Designing a fusion power plant requires navigating a high-dimensional space of physics, engineering, and economic trade-offs. This talk presents two complementary approaches to exploring that space, both built on the FUSE (Fusion Synthesis Engine) integrated modeling framework.
In the first part, I will discuss a comparative optimization study of positive and negative triangularity tokamak power plants (Slendebroek et al., Nucl. Fusion 66, 026032, 2026). Using multi-objective optimization over coupled physics and engineering models, we identify where negative triangularity configurations are competitive with their positive triangularity counterparts, and where they are not, with particular attention to confinement assumptions, divertor constraints, and capital cost drivers.
In the second part, I will present a large-scale parametric study of ITER-like plasmas spanning roughly 1.5 million cases across 13 input dimensions, using the TGLF SAT1-EM and SAT3-EM saturation rules for turbulent transport. This database-driven approach complements targeted optimization by revealing the global structure of the design space, identifying sensitivities and degeneracies that local studies can miss, and providing training data for surrogate models.
Together, these studies illustrate how integrated modeling at scale can shift fusion device design from point-design iteration toward systematic exploration, and I will close with thoughts on how agentic and autonomous workflows may extend this further.