A decision analysis framework for the U.S. nuclear fuel cycle

Abstract

If we are willing to pay a premium, we may be able to mitigate some of the long-lasting impacts of nuclear waste. Deciding how to navigate this tradeoff, between cost and waste, is a central challenge for stewards of nuclear power. It is made more difficult by uncertainties that characterize the global future of nuclear electricity generation. The recent increase in concern about climate change has prompted U.S. policymakers to back strategies favorable toward nuclear power, so much so that some experts see a "nuclear renaissance" on the horizon. Whether such a renaissance will come to pass, involving the construction of a vast new fleet of nuclear plants, is unclear - especially in light of the March 2011 nuclear accident at the Fukushima Daiichi reactors in Japan. Even more unclear is what should be done with the commercial U.S. nuclear waste, given an array of technical options and a large amount of uncertainty about how much waste will ultimately need to be managed. This study introduces a framework for analysis of strategies to evolve the nuclear fuel cycle which may be helpful in analyzing decision problems for similarly complex, long-lived technical infrastructure systems. The framework consists of a system dynamics simulation coupled with a decision analysis model. The system dynamics code is developed specifically for this study to be simple, fast-running, and also to echo the results of many previous nuclear fuel cycle simulations in demonstrating how various technical options impact important parameters (like uranium consumed, waste generated, etc.). Code results are benchmarked to more complex fuel cycle simulations for the parameters relevant to the decision space. The decision analysis model takes information from the simulation and makes it useful to policymakers, by allowing the explicit analysis of desirable decision pathways under uncertainty, and also considering tradeoffs among system goals. The framework is applied to three nuclear systems, the light-water reactor (LWR) once through fuel cycle, which represents the status quo, an advanced, traditional, plutonium-fed self sustaining fast reactor fuel cycle, and a fast reactor fuel cycle for which initial fast reactor cores are composed of enriched uranium rather than recycled LWR fuel. Fast reactors are highly likely to cost more than LWRs, but they can produce electricity from some of the elements that most plague the long-term management of a nuclear waste repository. A value function compares how these options fare under different scenarios, incorporating system-wide costs and the system waste burden as the two attributes in the function. The primary result is that the best strategy, under a strong preference for eliminating LWR spent nuclear fuel waste, consists of building a few traditional fast reactors now, and then building a full fleet more rapidly later in the century. This allows both for a significant amount of waste mitigation compared to an all-LWR fuel cycle, and for the costs associated with the more expensive fast reactor technology to be incurred primarily later in the century. On the other hand, if cost is the main consideration, the framework advises moving forward with the once-through LWR fuel cycle and avoiding fast reactors altogether, or at least until later in the century. These results are examined from a traditional decision analysis perspective, and then from one that departs somewhat from the assumption of a fully powerful decision maker. In reality, a government decision maker can only offer incentives to industry in order to induce a strategy change. Changing the decision model to reflect this reality causes the framework to more strongly advise moving forward with traditional fast reactors. This occurs because any single attempt at offering incentives to industry might be unsuccessful, and thus prevent a waste concerned government from achieving any significant mitigation. The most important contribution of the methodology is its ability to illuminate which parameters represent strong drivers of system decisions. Preferences across competing attributes are always important: in general, if decision maker preferences for reducing cost vs. waste were to shift significantly, the framework would show a change in the desirable decision strategy. Decision results are not very sensitive, on the other hand, to the rate of nuclear power growth or to the cost of fast reactor technology. A second contribution comes from the initial foray into studying a more complex decision maker perspective, and shows how a different view can complement results using the traditional decision analysis assumption of an "ideal" decision maker. Ultimately, the system dynamics/decision analysis framework presented here helps identify desirable pathways for complex system evolution, identifies factors that bear strongly on decisions and which are deserving of more study, and begins to show how strategy implementation can be considered within the framework in order to further improve decision-making.