Model-based optimization of drilling fluid density and viscosity

Abstract

Optimization of drilling fluid properties is an essential part of cost effective drilling operations and process safety. Currently fluid properties are measured and optimized manually by human engineers with different skills and experience which might lead to nonoptimum drilling fluid properties that deteriorate its functionalities. Automated drilling fluid management is still at an early development stage. Several vendors are actively developing automated skids to measure drilling fluid properties in real time [1] [2], and several authors also have published scientific work on the use of the real-time measurement as a component of automated control systems that dose mud additives automatically to meet the mud specifications or setpoints defined by human engineers [3] [4]. During the well planning stage, the design process of mud specifications is carried out by engineers checking several scenarios using well planning software and their experience to come up with drilling fluid specifications. When hole cleaning and/or borehole stability conditions change during the actual drilling process that warrant updates or changes of drilling fluid properties, the specifications are updated in an ad-hoc manner, relying on the skills of human engineers. This thesis focusses on the development of a model-based optimization module for drilling fluid properties to help engineers in the planning and drilling phase to automatically derive drilling fluid specifications that meet the hole cleaning criteria, and satisfy the downhole pressure requirement and constraints set on the operating ranges of drilling parameters. The optimization framework will use proxy models derived from well hydraulics software that predicts cuttings concentration and downhole pressure as a function of the drilling fluid properties. Three objective functions for the optimization module are given as examples in this thesis. The first two objective functions deal with the hole cleaning criteria while the last one is a cost function that combines the cost of hole cleaning and downhole pressure management. The optimization module has been tested on a case study based on real field data. Given an objective function, multiple constraints, and proxy models, the module takes only a few seconds to find the optimum mud property values and drilling parameters such as flow rates, rotary speed and rate of penetration. A benchmark with the field data shows that the optimum drilling fluid properties and parameters result in significant improvement of the hole cleaning state while the downhole pressure requirement and constraints on the drilling parameters can still be satisfied. When a cost function is defined as a combination of hole cleaning and downhole pressure management, the module also gives a quantified benefit of the trade-off between maximizing hole cleaning and minimizing losses. Since this module can perform optimization very efficiently compared to the ad-hoc processes done by human engineers, this module may be of significant value for operating units to use in the planning and drilling phase and also in the future as an outer optimization loop for automatic drilling fluid control systems.