision-making of individuals in adopting (and using) smart grid appliances and
the transfer of information:
is extremely difficult to understand
might change over time
can not be represented by one single conceptualisation, since this conceptualisation might not be accurate for all individuals
8
3. TYPES OF UNCERTAINTIES
Parameter uncertainty Incertitude about parameters describing the behaviour and the decision performed by individuals, and the
evolution of the environment in which individuals evolve.
Structural uncertainty Incertitude about how to represent a certain system, in this case the decision-making of households, in a
simulation model.
Heterogeneity Incertitude about the degree to which individuals differ in characteristics and preferences.
Stochastic uncertainty Set of probabilistic distributions used to run the simulation model, for example those determining the order
of the execution of actions by the individuals in the model (proper to discrete models)
Briggs, et al, (2012)
9
4. SCENARIO-DISCOVERY
Combination of:
EMA (Exploratory Modelling and Analysis) Inclusion of parametric uncertainties in a
model: ‘instead of assuming an 5 interactions per household per month in the model, let’s assume a range of [3-7] interactions’
Inclusion of structural uncertainties in a model: ‘instead of assuming that one structural conceptualisation is best, let’s try different ones’
PRIM (Patient Rule Induction Method) Find out which variations of input
parameters or structures lead to a specific desired or undesired model output
10
5. ADDED-VALUE OF INCLUDING UNCERTAINTIES
Creation of policies that are: robust to various future developments of the system (e.g. the future of the system ‘adoption of
smart grid appliances by households’
robust to various imperfections about the understanding and representation of that system
wider acceptation of model outcomes among various stakeholders
11
5. ADDED-VALUE OF INCLUDING UNCERTAINTIES
An example with parameter uncertainty
Adoption curve without Policy – single run
Adoption curve with purchase subsidy – single run
Adoption curve with purchase subsidy – with EMA
One policy is identified but the effectiveness to support smart grid appliance adoption appears to be limited in the model.
5. ADDED-VALUE OF INCLUDING UNCERTAINTIES
12
An example with parameter uncertainty
1. Low social value experienced by early adopters in smart grid appliance adoption
2. High degree of suspiciousness experienced by early adopters towards the reliability of information transferred between household
3. High degree of suspiciousness experienced by early majority population towards the reliability of information transferred between household
4. Slow decrease of smart grid appliance prices
5. Long growth phase (duration before the complexity of purchasing and using smart grid appliances is acceptable for all type of households)
6. Low amount of interactions between early majority population and early adopters
7. Low amount of interactions between late majority population and early adopters
Directions for policy Policy examplesEncourage communication between innovators and early adopters
Nomination of product ambassadors, creation of consumer groups
Promotion of smart grid appliances to early adopters
Nomination of product ambassadors, creation of consumer groups
Decrease adoption costs Purchase subsidyWhen smart grid appliances may become interesting for early majority population and later adopters, reinvent the product
Redesign the product to underline ease of use, savings that can be made to change product perception
Make product usage and the added value of owning it visible to others
Redesign the product to make it visible to others
After the usage of scenario discovery:
Scenarios identified: Directions for policy applied: Resulting adoption curves:
13
6. CONCLUSIONS
In the modelling of systems with high level of uncertainty, the use of the scenario discovery allows for: the identification of more robust policies
possibilities to achieve a wider commitment of various stakeholders to model outcomes
This is particularly true in the case of the modelling of the adoption of smart grid appliances since the behaviour of humans is fundamentally uncertain
Limitations: the policies identified still only say something about how to increase adoption in the model and not
in reality. An extended reflection about their effectiveness in real world is critical.
the inclusion of a large amount of uncertainties strongly increases the computation costs of model analysis
Uncertainties should hence only be added if they cannot be rejected during the verification and validation phase of the model
14
6. CONCLUSIONS
For more information about the project: http://repository.tudelft.nl/view/ir/uuid%3A07b27819-1e34-4a36-848b-29858f5139be
Or for any other questions: Tristan de Wildt: [email protected]
Yvonne Boerakker: [email protected]
15
LITERATURE
Bankes, S., 1993. Exploratory modeling for policy analysis. Operations Research, Volume 4, pp. 435-449.
Briggs, A., Weinstein, M. & Fenwick, E., 2012. Model Parameter Estimation and Uncertainty: A Report of the ISPOR-SMDM Modeling Good Research Practices Task Force-6. Value in Health, Volume 15, pp. 835-842.
Dam, K. v., Nikolic, I. & Lukszo, Z., 2013. Agent-Based Modelling of Socio-Technical Systems. 1 ed. s.l.:Springer.
Kwakkel, J. H., Auping, W. L. & Pruyt, E., 2013. Dynamic scenario discovery under deep uncertainty: The future of copper. Technological Forecasting & Social Change, Volume 80, pp. 789-800.
Kwakkel, J. & Pruyt, E., 2012. Exploratory Modeling and Analysis, an approach for model-based foresight under deep uncertainty. Technological Forecasting & Social Change, 80(3), pp. 419-431.
Lempert, R., Popper, S. & Bankes, S., 2003. Shaping the next one hundred years: new methods for quantitative, Long-Term Policy Analysis, Santa Monica, California: RAND.
Shalizi, C., 2006. Methods and techniques of complex systems science. In: T. Deisboeck & J. Yasha Kresh, eds. Complex Systems Science in Biomedicine. New York: Springer US, pp. 33-114.
Top Related