Zhi Li

Overview

Salinity Evolution in Coastal Wetlands

I develop numerical models to simulate surface flow, surface-subsurface exchange and scalar transport in coastal wetlands. The focus is on (1) representing small-scale environmental processes in corase-grid models and (2) optimizing model coupling to enable fast 3D, field-scale simulation.

A subgrid topography model is designed to parametrize small-scale topographic features on much coarser grid cells. This method allows hydrodynamic simulation on efficient coarse grids without losing the small-scale flow-topography interactions. It is useful for shallow wetland simulations where water flow is strongly affected by the complex topography.

For more info:

10.1016/j.advwatres.2019.05.004
10.1016/j.advwatres.2019.103465

A coupled 2D surface flow and 3D variably saturated subsurface flow model (Frehg) is developed. The purpose of developing Frehg is to model salintiy evolution in shallow wetlands that is affected by tidal intrusion, evaporation, surface-subsurface exchange, and the complex 3D topography. The codes of Frehg can be found on my Github repo.

For more info:

10.3390/w13070902
10.1016/j.jhydrol.2023.129268

Multi-phase Flow in Fractured Shale Reservoirs

I develop numerical models to simulate multi-phase flow in porous media. Specifically, I focus on simulating water, gas and oil production from shale reservoirs, representing the multi-scale fracture system in the multi-phase flow model to achieve reasonable estimation of reservoir production.

One challenge in shale reservoir simulation (and perhaps any flow simulation of underground reservoirs) is to model the multi-scale fracture network. Our approach is to model the primary hydraulic fractures using a geomechanics model (GEOS), and represent the secondary fractures as non-uniform dual-permeability elements. We found that the spatial heterogeneity of the fracture aperture is critical in matching the production data for shale reservoirs.

For more info:

10.3390/en14030641
10.1007/s11242-022-01841-0

Parallel Solvers for the Richards Equation

I develop parallel numerical solvers to solve the Richards equation, which is a key component of catchment-scale hydrological models. Common numerical schemes for the Richards equation are mostly iterative (e.g., the Picard scheme and the Newton-Raphson scheme). However, it remains a question if these schemes scale well on multi CPUs or GPUs. An alternative is the non-iterative predictor-corrector scheme, which is robust, but suffers from a small dt. The attached figure shows that as dt increases, more linear solver iterations are required, indicating that we cannot simply use dt to evaluate the efficiency of a scheme. The non-iterative scheme might outperform the iterative schemes when parallelized, if the non-iterative scheme requires fewer linear solver iterations. We are still testing if this is the case...

For more info:

10.1016/j.jhydrol.2020.125809 https://doi.org/10.1016/j.envsoft.2023.105900

Urban Floods and Resillient Cities

I develop numerical models to simulate urban floods. I focus on multi-hazard scenarios such as flooding over previously damaged areas (e.g., by earthquake), as well as on the underground infrastructures (e.g., urban metro stations). These scenarios are rarely studied, but they are closely related to city resillience. I use SERGHEI model for process-based urban flood simulation. SERGHEI is a parallel model that uses GPUs to perform high-resolution hydrodynamic simulations. This work is still on going and more results should be expected in the future.

For more info:

SERGHEI

Research Scientist

Postdoctoral Scholar

Graduate Research Assisstant

Undergraduate Research Assisstant

The University of Texas at Austin

University of California Berkeley

Shanghai Jiao Tong University

University of Michigan Ann Arbor

Our group photo taken on Sept. 2023 (when we were watching Oppenheimer)

Openings

MS, PhD and Postdoc positions are all available. Please contact by email if you are interested. Note that for applicants whose first language is not Chinese, an HSK score is mandatory.