· · · Institutional Repository

Uncertainty manifests itself in almost every aspect of decision making. Adaptive and flexible policy design becomes crucial under uncertainty. An adaptive policy is designed to be flexible and can be adapted over time to changing circumstances and unforeseeable surprises. A crucial part of an adaptive policy is the monitoring system and associated pre-specified actions to be taken in response to how the future unfolds. However, the adaptive policymaking literature remains silent on how to design this monitoring system and how to specify appropriate values that will trigger the pre-specified responses. These trigger values have to be chosen such that the resulting adaptive plan is robust and flexible to surprises in the future. Actions should be neither triggered too early nor too late. One possible family of techniques for specifying triggers is optimization. Trigger values would then be the values that maximize the extent of goal achievement across a large ensemble of scenarios. This ensemble of scenarios is generated using Exploratory Modeling and Analysis. In this paper, we show how optimization can be useful for the specification of trigger values. A Genetic Algorithm is used because of its flexibility and efficiency in complex and irregular solution spaces. The proposed approach is illustrated for the transitions of the energy system towards a more sustainable functioning which requires effective dynamic adaptive policy design. The main aim of this paper is to show the contribution of optimization for adaptive policy design.

The treatment of uncertainty in the long-term planning of infrastructures in general and of mainports such as airports and seaports is a key challenge for decisionmakers. Moreover, these uncertainties have increased over the last decades due to changes in owner structure, changes in rules and regulations, and the ever increasing connectedness of the world. This dissertation explores how the treatment of uncertainties in airport planning can be improved. Currently, the treatment is limited to one or a few forecasts for the future. Such an approach limits the exploration of the multiplicity of futures to those that are judged to be most likely. However, if the last decade has taught is anything, then it is that the future will be substantially different from the one we are anticipating now. The implication of this for decisionmaking is that any plan or policy optimized for one or a few forecasts is likely to perform poorly. An alternative approach that is capable of handling the multiplicity of futures and accepts the limits on predictability is needed. Such an approach should result in a plans consist of time-urgent low regret options that can be taken immediately, while establishing a framework for guiding future actions. Thus the decisionmaker is able to adapt the plan to the way in which the future unfolds. This dissertation presents such a dynamic adaptive planning approach, tailors this approach to the specifics of airport planning, and provides computational evidence for the efficacy of plans that are designed utilizing this approach.

Airport Strategic Planning (ASP) focuses on the development of plans for the long-term development of an airport. The dominant approach for ASP is Airport Master Planning (AMP). The goal of AMP is to provide a detailed blueprint for how the airport should look in the future, and how it can get there. Since a Master Plan is a static detailed blueprint based on specific assumptions about the future, the plan performs poorly if the real future turns out to be different from the one assumed. With the recent dramatic changes occurring in the context in which an airport operates (e.g., low cost carriers, new types of aircraft, the liberalization and privatization of airlines and airports, fuel price developments, the European Emission Trading Scheme), the uncertainties airports face are bound to increase. Hence, there is a great need for finding new ways to deal with uncertainty in ASP. An alternative direction is to develop an adaptive approach that is flexible and over time can adapt to the changing conditions under which an airport most operate. Three adaptive alternatives to AMP have been discussed in the literature. This paper explores these three alternative approaches. Based on this, it concludes that these approaches are complementary and that it might be worthwhile to combine the three into a new, adaptive approach to ASP. A design that integrates the key ideas from the three alternative approaches is presented and illustrated with a case based on Amsterdam Airport Schiphol.

