Friday, May 26th, 2023. Time: 12 noon
Hybrid! O.C. Zienkiewicz Room, C1 Building, UPC Campus Nord, Barcelona
ABSTRACT
Reservoir simulation is used by practically all oil companies to assess reserves and optimize field development plans. However, the process of constructing simulation models with strong forecasting capabilities is challenging due to limited data. The talk will focus on the sources of error in simulation: input, mathematical model, discretization, solution and future conditions. We will discuss the relative impact in forecasting based on the author’s experience but also on multiple papers on the topic. With this understanding, we will discuss strategies to maximize the predictive power of models. The counter-intuitive fact is that models with excellent history-match can in fact be very poor prediction tools due to overfitting. Blind testing strategies and automated workflow to avoid overfitting as well as the possible role of stochastic modelling to quantify error will be discussed.
SPEAKER CV
Andino Saint Antonin is a Petroleum Reservoir Engineer with 20 years’ experience who was worked in characterization, simulation and development optimization of some of the largest fields in the world from the North-Sea to Russia and Saudi Arabia.
In his current job, he combines specialist knowledge of reservoir engineering with data analytics of subsurface data and reservoir model building, history match & prediction in order to tackle specific challenges of the fields he works on as an internal consultant.
Alternating roles between hands-on staff and reluctant team leader, his many contributions include multiple successful reserves’ certifications, portfolio assessments and development plans for mayor fields, upwards of 15 full-field simulation models characterized and matched and at least half a dozen machine learning studies.
He holds a BSc & MSc in Petroleum Engineering (Heriot Watt), an MSc in Numerical Methods (UPC) and is studying an MSc in Data Analytics (GaTech)
<!–
–>