Chasing fluid pathways: using multiscale modelling of subduction zones to unravel the role of fluid dynamics on fluid flow across slab-wedge-crust domains - PhD

University of Glasgow

Glasgow, UK 🇬🇧

Project institution: University of Glasgow

Project supervisor(s): Dr Chun Hean Lee (University of Glasgow), Dr Antoniette Greta Grima (University of Glasgow), Prof Antonio J. Gil (Swansea University) and Dr Tobias Keller (University of Glasgow)

Overview and Background

Overview: This PhD studentship focuses on developing GPU-accelerated models of subduction dynamics and surface evolution, emphasising fluid release at the slab-mantle interface, volatile transport, and melt dynamics, with implications for volcanic hazard and critical resource formation potential. Subduction processes, fluid release and flow, and the resulting surface response of our planet operate across different scales, spanning from grain size to regional scale dynamics and span across quasi-instantaneous timescales to millions of years. This project will contribute to a multiscale framework by using advanced GPU-based techniques to better understand how fluids released at the slab-mantle interface migrate into the mantle wedge and influence subduction processes and melt generation, and interactions with the overriding crust. This flow is driven by significant pressure gradients, thermal gradients, and mechanical deformation. The study will involve phase changes (i.e., solid to liquid transitions) and multi-phase flow interactions between fluids and (viscoplastic) solids (plates and mantle wedge). From a modelling perspective on fluid flow across slab-wedge-crust domains, we will build upon the Smoothed Particle Hydrodynamics (SPH) methods recently developed by the supervisory team and extend them to simulate two fluid-solid phase, multi-scale non-Newtonian fluids, For large-scale simulations, these SPH developments will be integrated in the GPU-accelerated, open-source SPH code “DualSPHysics”. Importantly, these developments will be carefully linked to subduction models generated using ASPECT, from which the initial and boundary conditions for the problem will be extracted. A feedback loop between ASPECT and DualSPHysics will be established to iteratively refine the modelling approach. This project will include software development, integrating and interpreting field and experimental data sets, attending regular seminars, collaborating within a research team, and receiving training through ExaGEO workshops.

Background: Subducting slab transport altered near-surface rocks into the Earth’s mantle, introducing volatiles that sustain the deep water and carbon cycles and are crucial in generating melt and magmatic processes. Subduction links shallow and deep Earth systems, maintains conditions essential for a habitable planet (e.g., Tian et al., 2019) and at shallow depths has a critical influence on mineral resource emplacement.

Furthermore, the volatiles and fluids carried by the subducting slab influence subduction style, weaken and fracture the overriding plate, induce fluid release from the slab and into the mantle wedge, and control the location, timing, composition, and volume of arc magmatism (Nakao et al., 2016). These processes in turn govern volcanic hazards and the formation of critical metal deposits (Faccenda, 2014). Despite the key role of fluids in subduction zones, fluid release and the mechanisms controlling volatile and melt dynamics, and reactive fluid transport in subduction dynamics remain poorly understood. This project will utilise high-resolution, GPU-accelerated simulations to investigate the interaction between fluid and subduction dynamics and their expression at the Earth’s surface.

Methodology and Objectives

Modelling subduction processes is a challenging scientific and computational problem especially when this is coupled with free surface deformation and fluid release and transport. This is due to the multi-scale, multi-component and multi-phase processes and feedbacks which operate across different timescales and variable rheologies. To tackle this complexity, this project will adopt a multiscale modelling approach combining small-scale simulations of reactive fluid transport and lithospheric dynamics with large scale 2-D and 3-D subduction models with a free surface evolution. These models will capture the effects of dynamic mechanisms, strong thermal gradients, phase changes, and two-phase flow, aiming to provide a reliable representation of fluid migration from the slab-mantle interface into the mantle wedge. GPU architectures will be used to couple local and regional scale models with adaptive refinement to ensure sufficiently high resolution at the various process scales. 