Intelligente Snelheid Assistent of ISA is de benaming voor een categorie in-vehicle systemen die bestuurder helpen om zich te houden aan de lokale snelheidslimiet (m.a.w. die er voor zorgen dat bestuurder niet te hard rijdt, of er zelfs voor zorgen dat de bestuurder nooit meer te hard kan rijden). De vele (veld) testen die in het verleden gedaan wijzen er allemaal op dat ISA een grote bijdrage zou kunnen leveren aan de verkeersveiligheid. Ondanks het grote potentieel (experts schatten dat een begrenzende ISA in Nederland jaarlijks meer dan 200 doden kan schelen) is ISA tot op de dag van vandaag niet geïmplementeerd. Om beter om te gaan met de onzekerheden die nog spelen rond de implementatie van ISA systemen en om uiteindelijk te komen tot duurzaam beleid met betrekking tot ISA implementatie wordt in het proefschrift een conceptuele aanpak gehanteerd die Adaptive Policymaking (APM) heet. APM is erop gericht om adaptief beleid te maken door vooraf na te denken over de onzekerheden die spelen en de manier waarop het beleid kan falen. Vervolgens wordt het beleid adaptief gemaakt door te bepalen op welke manier er gereageerd moet worden om de uiteindelijke beleidsdoelen te halen (beleid aanpassen, flankerend beleid maken, etc.). In dit proefschrift wordt onderzocht of APM geschikt is voor het ontwerpen van ISA implementatiebeleid voor Nederland. De resultaten laten zien dat ISA klaar is om geïmplementeerd te worden. Beleidsmakers zouden om kunnen gaan met de onzekerheden die spelen door op kleine schaal te beginnen met implementeren en als de tijd verstrijkt geleidelijk het beleid aan te passen aan de nieuwe kennis en omstandigheden (adaptief beleid). APM is een beleidsaanpak die daarbij zou kunnen helpen. De resultaten laten zien dat het ontwerpen van adaptief ISA implementatiebeleid met behulp van APM de kansen vergroot dat het ontworpen beleid, ondanks de onzekerheden die er nog zijn, de vooraf gedefinieerde beleidsdoelen haalt (in het geval van ISA een reductie in het aantal verkeersdoden, gewonden en ongevallen met schade). Desondanks geven de geraadpleegde experts ook aan dat ontwikkelde adaptieve ISA implementatiebeleid hoogstwaarschijnlijk strandt in de besluitvormingsfase (dus dat er geen beslissing over implementatie kan worden genomen). Dit komt omdat het expliciteren van de onzekerheden die nog spelen rondom de implementatie van ISA ertoe zal leiden dat politici helemaal geen beslissing kunnen of durven nemen.   Intelligent Speed Adaptation (ISA) is an in-vehicle system that supports the driver of a vehicle in complying with the local speed limit (In other words, that helps the driver to comply with the legal speed limit, or make sure the driver cannot drive faster than the speed limit). There is strong evidence that ISA has a great potential when it comes to contributing to traffic safety. However, ISA implementation is being delayed because of many uncertainties. Despite the large potential (it is estimated that a restricting ISA could save up to 200 lives per year in the Netherlands), ISA systems are not implemented yet. This dissertation focuses on the application of a conceptual approach called Adaptive Policy making (APM). APM is designed to develop policies that can be adapted over time, adaptive policies change as the external conditions change. In this dissertation the applicability of APM for ISA implementation in the Netherlands is researched, by operationalizing, applying, and evaluating the APM approach (for the case of ISA). The results show that ISA is ready to be implemented. Policymakers can deal with the uncertainties that still exist by starting to implement ISA on a small scale, and, as time proceeds, gradually adapt the ISA implementation policy to changing conditions. APM is an approach that could support that process. The results also show that designing adaptive ISA implementation policies with APM increases the chance that the policy will be a success, and reaches the predefined goals. (In case of ISA these goals would be: a reduction in the number of accidents, reduction of fatalities due to road accidents, etc.) Despite this experts also indicate that the developed adaptive ISA implementation policy will cause difficulties in the decision making process, and probably results in the fact that decision makers cannot take a decision at all. (It is indicated that making the uncertainties that surround ISA implementation explicit will be counter-productive for the decision making process).

