About me

I am a first-year postgraduate researcher in AI and part of the third cohort in the Centre for Doctoral Training in Decision Making for Complex Systems, Department of Computer Science, University of Manchester. I am supervised by Dr. Mingfei Sun and Dr. Mauricio Álvarez as part of the Agent lab .

My research will integrate reinforcement learning with CFD workflows, where intelligent agents refine meshing strategies and solver configurations based on numerical performance and best practices extracted from the literature. This is expected to transform traditional computational simulation methods by introducing artificial intelligence into the configuration of simulations. This project will allow me to validate my research with Cummins against cases, including isolating flows, microchannel flows, and problems with deforming meshes.

Research interests

  • Reinforcement Learning
  • Computational Fluid Dynamics
  • Mesh Generation
  • Scientific Machine Learning
  • Numerical Simulation

A finite-difference numerical solution for the temperature field in a cutting tool while it is removing material with its tip and is simultaneously being cooled.

Master's Thesis

Meshing Methodology for Numerical Simulations of Turbulent Flows at Low Computational Cost

Overview: I earned a master's degree at the National University of Colombia, where I focused on computational meshing strategies for a compressible supersonic jet from a nozzle. The objective was to test resolution effects and mesh refinement regions to improve the numerical accuracy of computational simulations based on experimental data and LES simulations carried out by other researchers. By developing optimized low-cost meshes under the UDNS framework, I found that first- and second-order turbulent flow characteristics could be predicted without explicit turbulence models.

Parameters such as mesh size, axial and radial refinement, and the computational domain size played a critical role in accurately predicting key turbulence statistics, including energy dissipation, turbulence intensity, and Reynolds stresses. The academic contribution of this work lies in demonstrating how mesh topology, resolution strategy, and domain design influence the predictive capability of CFD simulations. These findings provide insight into mesh sensitivity and highlight effective refinement strategies, rather than uniform refinement.

I used OpenFOAM to construct structured computational meshes based on meshing parameters. I used ParaView as my primary post-processing tool to analyse turbulence fields and velocity gradient isocontours. Moreover, I developed Python scripts to export numerical data, wrote Fortran programs to perform statistical analyses, and used Gnuplot to plot velocity curves and turbulent intensities. These experiences were crucial in enhancing my computer programming skills in a Linux environment, from setting up simulations and automating statistical calculations to analysing large datasets and presenting results.

  • Degree: MSc Mechanical Engineering
  • Institution: National University of Colombia
  • Year: 2022
Read my thesis →

Background & Previous Courses

National University of Colombia (2019 – 2022)

Master's Degree in Mechanical Engineering

My academic background in mechanical engineering, with a strong emphasis on computational fluid mechanics, provided me with a solid foundation in numerical methods, computer programming, and algorithm implementation for solving applied engineering problems.

Previous Courses

  • Finite Element Applications

    • I studied the Galerkin method for solving elliptic partial differential equations, particularly in elasticity. I became proficient in using Lagrange polynomials to define basis functions and in applying Gauss-Legendre quadrature for numerical integration in triangular and quadrilateral elements of both first and second order.

  • Modelling of Thermal Systems

    • I tackled steady-state and transient heat transfer problems using second-order finite difference schemes and time integration methods for linear and nonlinear partial differential equations.

  • Computational Fluid Dynamics

    • My knowledge of the finite volume method gained through the CFD course has equipped me to simulate complex flow phenomena. Using OpenFOAM, I successfully analyzed first- and second-order statistics of turbulent flows, generated flow visualisations in ParaView, and developed custom post-processing tools in Fortran and Python.

  • Engineering Optimization with Bio-Inspired Techniques

    • I studied multi-objective optimization using the Bacterial Chemotaxis Algorithm, which models chemotaxis behavior mathematically.

  • Combustion

    • This course exposed me to chemical equilibrium for calculating molar fractions and adiabatic flame temperature using iterative methods such as Newton-Raphson, and numerical solutions of chemical kinetics models to compute the temporal evolution of all species in a reaction.

Transient Simulations

This section presents examples of transient simulations related to my research and previous coursework.

Supersonic jet

Computational simulation using the finite volume method in OpenFOAM of the instantaneous axial velocity and temperature fields of a supersonic round jet (Mach 1.4 at exit) generated by a nozzle.

Transient of the Diffusion-convection equation

Numerical solution of the concentration of a substance from the diffusion-convection equation with the finite element method. An implicit method for time integration and a Peclet number less than 1 were employed. In advance, the numerical solution of a potential flow in the computational domain was computed.

Elastodynamics with FEM

Numerical solution of normal stress using the finite element method to solve the elasticity equation with hexahedral elements of second order. Damped oscillation of a tuning fork after it has been moved from its equilibrium position.

Teaching Experience

Mathematical Modelling

National University of Colombia (2021–2022)

Teaching assistant for a Mathematical Modelling course where I guided undergraduate students in programming and mathematical problem-solving, designing quizzes and workshops that presented engaging engineering challenges in the fields of fluid mechanics, heat transfer, mass transfer, and chemical kinetics.

Technical Skills

Programming: Python · Fortran · Bash

CFD: OpenFOAM

Visualisation: ParaView · Gnuplot

Methods: Finite Volume · Finite Element · Numerical Optimisation · CFD