Hanni Lux

The Spherical Tokamak for Energy Production (STEP) programme is a world-leading fusion power plant programme that has embedded a cost conscience in its design from the early phases. This firmly addresses the attitude of cost complacency of which many major infrastructure projects have historically been accused. While a detailed and highly accurate whole life cycle cost analysis is not possible, or even valuable, during the conceptual design stage, this early design phase is still the most critical programme phase where a focus on costs can drive longer term reductions and impact whole life cycle costs at the high level. Consequently, appropriate estimating methods for these early-stage designs and lessons learned from other industries are used to inform design decisions and ensure cost is part of the overall option analysis. Hence, while the overall programme cost estimate is too immature to be a reliable indicator for the final programme costs, significant effort has been undertaken to understand the major cost drivers and take action to make the STEP design as cost-effective as possible. This article is part of the theme issue ‘Delivering Fusion Energy – The Spherical Tokamak for Energy Production (STEP)’.

Developments in fusion energy technology and the aspiration to build and run commercial fusion energy power plants have seen the commencement of numerous publicly and privately funded projects in recent years. Megaprojects, like fusion power plants, by their very nature are inherently complex and risky, therefore, providing a robust cost estimate in the early stages is challenging. In the fusion sector, this is amplified by the fact that very little data exist on which to base an estimate due to the novel nature of the materials and technologies involved. A common phenomenon during the estimating phase of a new project is the concept of optimism bias (OB), where underestimation of time, cost, and risk can cause impacts on the baseline cost and schedule, leading to significant over-runs during the project lifetime. Reference class forecasting (RCF) is a method used to mitigate against these factors, based on using actual performance data in a reference class of comparable projects to improve forecasting accuracy. This article will discuss RCF, how it has been used in recent megaprojects, and how it is intended to be used in the Spherical Tokamak for Energy Production (STEP) program to provide a full program cost model for a prototype fusion energy plant.

In the early stages of any large engineering/construction project, a standard way of estimating the cost is a top–down approach, comparing the design with similar existing projects with known costs. As many international fusion power plant projects are currently in their concept design phase or early engineering design phase any whole program cost estimates can be expected to transition from the top–down to more bottom–up cost approaches. While multiple publications of top–down cost estimates for fusion power plants exist in the literature see and some isolated bottom–up cost models for components have been presented, Bonavolonta et al., 2023 are not aware of any published work that demonstrates how to transition from the top–down to the bottom–up cost estimates and how these are used within the design of the fusion plant. The Spherical Tokamak for Energy Production (STEP) program has just completed its concept design phase where top–down parametric costing has been used in addition to comparable costing to generate broad estimates. It consciously uses these estimates throughout its design phase to ensure a cost-optimized design. To achieve a smooth transition, the total project cost is split up according to a unique cost breakdown structure (CBS) (such as the one defined in STARFIRE). As the STEP design matures and becomes more defined, it is beneficial to replace sections of the estimate with more detailed bottom–up estimates as the design detail is developed. The Excel add-in software CALC4XL supports this approach, by enabling the whole project cost to be split into sections, estimated in top–down or bottom–up methods (incorporating costs of materials and manufacturing). CALC4XL can switch between the two calculation methods for an easy comparison and each section is assigned an uncertainty class to calculate a high/low view of the costs. STEP uses a special bespoke version of the software, adapted for the particular needs of the unique STEP project, making it flexible to any level of design detail. The development and accuracy level of the project estimate can be tracked and plotted via this software, where it can easily be seen how the estimates are generated. A quality control procedure has been employed to make sure all cost changes can be attributed accurately. This system could be employed in many other complex mega-projects, especially fusion.