The recent flu pandemic in 2009 caused a panic about the possible consequences due to deep uncertainty about an unknown virus. Overstock of vaccines or unnecessary social measures to be taken were all due to uncertainty. However, what should be the necessary actions to take in such deeply uncertain situation where there is no or very little information available? For uncertain and complex future, adaptivity and flexibility should be the main aim for designing robust policies. Here, we propose an iterative approach for designing adaptive and robust policies in the presence of deep uncertainty. A crucial part of this approach is the use of monitoring systems that provide the adaptivity and flexibility of the policy design. In the monitoring system, signposts to track specific information are defined. Specific values of these signposts are called triggers and they are triggered when pre-specified conditions occur in the system. The specification of trigger values is crucial for the policy performance but has not been studied in depth. Here, we use robust optimization to optimize the trigger values. This paper shows that our proposed approach with robust optimization improves policy design in deeply uncertain and complex situations where very little information is available.

Decision-making for flood defense while facing a considerable change in driving forces demands other methods than the traditional approach of forecasting and optimal policy selection. Exploratory modeling can be a candidate for helping adaptive policymaking to deal with the uncertainties that confront decision-makers. In adaptive policymaking where changes , policies are considered that respond to changes over time. This thesis addresses the question whether exploratory modeling is appropriate to support the design of flood defense strategies and in particular to assess the value of flexibility in such designs. The literature review of this thesis explores the concept of flexibility and shows that exploratory modeling as a method for handling uncertainty can contribute to system control as well as system resilience, and to scientific analysis as well as process management. The case study of this thesis demonstrates the application of exploratory modeling to flood defense strategies. It shows (1) how alternative strategies can be compared and evaluated while considering seven uncertain system parameters, (2) how the relative performance of strategies can be expressed as a regret value, and (3) how these values can be visualized to let decision-makers see how changes in parameters impact on their regret of decision-maker. These 3 goals reached by an exercise of exploratory modeling. Interpretation of computer model of a pre-investment analysis leads to selection of most robust strategy. Beside the help of visualization technique to see and compare performance of different strategies, a tool developed that counts and compare performance of strategies based on their regret value. I call this tool “Regret Frequency Table”. After finding the robust strategy which in this case was “Dike relocation” there was a need to see is this strategy flexible in terms of design to let future developments or not. Therefore based on interviews and gathered information, a decision tree has been draw to see if selected strategy can let future developments. It emerged that it can support couple of available measures in project in future. This conclusion leads to an extension of practice of exploratory modeling to apply adaptive policymaking approach to see the performance of selected strategy after taking the next step. This extension of exploratory modeling included the selection of a signpost for taking the next step. Signpost selection depends on (1) risk tolerance of decision maker (2) and model structure. The main conclusion in this step is that the sign post should be a part model; either a component of model or a parameter of model. This outcome leads to selection of a component and a parameter in model with two different values corresponding to different risk attitudes of decision maker. The results again fitted into regret frequency table and the outcome has been compared based on regret value of strategy before taking the next step. This allows seeing the value of flexibility relatively in terms of how many percent it improves the no regret values of a flexible strategy after implementing future developments. This practice lets decision maker to make his final decision about next step based on his attitude toward risk. The final outcome of research is a new approach of practicing exploratory modeling tailored for adaptive approach of policy making. I call this method “Exploratory Modeling for Adaptive Policies”. It borrows seven steps from exploratory modeling analysis (EMA) and adds three more steps to adjust the practice for adaptive policy making.

There is increasing interest in long-term plans that can adapt to changing situations under conditions of deep uncertainty. We argue that a sustainable plan should not only achieve economic, environmental, and social objectives, but should be robust and able to be adapted over time to (unforeseen) future conditions. Large numbers of papers dealing with robustness and adaptive plans have begun to appear, but the literature is fragmented. The papers appear in disparate journals, and deal with a wide variety of policy domains. This paper (1) describes and compares a family of related conceptual approaches to designing a sustainable plan, and (2) describes several computational tools supporting these approaches. The conceptual approaches all have their roots in an approach to longterm planning called Assumption-Based Planning. Guiding principles for the design of a sustainable adaptive plan are: explore a wide variety of relevant uncertainties, connect short-term targets to long-term goals over time, commit to short-term actions while keeping options open, and continuously monitor the world and take actions if necessary. A key computational tool across the conceptual approaches is a fast, simple (policy analysis) model that is used to make large numbers of runs, in order to explore the full range of uncertainties and to identify situations in which the plan would fail.