Research

How biofilm and elemental sulfur reorganise inside a fixed-bed reactor, and what that does to nitrate removal from groundwater.

Nitrate in groundwater is a growing public health concern, and elemental sulfur (S⁰) based autotrophic denitrification is one of the most cost-effective alternatives to treat it. The catch is that the bed is not static: sulfur is consumed, biofilm grows and the pore space changes. That coupling between biology and flow is what I study.

The project combines three-dimensional CFD simulations of the reactor with advanced digital imaging techniques that track, non-destructively, how the internal structure of the bed evolves. The goal is a predictive model that can be used to operate and scale these reactors, instead of tuning them by trial and error.

Validating what simulation says about the real flow inside a packed bed, using residence time distributions.

A CFD model of a packed bed is only as good as its validation. A central part of my work has been developing methods that confront simulation with experimental residence time distribution (RTD) measurements — an observable that integrates everything happening inside the reactor: channelling, dead zones, mixing.

I have applied this approach to biotrickling filters for waste gas treatment, combining computed tomography of the packing material with CFD simulation, and developed hybrid section-based models that capture the effect of the packing on hydrodynamics at a reasonable computational cost.

Try the simulation →

What happens to flow, mixing and reaction when biomass progressively clogs a porous medium.

In collaboration with Dr. Verónica L. Morales’ group at the University of California, Davis, I study bioclogging: the progressive accumulation of biomass in a porous medium and the flow rerouting it causes. It is a central phenomenon in aquifers, bioremediation and soil recovery, and it remains hard to predict.

The work combines micromodel experiments with pore-scale simulation, and aims to understand how biofilm permeability and the accumulation pattern control reactive transport. Two manuscripts in preparation and presentations at AGU and InterPore came out of this line.

Where this is going: turning expensive simulations into fast models that can drive operation and design in real time.

The three lines above converge on a single medium-term goal: building digital twins of reactors — computational representations fast and faithful enough to accompany operation and design in real time, fed by high-fidelity CFD simulations and imaging data.

Further ahead, I want to extend these tools to natural aquatic systems — lakes and rivers — where the same couplings between flow, biology and transport govern water quality.