In this project, the candidate will build upon the explicit Smoothed Particle Hydrodynamics (SPH) formulation for Newtonian fluid flows proposed by the supervisory team (Low et al., 2021). This formulation has demonstrated the ability to mitigate long-standing SPH numerical artefacts, such as tensile instability, through the introduction of a novel entropy-stable stabilisation method (Lee et al., 2023). The candidate will first extend the formulation to model single-phase non-Newtonian fluid flow. To improve computational efficiency, an implicit formulation will be implemented. The formulation will then be extended to two-phase flows, focussing on non-Newtonian fluids interacting with viscoplastic solid materials. To enhance the accuracy and robustness of the algorithm, thermodynamically compliant particle-shifting techniques based on the recently developed ALE framework (Lee et al., 2024) will be introduced. Nonlinear Riemann solvers (Runcie et al., 2022) will be developed to accurately resolve sharp phase transitions, as well as pressure and thermal gradients. Additionally, well-posed material models, such as thermo-viscoplasticity, across any coupled deformation states will be incorporated. Finally, the method will be implemented in the GPU-accelerated, open-source SPH code “DualSPHysics” to effectively handle a wide range of spatial and temporal scales. A feedback loop will be established between ASPECT (subduction models) with DualSPHysics (fluids migration) to iteratively exchange data, enabling continuous refinement of the modelling approach and ensuring consistency between the subduction models and SPH simulations.   

Within this framework, the student will start by working on two “teaser” projects to gain familiarity with the underlying theory, SPH techniques and relevant data. The student will then decide how to further develop their research, focus their efforts, and implement their work on GPUs.

Teaser Project 1:

The objective of teaser project 1 is to enable the candidate to understand the mechanics and physical behaviour of both Newtonian and non-Newtonian flows through a simple implementation. This sub-project, conducted during months1-6 of the first year, will focus on two primary areas: (1) computational mechanics, including conservation laws (e.g., Total Lagrangian, Updated Lagrangian, and Arbitrary Lagrangian Eulerian (ALE)), interface resolution using Riemann solvers, hyperbolicity/stability, and material modelling; and (2) SPH schemes, covering stabilisations, consistency, stability, and convergence. The student will begin by working with a simple explicit-based SPH MATLAB code for single-phase Newtonian flows. The candidate will then extend the code to handle non-Newtonian single-phase flows using an implicit formulation. The student will also identify challenges in implementing SPH on GPUs, focussing specifically on aspects such as such the scalability of the neighbouring search, force calculations, and SPH kernel and gradient corrections.

Teaser Project 2:

The objective of teaser project 2 is to enable the candidate to understand the capabilities of DualSPHysics in simulating multi-phase flows on CPU and GPU platforms, which will be essential for understanding flow along the slab-mantle wedge-continental crust interface. This sub-project, conducted during months 7 -12 of the first year, will focus on familiarising the open-source SPH code “DualSPHysics”. The candidate will test and evaluate the main functionalities of the code, including the single-phase free surface fluid solver, and Newtonian/Newtonian multi-phase solver. Several benchmark test cases will be tested and compared on both CPU and GPU platforms. Performance metrics, such as accuracy and computational efficiency will be assessed to understand the advantages and limitations of GPU acceleration.

The proposed project is designed as a standalone initiative, but it includes a cohort-based learning experience for linked projects related to subduction modelling.

References and Further Reading

  1. Low, K., Lee, C.H., Gil, A.J., Haider, J. & Bonet, J. (2021). A parameter-free Total Lagrangian SPH algorithm applied to problems with free surfaces. Computational Particle Mechanics, 8, 859892 (click here)
  2. Lee, C.H., de Campos, P.R.R., Gil, A.J., Giacomini, M. & Bonet, J. (2023). An entropy-stable Updated Reference Lagrangian SPH algorithm for thermo-elasticity and thermo-visco-plasticity. Computational Particle Mechanics, 10, 1493-1531 (click here)
  3. Lee, C.H., Gil, A.J., de Campos, P.R.R., Bonet, J., Jaugielavicius, T., Joshi, S. & Wood, C. (2024). A novel ALE SPH algorithm for nonlinear solid dynamics. CMAME, 427, 117055 (click here)
  4. Runcie, C.J., Lee, C.H., Haider, J., Gil, A.J. & Bonet. (2022). An acoustic Riemann solver for large strain computational contact dynamics. IJNME, 123, 5700-5748 (click here)
  5. Faccenda, M. (2014). Water in the slab: A trilogy. Tectonophysics614, 1–30 (click here)
  6. Heister, T., Dannberg, J., Gassmöller, R., & Bangerth, W. (2017). High accuracy mantle convection simulation through modern numerical methods – II: Realistic models and problems. Geophysical Journal International, 210(2), 833–851 (click here)
  7. Nakao, A., Iwamori, H., & Nakakuki, T. (2016). Effects of water transportation on subduction dynamics: Roles of viscosity and density reduction. Earth and Planetary Science Letters454, 178–191 (click here)

3 days remaining

Apply by 17 February, 2025

POSITION TYPE

ORGANIZATION TYPE

EXPERIENCE-LEVEL

DEGREE REQUIRED

You ad could be here!