The energy transition toward a carbon-neutral energy system presents complex challenges that require reliable system-level insights to guide investment and policy. Energy system models are essential tools in this context. They support planning by simulating interactions between t
...
The energy transition toward a carbon-neutral energy system presents complex challenges that require reliable system-level insights to guide investment and policy. Energy system models are essential tools in this context. They support planning by simulating interactions between technologies, markets, and policies under various future scenarios. Their strength lies in their ability to highlight structural system relationships and test the feasibility of different energy strategies.
At the foundation of these models lie assumptions and simplifications that define the internal logic of an energy system model. Importantly, a distinction must be made between assumptions (e.g., cost or efficiency parameters) and simplifications (e.g., ignoring demand fluctuations or omitting battery interaction). While simplifications make models tractable and transparent, they also risk overlooking key real-world constraints. This is why testing the impact of these assumptions and simplifications is critical: doing so ensures that model outcomes are robust and that their conclusions remain meaningful in practical applications.
Energy modelling simulates the operation and evolution of energy systems to support decision-making and policy planning. It helps simplify complex systems, forecast scenarios, and evaluate the effects of different strategies. While models are never perfectly accurate, their usefulness depends on data quality, transparent assumptions, and iterative refinement. These assumptions directly shape model credibility and must be rigorously tested to avoid the risk of unvalidated assumptions becoming accepted truths that undermine decision-making.
One such model is the Kramer and Koning Model (KKM), a stylised energy model developed to analyse the relationship between renewable electricity generation and hydrogen capacity. The KKM is appreciated for its simplicity and its capacity to clarify the fundamental relationship between renewable energy generation and electrolyser capacity - the r : e relationship. However, this simplicity raises the question of how sensitive its results are to added real-world complexities and how valid its outcomes remain. This study addresses that knowledge gap by investigating: "How Do Key Model Assumptions in the KKM Influence the Relationship Between Renewable Energy and Electrolysis Deployment?".
To evaluate the validity of KKM outcomes, this study introduces the Electrolyser Battery Balancing Model (EBBM) - a more detailed cost optimisation model operating under the same logic as the KKM, but with extensive additional parameters. The EBBM simulates hourly interactions between renewable supply, demand, electrolysers, and batteries. Developed in collaboration with Gasunie, a key player in the Dutch gas infrastructure and hydrogen transition, the EBBM is specifically designed to test real-world factors and find the cost-optimum interplay between renewable, electrolysis, and battery capacity. It is well-suited to validate the simplified relationships modelled by the KKM.
Firstly, a systematic identification of assumptions in the KKM was made. These were categorised as either explicit or implicit. Implicit assumptions were further divided into (1) real-world system simplifications (e.g., omitting compressors, conversion losses), and (2) wider context simplifications (e.g., sector coupling, market conditions). Based on their role in the model and feasibility for testing in the EBBM, a focused selection of assumptions was made, grouped into four categories: renewable energy, hydrogen, cost, and system simplifications. The eventual selection consisted of:
• Generation Mix;
• Electrolyser Efficiency;
• Electrolyser Limitations;
• Hydrogen Storage Cost;
• Cost Ratio between Renewables and Electrolysers;
• Neglect of Demand Fluctuations;
• Battery Interaction Exclusion;
• Demand Flexibility.
Moving on with the selected set of assumptions and simplifications, a sensitivity analysis was first conducted by incrementally reintroducing high-certainty system simplifications to the KKM base case. This included adding demand fluctuations, battery interaction, electrolyser efficiency curves, hydrogen storage cost and electrolyser limitations to create a new, more realistic base case. This updated case was then used to test the impact of four key parameters: electrolyser efficiency, demand flexibility, solar share, and the cost ratio between renewables an electrolysers. In each case, a high and low value was tested. These variations were used to assess how much each assumption shifts the r : e relationship, battery sizing (r : b), and total system cost (c).
Firstly, the incremental addition of complexities resulted in a flatter slope and lower overall system cost compared to the original KKM. Further results showed that parameters like solar share and cost ratio significantly affect infrastructure allocation between batteries and electrolysers, while demand flexibility and efficiency assumptions moderately shift total system cost and capacity sizing. The r : e relationship remained structurally linear in all cases but varied in slope and magnitude. Notably, the combination of battery interaction and electrolyser efficiency assumptions produced the largest cost savings, lowering total decarbonisation cost by several hundred euros per kW relative to the KKM.
A robustness analysis followed, designed to assess whether model outcomes remain valid under extreme input conditions (edge cases). These edge cases were selected for the same assumptions as for the sensitivity analysis. The aim was to evaluate whether the KKM’s simplified relations hold up under stress. The results indicated that while the relationship itself remains observable, its quantitative implications (e.g., cost and deployment levels) vary substantially, suggesting that the relation needs to be interpreted as directional rather than predictive.
To further contextualise the findings, a comparative model analysis was conducted. This compared the r : e relationship in the KKM against other existing energy system models. A longlist was developed and refined to three studies: CE Delft, E-Bridge, and a NSWPH study. Extracted data confirmed that while each model uses different frameworks, a consistent structural trend in the r : e relation is present, supporting the underlying logic of the KKM, albeit under different boundary conditions.
Despite differences in geography, modelling scope, and sectoral integration, all three studies showed a similar acceleration in electrolyser deployment relative to renewable generation, particularly beyond 2040. This convergence across models suggests that the r : e relationship is a robust feature of future energy system dynamics, rather than an artefact of a specific model setup. It reinforces the validity of the KKM’s structural assumptions, even if absolute outcomes vary. As such, the r : e relation emerges as a valuable comparative indicator for system modellers and energy planners aiming to align infrastructure scaling with decarbonisation timelines.
In the discussion, the findings reveal that while the KKM offers a robust conceptual tool, its practical outputs are assumption-sensitive.. Key limitations include the use of a single weather year to simulate renewable variability, a strictly unidimensional approach to parameter varying, and the degree of certainty with which a particular impact can be attributed to an assumption in another model. These issues are particularly important for policymakers or investors relying on model outputs for long-term infrastructure decisions.
The conclusion confirms that the KKM captures a fundamental structural relationship between renewables and hydrogen capacity, which reappears when evaluating other models. However, the outputs of the KKM are highly dependent on assumption quality and scope, especially regarding solar share and the cost ratio between renewables and electrolysis. The research shows that integrating high-certainty simplifications and testing uncertain variables adds valuable depth. Therefore, the KKM proves useful for identifying strategic trends in the r : e relation. Future research should extend this work by incorporating power-to-heat, more detailed battery interaction, and policy scenarios to increase applicability in real-world system design.
