Mapping the near-optimal design space of off-grid green hydrogen systems using reduced linear programs
Year: 2026
Type: preprint
Abstract: Near-optimal design regions identify feasible equipment alternatives as project requirements change. ORACLE and G-ORACLE repeatedly solve the full operating model to identify the nearest feasible design, an approach that can be computationally costly for hourly planning models. This study enables distance-based exploration of near-optimal green hydrogen design regions incorporating load-dependent electrolyzer efficiencies and integer wind turbine counts. For a concave piecewise-linear production model, we establish conditions under which the integer sections of a convex continuous representation exactly recover the original budget-feasible integer design set. Annual production and cyclic storage cuts eliminate hourly operating variables from the nearest-design linear program (LP), whereas vertex LPs guide maximum-distance refinement. Independent exploration of integer sections yields geometric error bounds and finite termination under the stated assumptions. Case studies in Jeju and Texas use three complete weather years to explore regions within 1% and 2% of the minimum levelized hydrogen cost. For the joint continuous design region, the proposed method is 625-1,453 times faster than G-ORACLE and 1,023-5,076 times faster than ORACLE. In both regions, designs with three fewer wind turbines than the minimum-cost design met the same demand at a cost premium below 1%, although the remaining capacity choices are more limited. The explored design regions enable rapid comparison of feasible equipment combinations and their costs as site and procurement conditions change.
Authors: Kim, Sunwoo, Park, Joungho
Institution: Korea Institute of Energy Research
Language: English
Subfield: Energy Engineering and Power Technology
Field: Energy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236107330