Being a novel technology, estimating costs for fusion power plants comes with large uncertainties. Cost uncertainties in prototypes arise from various sources and reduce with program maturity (including the selection of a site), technical and design maturity, as well as the maturity of the commercial strategy (e.g., make versus buy decisions and partnering decisions). In the early design phases, where not much detail is fixed, the cost is typically evaluated via analogy/comparator to existing technologies, which are typically associated with high uncertainties due to the innovative nature of fusion technology. Early cost estimates for fusion power plants are broken down using a top-level cost breakdown structure as this allows for cost uncertainty evaluation over the various fusion components. The STEP prototype fusion power plant is nearing the end of conceptual design with significant uncertainties but seeks to follow best practices to effectively manage costs. Best practice in cost estimating involves evaluating the base cost estimate, corresponding uncertainties, and project risks and correcting for optimism bias as defined by the Infrastructure and Project Authority (IPA). The overall objective is to illustrate the importance of uncertainty definition within fusion prototype/demonstration power plants and their impact on program costs.

For mega-projects like fusion power plants, modularity is a key enabler to cost and schedule efficiency. One way of achieving more modularity is aiming for higher numbers of smaller fusion reactors. Previous work has demonstrated that the levelized cost of electricity (LCOE) of commercial magnetic confinement fusion power plants falls at a decreasing rate with increasing net electric power. Furthermore, net electric power increases more rapidly than size/cost. This is because as fusion power increases the proportion of energy being exported as net electric power plateaus but the size of the plant required increases linearly. Increases in plant size increase upfront capital costs and project complexity. Therefore, there is an optimal design point beyond which any increases in net electric power continue to increase the project cost and complexity but deliver only marginal gains in LCOE. This helps identify a sweet spot between anticipated, better economy of size (cost per unit being smaller at larger unit sizes), and economy of scale (cost per unit being smaller at a higher scale of production).

With the ITER project entering the second half of the construction phase and various national and international plans for fusion prototype power plants being in early to advanced stages, fusion has entered the delivery era. With the urgency of the climate crisis being at the forefront of government policy agendas, it is important to focus on the commercialization of fusion power plants to support decarbonization commitments. This work argues that the cost optimization of fusion as well as the development of a compelling value proposition enhances the contribution that fusion can bring to target energy markets. In particular, it focuses on how both aspects are being considered for the Spherical Tokamak for Energy Production (STEP) program (Wilson et al., 2020) which is currently developing both the conceptual design for a prototype power plant as well as a pathway to commercialization on its mission to “Deliver a U.K. prototype fusion energy plant, targeting 2040, and a path to commercial viability of fusion.”

During the pre-conceptual design phase of fusion devices such as the European demonstration fusion power plant (DEMO), systems codes provide a fast evaluation of optimal design points and highlight high impact areas. However, determining or evaluating a design point at such an early stage comes with uncertainties in many of the design parameters. These uncertainties are both associated with the physics as well as the engineering basis of the European DEMO design. The work applies an uncertainty quantification analysis to the 2017 pulsed European DEMO design using the PROCESS systems code. It assumes that DEMO will be built as suggested by the baseline and explores what implications the currently known physics and engineering uncertainties have on the expected performance parameters (net electric output and pulse length), while optimising the fusion gain Q. A more detailed single parameter analysis clearly identifies high impact parameters. This confirms previous investigations as well as revealing new areas that warrant deeper investigation in particular in the technology area.

In systems modelling for fusion power plants, it is essential to robustly predict the performance of a given machine design (including its respective operating scenario). One measure of machine performance is the energy confinement time τ E that is typically predicted from experimentally derived confinement scaling laws (e.g. IPB98(y,2)). However, the conventionally used scaling laws have been derived for ITER which—unlike a fusion power plant—will not have significant radiation inside the separatrix. In the absence of a new high core radiation relevant confinement scaling, we propose an ad hoc correction to the loss power P L used in the ITER confinement scaling and the calculation of the stored energy W t h by the radiation losses from the ‘core’ of the plasma P r a d , c o r e . Using detailed ASTRA / TGLF simulations, we find that an appropriate definition of P r a d , c o r e is given by 60% of all radiative losses inside a normalised minor radius ρ c o r e = 0.75 . We consider this an improvement for current design predictions, but it is far from an ideal solution. We therefore encourage more detailed experimental and theoretical work on this issue.