[Paper Review] Hedging Hydrogen: Planning and Contracting Under Uncertainty for a Green Hydrogen Producer

Owen Palmer, Hugo Radet+3 more|arXiv (Cornell University)|Jul 31, 2024

Energy, Environment, and Transportation Policies4 citations

TL;DR

This paper proposes a two-stage stochastic optimization model that co-optimizes equipment sizing and energy hedging strategies—using power purchase agreements (PPAs) and power futures—for electrolytic green hydrogen producers under uncertainty. It demonstrates that stochastic methods are essential when demand is uncertain, while rule-based hedging suffices in simpler cases, and that RFNBO subsidies drive demand for geographically and technologically diverse PPA portfolios to meet regulatory requirements.

ABSTRACT

Green hydrogen production by water electrolysis using renewable electricity is considered essential for decarbonisation of certain sectors of the global economy, however development of the industry is lagging behind expectations due to the perceived financial risk for individual projects. This risk stems from a number of uncertainties, including future hydrogen demand, variable renewable energy sources, and volatile energy market prices. The interaction of these uncertainties is complex, yet the analysis of hydrogen projects is often carried out using simplified modelling that often omits uncertainty and/or energy hedging practices which are typical for intensive power consumers. In this study, we define a set of planning methods (planning policies) in order to compare the effectiveness of different modelling approaches. We propose a 2-stage market-focused stochastic program to represent a hydrogen producer supplying an industrial customer through a hydrogen offtake contract (a Hydrogen Purchase Agreement, or HPA). The model can be used to obtain equipment sizing decisions, as well as energy hedging decisions using Power Purchase Agreements (PPA's) and power futures. We find that for some HPA contract types, failure to use stochastic modelling can lead to planning decisions that result in 30% higher production costs during scenario stress-testing for the same project. This could lead to some projects being discarded by developers, incorrectly deemed to be unviable due to cost projections being too high. The results also show the importance of HPA contract volumetric obligations in limiting demand uncertainty.

Motivation & Objective

• Address the financial risk in green hydrogen projects stemming from uncertain electricity prices, renewable generation, and variable demand.
• Overcome the limitation of treating asset sizing and energy hedging as separate decisions by co-optimizing them in a single model.
• Evaluate the effectiveness of deterministic vs. stochastic planning methods under different uncertainty regimes.
• Assess the impact of European RFNBO classification rules and associated subsidies on optimal PPA portfolio design and hydrogen cost.
• Provide a decision-support framework for hydrogen producers and offtakers to structure bankable Hydrogen Purchase Agreements (HPAs) under price and demand risk.

Proposed method

• Develops a 2-stage market-focused stochastic program modeling a green hydrogen producer supplying an industrial customer over a long-term horizon.
• Incorporates key decision variables: electrolyzer capacity, hydrogen storage size, PPA volume, and power futures position.
• Uses risk-averse stochastic optimization to minimize the expected Levelized Cost of Hydrogen (LCOH) while accounting for day-ahead market price volatility.
• Models uncertainty in three key drivers: renewable generation (wind and solar), day-ahead electricity market prices, and hydrogen demand.
• Applies scenario-based decomposition with Monte Carlo sampling to represent temporal correlations and uncertainty realizations.
• Validates results against deterministic rule-based hedging heuristics and benchmarks performance across different regulatory contexts (RFNBO compliance levels).

Experimental results

Research questions

• RQ1 How does the integration of energy hedging (PPAs and futures) with optimal equipment sizing affect the Levelized Cost of Hydrogen (LCOH) under uncertainty?
• RQ2 Under what conditions do stochastic optimization methods outperform deterministic, rule-based hedging strategies in hydrogen project planning?
• RQ3 How does demand uncertainty influence the need for flexible assets and the resulting hydrogen sale price?
• RQ4 What impact do RFNBO eligibility rules—requiring hourly correlation between renewable production and hydrogen output—have on PPA portfolio design and diversification?
• RQ5 To what extent do green hydrogen subsidies stimulate investment in geographically and technologically diverse PPA portfolios?

Key findings

• In the absence of demand uncertainty, simple rule-based hedging (e.g., full hedging via PPA) can achieve near-optimal results, but stochastic models are essential when demand is uncertain.
• When demand uncertainty is present, stochastic optimization leads to higher equipment sizing (especially hydrogen storage) and increases the mean LCOH to 6.54 €/kg in Context A and 4.67 €/kg in Context C, with worst-case LCOH reaching 7.17 €/kg.
• The model shows that RFNBO compliance (100% renewable correlation) increases the need for diversified PPA portfolios, particularly in high-RFNBO contexts like Context C, where 100% of production must be RFNBO-compliant.
• In Context B, where 81% of production must be RFNBO-compliant, the model indicates a strong preference for diversified PPA portfolios to meet regulatory and cost targets.
• Without RFNBO subsidies, only 40% of production is RFNBO-compliant in Context A, suggesting that current market conditions alone are insufficient to drive high RFNBO penetration.
• The results suggest that green hydrogen subsidies are likely to stimulate demand for technologically and geographically diverse PPA portfolios to meet RFNBO criteria and reduce cost volatility.