QKP Well Placement Optimizer

A computational framework for optimizing well placement under uncertainty using Gaussian Process models and Quadratic Knapsack Problem solvers

Learn More View on GitHub

Project Overview

The QKP Well Placement Optimizer implements a mean-variance portfolio optimization approach for petroleum well placement:

  • Uses Gaussian Process models to capture spatial correlations in well productivity
  • Solves a Quadratic Knapsack Problem (QKP) to select optimal well locations
  • Employs Randomized SVD (RSVD) for computational efficiency
  • Allows trade-off between expected returns and risk via a risk aversion parameter (λ)

Visualization of well placement optimization

Key Features

Gaussian Process Models

Captures spatial correlations in well productivity with multiple variogram models.

QKP Solver

Advanced optimization algorithms to maximize returns while managing risk.

RSVD Approximation

Low-rank approximation techniques for computational efficiency.

Risk Management

Flexible risk aversion parameter to balance returns and uncertainty.