Banca de DEFESA: DILAYNE SANTOS OLIVEIRA

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : DILAYNE SANTOS OLIVEIRA
DATE: 02/08/2021
TIME: 15:00
LOCAL: Auditório Pós-Graduação Engenharia Civil - CTG - UFPE
TITLE:

FRACTIONAL FLOW MODELS BASED ON WELL PRODUCTION DATA FOR WATERFLOODING OPTIMIZATION

KEY WORDS:

CRMP. Fractional Flow Models. Optimization. Waterflooding. Optimum Reservoir Management.

PAGES: 96
BIG AREA: Engenharias
AREA: Engenharia Civil
SUMMARY:

This work applies the Capacitance and Resistance Model based on the producer (CRMP) and the fractional flow models Koval and Gentil to determine water and oil production in reservoirs with waterflooding. These models only use the history well rates data as inputs. In addition, it presents a fractional flow model called Kogen, which is formulated as a non-linear constrained problem and solved with the Sequential Quadratic Programming (SQP) algorithm. The CRMP, Koval, and Gentil parameters are calculated using a Nonlinear Least Squares (NLS) algorithm. The comparations between each model performance are made with an optimization problem to maximize the reservoir rentability. The objective function is the Net Present Value (NPV) of the reservoir, and the control variables the injection rates. This optimization problem is solved applying the SQP method, and gradients are approximated with an ensemble-based method. To validate the proposed strategies, two reservoir models are used: Brush Canyon Outcrop and Brugge. The results show the capacity of CRMP to forecast the producer’s liquid rate and to indicate the influence of the injector on producer wells quantitatively. They also validate the applicability of Koval, Gentil, and Kogen models to estimate oil and water rates. It is observed that Kogen forecasts oil and water rates with more accuracy for wells with different water cut levels, while the other models show limitations. Comparing theproxies CRMP+(fractional flow models) performance with a high fidelity optimization process, it is observed that CRMP+Kogen presents better results. To BCO, the final VPL calculated with CRMP+Kogen is increased by 20.2% e 2.65% in comparison to Koval and Gentil models, respectively. This proxy also generates a good start point to high fidelity optimization processes, increasing its results and decreasing the number of simulation executions. The use of the proxies generated to optimization problems is demonstrated with satisfactory results to forecast oil and water productions.

BANKING MEMBERS:
Interno - 1130915 - BERNARDO HOROWITZ
Externo ao Programa - 2695448 - LEONARDO CORREIA DE OLIVEIRA
Interna - 2223613 - SILVANA MARIA BASTOS AFONSO DA SILVA