진학프로석·박사 채용 접수 플랫폼

Natural resource accumulations (such as natural hydrogen, lithium and copper) result from fluid flow, heat transfer and fluid–rock interactions over geological timescales. In order to understand these systems and predict the location of resources more accurately, it is necessary to model these processes jointly at the basin scale.

However, directly solving geochemical equations is computationally expensive and difficult to couple with basin simulators. Geochemical metamodels, particularly those based on machine learning, can significantly reduce computation times while maintaining physico-chemical consistency. While recent advances have explored using such learning methods to accelerate calculations, their application remains limited to simple coupling schemes and narrow physicochemical ranges. This often makes these models case-specific (Guerillot & Bruyelle, 2020; De Lucia & Kühn, 2021; Demirer et al., 2023).

This project aims to address these challenges by optimising the choice of input and output variables, and by developing more complex coupling schemes. The aim is to improve the prediction of natural resources in sedimentary basins.

The PhD must have been awarded no more than three years before the start of the postdoctoral contract.

Main missions and activities

The postdoctoral candidate will develop robust metamodels for each of the main geochemical reactions controlling hydrogen release and trapping in the subsurface, using machine-learning approaches. Key tasks will include:

1. • Developing an in-depth understanding of the chemical processes involved in the hydrogen cycle.
2. • Reducing complex processes to representative reactions suitable for modelling.
3. • Selecting and training metamodels (analytical laws, response surfaces, AI-based approaches), and validating them by comparison with full geochemical calculations.
4. • Performing control and accuracy analyses of the metamodels.
5. • Integrating the metamodels into the reactive transport module of a basin simulator and testing them on a real natural hydrogen system.

This postdoctoral project, carried out within the PEPR Sous-Sol programme in collaboration with Mines Paris – PSL, will be structured in two main phases: a learning phase on a well-documented case study, and a methodological transfer phase towards a more general topic: natural hydrogen (H₂) generation at the basin scale, with implementation in the simulator.

Travel

Occasional short trips (a few days) within France.

Technical skills

• • Solid experience in geochemical modelling
• • Previous experience with statistical or AI-based modelling methods is desirable (non-linear regressions, response surfaces, cross-validation, uncertainty analysis, design of experiments).
• • Advanced proficiency in at least one programming language (Python or C++).

Soft skills

Ability to work independently, to operate in an interdisciplinary environment, and to disseminate research through scientific writing in French and English and presentations at conferences.

Languages

English required (minimum C1 level).

Knowledge of French is desirable.

Degree

PhD in geochemistry or a related field.