Evaluation of characteristic diameter on barite settling in drilling fluids by Monte Carlo method

Abstract

Several researchers have discussed the settling behaviour of particulate materials because of the challenges and applications in different engineering fields. Particularly, settling of such weighting agents as barite, commonly added to drilling fluids, can cause problems in oil production, either on drilling processes or in production. Typically, sedimentation models consider a single value of particle size, which in this work, we refer to as characteristic diameter. This particle size can be chosen by statistical methods, such as the Sauter mean diameter, or it can be extracted directly from particle size distribution such as its mean, median, or mode. There are some conventions by how to choose the characteristic diameter; however, discarding the whole size variability from the particle size distribution is unwise. Regarding the literature, only a few works have, to the best of our knowledge, discussed the effects of this choice on the models' performance. This work then presents a new perspective on evaluating the uncertainty of the particle size distribution and choosing the best characteristic diameter, using a novel method based on Monte Carlo simulations. To evaluate the method, research team chose a barite settling model (recently published) for drilling fluids. Experimental data of barite concentration in batch settling test supported the Monte Carlo simulations. Five characteristic diameters were evaluated. Results showed different column heights have different characteristic diameters which differed from classical values such as Sauter's diameter by up to 200%, producing different model output variables. New particle size also affected estimation of remaining parameters and changed predicted properties such as media permeability by up to 90%. This work opens a discussion on how well-represented a particle size distribution is when researchers choose only a characteristic diameter is chosen without evaluating the whole distribution.