Joaquin Ambia Garrido

                     RESUME

Title     Research Engineering/Scientist Associate

Office Address:   PGE-5102

Email Address:    ambia@utexas.edu

Mobile Phone Number: 832-305-1141

LinkedIn URL: www.linkedin.com/in/joaquin-ambia-garrido-b057b46b

RESEARCH HIGHLIGHT

Most of my work revolves around our software “3D UTAPWeLS”, where we combine formation evaluation projects developed by multiple group members over the years and make them accessible to people in the industry, in a single program with a consistent common ground for all petrophysical aspects of their project. This allows users in the industry to test their hypothesis using our methods, and gives us feedback to further improve our own methods. The software is an excellent channel of communication between industry, and academia.

I also collaborate in the development of new algorithms that we expect to add into the software. My role in many cases, is to make sure the new models have rigorous scientific standards, while being of practical use in the industry.

RESEARCH PROJECTS

Project A

Direct inversion of petrophysical properties. These are properties that are tightly related to measurements i.e. well logs. We are constantly developing  multiple inversion techniques that can be optimized for different cases (e.g. thin layers), most of them based on variants of Monte Carlo, or steepest descent.

Project B

Joint inversion of petrophysical properties. When measurements are made in a well, we are interested in inferring petrophysical properties that will tell us whether there are hydrocarbons, and if they can be extracted. The problem is that the models to go from properties to measurements are not linear, hence not easy to invert.

This problem is usually approached by using the Levenberg-Marquardt method, but the results are not very satisfactory, because the invert problem is degenerate. We are developing a method that instead, uses Bayesian inference where we combine a likelihood function with a Monte Carlo method.

Project C

Parallelization of algorithms. Some of the methods most recently developed in our group can be greatly improved if they use multiple computational cores.