Summary Among all the systems that make up a drilling operation, the production and correction of drilling fluid can be considered the heart of the process. Among the main objectives of the drilling fluid are to cool the drill bit and maintain the pressure gradient inside the drilling well, which is done by controlling its density. Another important function is transporting the cuttings from the bottom to the surface and keeping them in suspension in case of stoppage, which directly depends on the viscosity of the drilling fluid. Density and viscosity must be constantly maintained within an operational window, and failures can lead to serious accidents, even the loss of the well. Currently, this control is done manually: An operator collects samples of the fluid and takes them for analysis in the laboratory and subsequently makes the necessary corrections by manually adding products to the fluid. To reduce process dead time, keep personnel on board, and increase operation safety, a control and monitoring system is necessary. Fuzzy logic was chosen because it can be combined with classical methods, is cheap to develop and implement, and can be customized in terms of natural language, capturing the knowledge acquired by operators from equipment operation, bench tests, etc. This work aimed to develop a novel real-time monitoring and fuzzy-based system for simultaneous control of the apparent viscosity and density of non-Newtonian fluids, dealing with the inevitable interactions between them in a pilot experimental unit. A pilot plant was built to evaluate the fuzzy system approach for modeling and controlling of density and apparent viscosity of drilling fluids. The pilot flow loop comprises a mixing tank, solids vibrating feeders, and a water-dosing pump. The unit was instrumented with online sensors to measure fluid density, temperature, flow rate, differential pressure, and viscosity. The apparent viscosity and density of the non-Newtonian fluid were controlled by manipulating the dosage of carboxymethylcellulose (CMC), barite, and water. The proposed methodology was compared to a classical proportional-integral-derivative (PID) controller in servo and regulatory scenarios for apparent viscosity and density. The results showed that the fuzzy controller dealt adequately with the effect of variable interactions, keeping both variables within their setpoint ranges, demonstrating the ability to control them individually despite their interactions. These results also showed that the fuzzy-based controller could easily be integrated into a diagnostic-predictive monitoring system to control fluid properties, accomplishing setpoint changes and rejecting undesirable disturbances presenting a maximum overshoot of 7.5% for apparent viscosity and 0.3% for density.
Read full abstract