Mixed-Integer Energy System Optimization Models for Germany: Seven Spatial Aggregations with Monte Carlo–Sampled UC, Dispatch, and Expansion Scenarios

This dataset provides seven sets of mixed-integer linear energy system optimization models (ESOMs) for the German power system, generated with the REMix framework. All models are formulated as mixed-integer programs (MIPs) that jointly optimize unit commitment, dispatch, and capacity expansion of renewable technologies, natural-gas-fired plants, storage units, and transmission lines across 8,760 hourly time steps (= 1 year; exception for miso_244k). Renewable generation from wind and solar is modeled with continuous variables. The full-resolution system comprises 488 nodes (477 within Germany and 11 neighboring countries) and 643 lines. The six aggregation levels (miso_244k has reduced horizon compared to miso_47M) reduce this spatial detail to varying degrees, resulting in models of different sizes (see Table 2 for details) and computational difficulty, ranging from large-scale MIP instances to more compact variants. Together, these datasets provide a scalable benchmark suite for energy system planning under uncertainty. Each set has been obtained using a different spatial aggregation of the complete system (see Table 1 for details). Each set includes multiple Monte Carlo–sampled instances to account for uncertainties in weather data and techno-economic parameters. The models follow the mathematical description of the expAll models in Wetzel et al. (2025) for the underlying LP part. Additionally, the following integer constraints are enforced: - Units under the column “discrete units” are modeled with integer variables enforcing integral expansion decisions. Lignite, coal, and OCGT units are included but cannot be expanded. - The unit commitment formulation enforces partial load constraints with two activities for CCGT, Coal, Lignite. Minimum up- and downtime constraints are enforced for CCGT, Coal, and Lignite. - Lithium-ion storage and pumped-storage power plants are modeled with continuous variables and do not involve discrete expansion decisions. Investment structure.Capacity expansion decisions are modeled using static investment variables defined for the full optimization horizon. Any added capacity is assumed to be available for the entire modeled year; the model does not include commissioning times or multi-period investment stages. The transmission topology is fixed and can only be expanded in capacity, technologies are not bound to their initial topology. Transmission modeling.AC transmission lines are represented using a DC load-flow approximation with a B–θ (angle) formulation. HVDC lines are modeled using a capacity-constrained transport representation. No N-1 contingencies or security-constrained unit-commitment features are included in these instances. Additional documentation on the remix modeling can be found under the REMix homepage. set discrete units spatial aggregation (nodes) number of lines number of nodes miso_244k* CCGT, Coal, Lignite, HVAC, HVDC Germany: 10; NRW** 32; Neighbors: 1 78 43 miso_2M CCGT, HVAC, HVDC Germany: 1; Neighbors: 9 9 10 miso_4M CCGT, Coal, Lignite, HVAC, HVDC Germany: 2; Neighbors: 1 3 3 miso_47M CCGT, Coal, Lignite, HVAC, HVDC Germany: 10; NRW: 32; Neighbors: 1 78 43 miso_82M CCGT, Coal, Lignite, HVAC, HVDC NRW: 130; Neighbors: 1 191 131 remix_105M CCGT, Coal, Lignite, HVAC, HVDC Germany: 10; NRW: 88; Neighbors: 11 167 109 remix_243M CCGT, Coal, Lignite, HVAC, HVDC Germany: 6; NRW: 253; Neighbors: 11 343 270 Table 1: model information.* miso_244k models only a period of 48 hours.** NRW = North Rhine-Westphalia (a German state). set number of instances number of constraints number of variables number of binaries/integers number of non-zeros miso_244k 100 73,518 67,386 5,445 244,293 miso_2M 100 928,790 880,606 35,052 2,829,975 miso_4M 96 1,279,124 1,103,903 157,692 4,590,593 miso_47M 100 16,313,022 13,597,290 972,477 47,310,667 miso_82M 20 24,698,935 24,356,961 867,465 82,773,747 remix_105M 100 33,194,746 28,455,474 3,364,138 105,960,838 remix_243M 100 77,498,047 66,714,360 7,726,975 243,495,063 Table 2: model sizes.