Procedure

Maon simulates electricity wholesale markets by deriving the following three key outputs.

• Short-term equilibrium: Determining dispatch schedules and wholesale prices using generation, grid, and consumption capacities as inputs through single and serial runs.
• Uncertainty distribution: Calculating uncertainties in short-term equilibria, along with associated outcomes like facility-specific costs and revenues with the help of Monte Carlo simulations.
• Long-term equilibrium: Deriving investment decisions in generation, grid, and consumption facilities based on anticipated prices, volumes, and emerging technologies via investment simulations.

All derivations rely on the same market simulation procedure. Single runs execute it once. Serial runs apply it to a sequence of years. Monte Carlo runs repeat it with varying inputs to generate output distributions. Investment runs use it iteratively to derive economically viable investment decisions.

Single

The procedure for single runs is applied at the single dispatch page by clicking on Run simulation.

Overview

Upfront input data is processed (see preprocessing).

The three-stage optimization procedure is illustrated in the following figure. First, on-off dispatch decisions are neglected, and nonlinear equations are linearized to solve the overall linear optimization problem. Second, unit states are fixed at the start and end of sub-time ranges (e.g., daily), and these subproblems are solved successively. Third, on-off dispatch decisions are fixed, and the overall optimization problem is linearized and optimized. Numerical issues are handled internally through automatic optimization solver restarts and recalibrations.

After the simulation procedure, the raw results like unit commitments and price estimates are processed for further usage (see postprocessing).

Procedure steps

The overall mixed-integer linear problem is solved through three successive procedure steps, each building on the previous one.

Linear optimization

In the first procedure step, on-off dispatch decisions are neglected and nonlinear equations are linearized. The resulting total linear optimization problem is solved. The barrier method is used by default for solving. The optimization solving method can be set for linear problems in the project configuration with the parameter solver_optimizer_linear_problem.

Mixed-integer optimization

After the total linear optimization problem is solved in the first step, determined unit states are fixed each simulation sub time range begin and end. The subproblems (e.g., daily) are solved successively via a branch-and-cut method by default.

The following image depicts the mixed-integer approach and its hand-overs used as presets in the subsequent simulations.

The total time horizon is set in the scenario settings (default is 8760 hours). The sub time range is configurable in the project settings (default is 24 hours). Longer sub-time ranges require more processing time, and the increase in time is disproportionate due to the combinatorial complexity of the mixed-integer optimization problem.

The mixed-integer problem solving algorithm can be set in the project configuration with the parameter solver_optimizer_mixed_integer_problem.

For large exchange combinations, such as those involving numerous bidding zones or nodal pricing, system decomposition can be applied in addition to time decomposition in 90_grid_bidding_zones.csv. This approach fixes groups of bidding zones or nodes with their net balances (exchange sums), effectively reducing the mixed-integer problem complexity to practical levels.

Reoptimization

After the mixed-integer optimization procedure step, derived on-off dispatch decisions are fixed. Lastly, based on the derived operation points, the global optimization problem is linearized and solved. The solving algorithm equals the linear solving method (see linear optimization).

Submodels

The submodels considered vary depending on the selected steps in the simulation procedure. Each subsequent step includes the features of the previous step. Detailed information about submodels can be found in the input data definition.

Inflexible demand

Inflexible supply

Flexible demand-side response

Flexible battery

Flexible hydro

Flexible thermal

Flexible grid

Fast mode

The total runtime for the three-step procedure is approximately 12 hours (for ENTSO-E region, 8760 hours, and 50000 components). Maon offers a fast mode that significantly reduces this time while still respecting all market equilibrium constraints, achieved by aggregating components. When battery, DSR, thermal, hydro, and grid components are aggregated to the maximum levels in all bidding zones, the procedure time drops to under five minutes. The higher the aggregation level, the faster the simulation. Intermediate aggregation levels allow choosing a specific trade-off between accuracy and speed.

Original and aggregated model data can be combined, if needed. For example, the aggregation method can be applied to all bidding zones except those in the focus region of the analysis. The aggregation level can be set in 90_grid_bidding_zones.csv. To further reduce simulation runtime, certain restrictions and functionalities (e.g., minimum down times, operation times, ramping constraints, or frequency reserve provision) can be disabled in the project configuration.

Serial

The procedure for serial runs is applied at the serial dispatch page by click on Run simulation.

Parametrizing multiple years can be done through annual scenarios and single runs (see single run). However, gapless simulations over longer periods (e.g., 2020 to 2040) require considerable effort, especially for many years (e.g., 21 years). To minimize this effort, serial runs can be applied. This approach utilizes the inheritance.

Serial runs require one initial scenario (e.g., 2020) and a derived scenario for a later year (e.g., 2040). The procedure automatically generates annual scenarios for each year in between (e.g., from 2021 to 2039), significantly reducing parametrization efforts. The following figure illustrates how the intermediate scenarios are created.

The serial run procedure is available in derived scenarios, where users can select the years in between and manage the runs. By default, the procedure starts with all years between the base and derived scenario.

Monte Carlo

The procedure for Monte Carlo runs is applied at the Monte Carlo simulation page by click on Run simulation. The procedure is available in base scenarios, where users can select the availability and climate years as well as manage the runs.

Parametrization

Parametrizing Monte Carlo runs can be done with the same data-sets as for single runs. However, the creation of single combinations of, for example, climate and outage years can require considerable effort, especially for many combinations (e.g., 1000 samples). To minimize this effort, Monte Carlo runs can be applied. This approach utilizes the inheritance.

Sampling

Monte Carlo runs require one base scenario (e.g., 2050) and at least one therefrom derived scenario for a climate year (e.g., 2012). The procedure automatically creates availability scenarios in the preprocessing as in single runs. Further, it combines climate and availability scenarios for all selected years, significantly reducing parametrization efforts.

Sampling is performed by creating and merging availability and climate scenarios. Availability scenarios are created with the availability preprocessing module in the base scenario. Climate scenarios represent other impact factors, such as climate impacts, found in the derived scenarios.

The components and their availabilities are sourced from the base scenario. All other data, including climate dependent impact factors, are loaded from the derived scenario, as illustrated in the table below.

Investment

The procedure for investment runs is applied at the investment page by click on Run simulation.

Investment simulation runs through an iterative approach can be performed through the available interfaces. The documentation here is currently under development. For more information, please contact the Maon-Team.