Real time prediction of suspended solids in drilling fluids

Abstract Drilling operations require the usage of drilling fluids to perform not only cleaning and cooling functions, but also to control the hydraulic pressure inside the oil well. Monitoring the quantity of solids suspended in such fluid is important because it is related to density and apparent viscosity. Both physicochemical data influence the determination of the frictional pressure loss in annuli, which is the first step of calculation while determining the hydraulic pressure. Automating the measurement of solids may not only improve drilling performance but also mark the initial steps toward attaining a fully automated drilling operation. This work measured real-time data of ultrasonic attenuation, velocity of sound, density and apparent viscosity in several distinct suspensions. Such data was used to determine the content of solids in water-based drilling fluids. The results showed that ultrasonic attenuation and density were proportional to the quantity of solids suspended (The largest density range evaluated was from 0.98 to 1.6 kg/l, water to heavy slurry; the concentration of solids suspended varied from zero to 1000 g per liter of suspension, which caused the ultrasonic attenuation range from 6 to 32 dB). The apparent viscosity appears to impact ultrasonic attenuation (1 cP to 135 cP caused the ultrasonic attenuation to vary from 6 to 22 dB, approximately). The velocity of sound was proportional to the quantity of dissolved solids (1500 m/s were raised to 1750 m/s, approximately, when 300 g/l of sodium chloride was added in water.) Results also showed that Newtonian and non-Newtonian suspensions vary due to the fact that the shear rate must be known for non-Newtonian fluids in order to correctly determine the solids content. To predict the solids content of water-based drilling fluids from real time data, the study successfully implements a statistical model. When the predicted and experimental values were compared in a case study, the proposed model presented accuracy of ±10%.