PowHyCEM is a spatially aware integrated power–hydrogen capacity expansion model for long-term energy system planning.
The model jointly optimizes investment, retirement, and hourly operation of power and hydrogen infrastructure while accounting for:
- power–hydrogen sector coupling;
- renewable and thermal generation;
- electricity transmission and hydrogen pipelines;
- electricity and hydrogen storage;
- emissions and carbon mitigation costs;
- hourly operational constraints; and
- spatial land availability and cumulative infrastructure land use.
This repository contains the model implementation and input data associated with the study:
When land becomes the bottleneck: Spatial feasibility determines decarbonization pathways
PowHyCEM/
│
├── PowHyCEM_BD.jl # Main execution file
├── Input_Data/ # Input datasets for the case study
│
└── src/
├── Config.jl # Solver and decomposition settings
├── Data_loading.jl # Data loading and preprocessing
├── Master_problem.jl # Investment/master problem
├── Sub_problem.jl # Weekly operational subproblems
├── Benders_loop.jl # Benders decomposition algorithm
├── Output.jl # Result processing
└── Write_*.jl # Output-writing routines
The model was developed and tested using:
| Software | Version |
|---|---|
| Julia | 1.9.2 |
| Gurobi | 12.0.1 |
| JuMP | Julia optimization package |
| CSV | Julia data package |
| DataFrames | Julia data package |
| Plots | Julia plotting package |
| Measures | Julia plotting utility |
A valid Gurobi license is required.
The required Julia packages can be installed using:
using Pkg
Pkg.add([
"JuMP",
"Gurobi",
"CSV",
"DataFrames",
"Plots",
"Measures"
])git clone https://github.com/rezfa/PowHyCEM.git
cd PowHyCEMThe datasets used for the case study are provided in:
Input_Data/
For reproduction of the submitted case study, the input files can be used without modification.
From the root directory of the repository, execute:
julia PowHyCEM_BD.jlThe main script performs the complete workflow:
- loads and preprocesses the input datasets;
- creates the investment master problem;
- creates the weekly operational subproblems;
- solves the model using regularized temporal multi-cut Benders decomposition;
- checks convergence of the upper and lower bounds; and
- processes and exports the model results.
Solver settings and parameters of the decomposition algorithm can be inspected or modified in:
src/Config.jl
For reproduction of the results reported in the paper, reviewers are advised to retain the default configuration and input datasets provided in this repository.
The Input_Data/ directory contains the main datasets required by PowHyCEM, including information on:
- power generation technologies;
- hydrogen production technologies;
- hourly electricity demand;
- hourly hydrogen demand;
- renewable generation availability;
- electricity transmission infrastructure;
- hydrogen pipeline infrastructure;
- fuel and technology costs; and
- zone-specific spatial characteristics.
The input assumptions correspond to those described in the manuscript and Supplementary Information.
PowHyCEM represents hourly system operation over a complete year.
To maintain tractability, the operational problem is decomposed into 52 weekly subproblems using a regularized temporal multi-cut Benders decomposition approach.
The main problem contains long-term planning decisions, while the weekly subproblems represent detailed system operation. The weekly subproblems can be solved in parallel.
The numerical experiments reported in the paper were performed on the Hábrók high-performance computing cluster at the University of Groningen.
The model can also be executed on other systems with Julia and Gurobi installed; however, computational time depends strongly on:
- processor performance;
- number of available CPU cores;
- Gurobi configuration; and
- problem size.
Running the complete 14-zone, hourly case study on a standard personal computer can therefore require substantially more computation time than on an HPC system.
To reproduce the submitted case study:
1. Clone this repository
2. Install Julia and the required packages
3. Install and activate Gurobi
4. Keep the provided Input_Data files unchanged
5. Keep the default model configuration
6. Run PowHyCEM_BD.jl
The mathematical formulation of the model, spatial methodology, scenario assumptions, and decomposition algorithm are described in detail in the manuscript and its Supplementary Information.
The output routines contained in src/ process the main model results, including:
- installed and retired generation capacity;
- power and hydrogen production;
- storage investment and operation;
- electricity transmission;
- hydrogen pipeline flows;
- emissions;
- system costs;
- land use; and
- other system-level and zone-level indicators.
For questions regarding the model, input data, or reproduction of the results:
Reza Fardi Asrami
Department of Operations
Faculty of Economics and Business
University of Groningen