Eventos Anais de eventos
ENCIT 2022
19th Brazilian Congress of Thermal Sciences and Engineering
MACHINE LEARNING MODEL TO DETERMINE THE CURVE-FITTING EQUATION FOR PREDICTION OF THE STARTUP OF NON-BROWNIAN SUSPENSIONS IN NON-NEWTONIAN FLUID
Submission Author:
LUIS HUMBERTO QUITIAN ARDILA , PR , Brazil
Co-Authors:
LUIS HUMBERTO QUITIAN ARDILA, Yamid José García Blanco, Angel De Jesus Rivera Jimenez, Eduardo Matos Germer, Admilson Franco
Presenter: LUIS HUMBERTO QUITIAN ARDILA
doi://10.26678/ABCM.ENCIT2022.CIT22-0406
Abstract
The rheological behavior of complex fluids such as non-Brownian suspensions is currently a required topic in the cosmetic, food, and petroleum industries. Since in processes where flow start-ups occur, high pumping pressures are necessary to exceed the gel strength. The interactions between the particles are one of the main problems for correctly obtaining the rheological properties of the non-Brownian suspension, especially in viscoelastic matrices with yield stress. Therefore, it is necessary to understand the transition from solid to liquid phase that occurs during flow start-ups in non-Brownian suspensions since variables such as the fraction of solids and particle size can influence the measurements. In the current work, controlled shear rate rheometric tests were conducted to investigate the yielding of Carbopol® gel solution with different spherical particles of sizes between 20 and 50 µm. The creep tests are performed with different constant shear stress values. A model based on machine learning is proposed, where the shear rate is considered. In order to find the best-supervised fit of the data obtained through rheological measurements. The analysis and adjustment of curves through machine learning allows to have a clearer idea of the variables that influence the behavior during the start of the flow. The results obtained through the machine learning model will facilitate the analysis of the rheological data. In addition, further, understand the behavior of the gel force during the transition from solid to liquid. In addition to having a dynamic and adaptive model that determines the maximum pressure peaks during the flow start-up processes. The findings will also contribute to the understanding and modeling of non-Brownian suspensions in the viscoplastic matrix. This can be seen in the standard pumping processes of various industrial processes, especially well drilling in offshore operations. In addition, it can open discussion about the problems that originate in the measurement of this type of suspension.
Keywords
machine learning, Non-Brownian suspensions, Non Newtonian Fluids, Startup Flow, Rheology
