InfraInt – Infrastructure development model for hydrogen and CO₂

The optimization model “InfraInt” models existing pipelines, the new construction of hydrogen and CO2 pipelines, and the repurpose of natural gas pipelines to hydrogen pipelines. In addition, the locations of electrolysis and methanation are determined in a cost-optimal way to give a complete picture of the transport infrastructure. The model integrates sources and sinks of energy-relevant gases into a holistic infrastructure. Accordingly, the name of the model is also derived from the terms “infrastructure” and “integrated”.

Context

The infrastructure for transporting energy-relevant gases is already partly in place today (natural gas grid) or must be newly constructed in the context of the energy transition (hydrogen, CO2). The transport effort depends on the location of production and consumption. There are basically two options for the production of renewable gases:

First, production can be carried out at the plant site of renewable power generation. To transport synthetic fuels to consumption centers, pipeline-based gas infrastructure is required. For synthetic methane production, the use of captured CO2 from industrial point sources is an option. Transport of the CO2 to the generation facilities can again be accomplished via pipeline-based infrastructure. If the CO2 is materially captured, it must also be transported to the consumers. Hydrogen can be fed into the existing natural gas grid to a certain extent. However, for transportation of pure hydrogen to, for example, steel industry load centers, new pipelines must be built or existing natural gas pipelines repurposed.

On the other hand, renewable fuel generation plants can be placed directly at load centers. In this case, a large part of the gas grid infrastructure would become obsolete.

 

What is InfraInt used for?

The model is intended to show the future expansion needs of H2 and CO2 pipelines in Europe. An example result for hydrogen infrastructure in 2030 is shown in Figure 1. The model indicates what capacities the pipelines must have to meet the demand of the counties. It also highlights the costs of building and operating the pipelines. The pipeline infrastructure is determined based on cost minimization, so the calculated transportation network specifies the least-cost option for pipeline-based transportation of hydrogen and CO2. Existing plans, such as the Hydrogen Backbone are included in the model as one option of future hydrogen transport. As an additional component, the locations and generation capacities of electrolysis and methanation are determined at the county level, specifying additional sinks of hydrogen. In particular, the location of the conversion plants provides information on whether electrolysis is placed close to generation when renewable electricity is produced, or close to consumption in counties with hydrogen demand, eliminating the need for connection through hydrogen pipelines. Again, the costs of electrolysis and methanation can be evaluated depending on their capacity.

 

Figure 1: New construction of hydrogen pipelines and repurposed natural gas pipelines in 2030, as well as electrolysis sites.

How does InfraInt work?

Figure 2: Methodology of the simulation in InfraInt

Input data regionalized to county level:

  • Electricity demand as well as electricity supply from renewable energies to indicate electricity availability for on-site electrolysis
  • Hydrogen demand from the sectors transport, building, industry and power generation
  • Demand for synthetic methane
  • CO2 capture from industrial waste gases
  • Model of the existing natural gas transmission network

A cost function is determined in which the following factors are included:

  • Costs for new pipeline construction depending on length and capacity (investment and operation)
  • Costs for repurpose of natural gas pipelines to hydrogen pipelines
  • Costs for electrolysis depending on generation capacity
  • Costs for methanation depending on capacity
  • Electricity grid costs for purchasing electrolysis electricity from the electricity grid
  • Costs for CO2 capture
  • Costs for methanation and electrolysis

When minimizing the cost function, the following constraints must be satisfied:

  • Meeting the demand of hydrogen in each county
  • Coverage of synthetic methane demand in each county
  • Maximum captured CO2 equals the potential of capture
  • Limitation of gas flows by capacities of pipelines

 

Technical details

The model corresponds to a node-edge model, where the nodes represent the centers of counties and the edges represent connections between counties, which can be connected by a pipeline. The coverage of hydrogen and CO2 demands by the modeled infrastructure is considered at the temporal resolution of one year.

The cost minimization corresponds to a mixed-integer linear optimization. The integer nature is necessary to represent the binary decision to build a pipeline connection, which only makes sense once a certain pipeline capacity is reached. Due to the inclusion of the binary variables, the cost function cannot be represented in a purely linear way and is thus much more difficult to solve. Currently, optimality gaps of about 10% are reached. The model is implemented in Python and uses either the solver CPLEX or Gurobi for optimization.

Technical details

The model corresponds to a node-edge model, where the nodes represent the centers of counties and the edges represent connections between counties, which can be connected by a pipeline. The coverage of hydrogen and CO2 demands by the modeled infrastructure is considered at the temporal resolution of one year.

The cost minimization corresponds to a mixed-integer linear optimization. The integer nature is necessary to represent the binary decision to build a pipeline connection, which only makes sense once a certain pipeline capacity is reached. Due to the inclusion of the binary variables, the cost function cannot be represented in a purely linear way and is thus much more difficult to solve. Currently, optimality gaps of about 10% are reached. The model is implemented in Python and uses either the solver CPLEX or Gurobi for optimization.