IAP-24-121

Modelling Natural Hydrogen systems using Agent Based modelling

In this project, we aim to pioneer a novel tool for natural hydrogen exploration, based on agent-based modelling (ABM), enabling the rapid screening of uncertainties in natural hydrogen systems. By modelling the interactions between fluid, rocks, and the subsurface biome as simple agents that interact with each other, we can efficiently capture the system’s complex behaviours at a low computational cost. A working tool will help to discover more potential reservoirs of naturally trapped/accumulated hydrogen and derisk drilling into these new resource plays to access this critical energy transition fuel. The key benefit of our approach is its high computational efficiency in capturing the complex migration of hydrogen as it interacts with the geology and the microbiome.

Hydrogen is a key resource for achieving net zero, and one novel, untapped source is natural or gold hydrogen. Similarly to hydrocarbons, gold hydrogen is produced in the subsurface via various geochemical processes and then migrates and accumulates in geological reservoirs. Natural hydrogen reservoirs are rare, with only one commercial operation in Mali. Recent discoveries in Australia, France, Spain, USA, and Albania suggest significant economic volumes of hydrogen, highlighting this resource’s global potential.

A barrier to successful exploration is a lack of modelling tools that account for generation mechanisms, recharge rates, and retention issues specific to natural hydrogen. Current software tools, designed for hydrocarbon exploration, are not well-equipped to handle the unique uncertainties of hydrogen systems but currently provide the only option for basin modelling.

Significant hydrogen accumulation occurs only when production exceeds consumption at the source, migration is faster than consumption on the way up, and consumption rates in the reservoir are low or non-existent. Consumption may be microbial or abiotic (e.g., Sabatier reaction, Fischer-Tropsch synthesis, or sorption to clays), with microbial activity being a key factor that hinders the formation of large hydrogen reservoirs, which is currently not modelled.

This project will include the microbial consumption of hydrogen in exploration models to forecast hydrogen reservoir potential. Various microorganisms, including sulfate reducers, methanogens, and acetogens, thrive in the subsurface and consume hydrogen, producing different byproducts such as sulfides, carbon dioxide, methane, and acetate. Each microbial process will only occur within a specific pH range, temperature, and salinity conditions. Additionally, some microorganisms produce hydrogen biotically, as noted by Gregory et al. (2019). We must model complex interaction of this microbial community with the geology and the migrating hydrogen to understand the conditions needed for hydrogen to accumulate.

We will build upon previous work by the Heriot-Watt (HWU) supervisory team on ABM of petroleum migration, as outlined in Steffens et al (2022), later demonstrated on the Wessex basin petroleum system by Kreiensiek et al. (2022). Our goal is to adapt our existing ABM code to incorporate a hydrogen-migrating agent, as well as secondary agents such as hydrogen-consuming microorganisms, and to model the interactions between them. This will require a multi-scale approach to accurately capture microbiome interactions at the small pore scale and then translate those insights into bulk behaviour. The models will use formal rules or reinforcement machine learning (ML) to represent these behaviours. To better capture migration pathways through geological formations, we will enhance our ABM (which constructs geological layers directly from seismic data) by incorporating the more rigorous GEMpy python framework (de la Varga et al., 2019).

Dr. Simon Gregory from BGS will provide expertise on the interactions between hydrogen and microbes that consume it. We will utilise existing knowledge and published microbiome data to construct our agent models, avoiding the need for new, complex experiments at this stage. Thaysen et al. (2021) for example, provides an approach to calculate the consumption and reproduction rates of hydrogen-consuming microorganisms from which to train agents. Agent rules can be hard-coded or learned through reinforcement learning where necessary.

The PhD program will finish with an application on a real gold hydrogen exploration area with our partner, H2Au, aiming to validate our approach and gain insights into a natural hydrogen system.

The project involves developing and testing agent-based models (ABMs) using open-access Python code. A powerful workstation beyond standard university specifications is required for this development.

The candidate must understand and potentially conduct experiments to elicit appropriate rules for microbial agents. The student will visit the British Geological Survey (BGS) laboratories in Keyworth at least once to learn from Dr Gregory and visit the laboratory. Relevant fieldwork will also help the student understand the geological factors influencing natural hydrogen systems and control the design of geological models.

A possible internship with our industrial partner H2Au would happen towards the end of the PhD, ideally applying the developed ABM tool to a real-world gold hydrogen play.

Year 1

In the first year, The project will address the question of how we can model hydrogen and microbial agents within a system. The focus will be on defining high, medium, and low thresholds for microbial activity and understanding the distribution of various species in the subsurface. The project will develop and test an agent-based model (ABM) framework for hydrogen exploration by updating the fluid agent from our hydrocarbon model to represent hydrogen (H2) and incorporating microbiological agents. Given that hydrogen behaves similarly to hydrocarbons in terms of migration, this process should be straightforward. The main effort will involve adding a competing microbiological agent and capturing their interactions using rules informed by prior research. The project will evaluate a simple microbial agent’s impact on hydrogen migration based on factors such as consumption rate, output, movement rate, reproduction rate, environmental limits (like pore size), sensitivity to geochemistry, and temperature.

Year 2

In year 2, the project will explore how the competition within the microbiome affects hydrogen migration and preservation. The project will enhance the hydrogen agent-based model (ABM) from year 1 to include stratigraphic and depth variations of the microbial community. This requires increasing microbial diversity, understanding their spatial arrangement, and incorporating geological and geochemical factors, such as temperature gradients and pore space depth. The project will use a multi-scale approach to modelling to capture the macro behaviour of a rock mass based on representative micromodels of the fluid and microbial community. The distribution of bacterial agents based on depth, temperature, and salinity is complex, as they often compete for space and resources. Any biome model must consider geological, geothermal, and water chemistry influences. The project will use rapid ABM simulations to examine the significance of these factors.

Year 3

In Year 3, the project will focus on quantifying uncertainty in the accumulation of natural hydrogen in a real geological system. H2Au, a UK-based hydrogen exploration company, is a collaboration partner on this project providing access to sources of real-world exploration data and internship opportunities. The project will aim to find suitable data to assess real hydrogen play concepts. This will, where possible, involve an internship. The collaboration will address questions about charge rates, geochemistry, and tectonic history’s effects on hydrogen emplacement, including conditions for hydrogen formation at shallower depths, as seen in Mali.

Year 3.5

The final 6 months will focus on writing the thesis and publications from years 2 & 3.

The candidate should have or develop coding skills in Python and learn about ABM and ML techniques, with support from HWU and relevant external partners. They will gain knowledge about microbial and hydrogen experimentation methods from BGS, although no experiments will be conducted. Additionally, the candidate will study natural hydrogen systems and geological factors influencing hydrogen production and migration to create geological models, using relevant seismic data.