Due to depletion of on-shore and superficial oil reservoirs, and impulsed by recent discoveries of oil reservoirs in off-shore ultra-deep waters, each of the processes and equipment in oil production required further improvements in order to save costs, space and to reduce weight off-shore. One way to accomplish this is without separators and with the use of online multiphase flowmeters. The most used flowmeter is the Venturi tube. Despite Venturi flowmeters having been used in almost all commercial multiphase flowmeters, there is not a single correlation that provides good results for predicting mass flow in each phase, for any flow pattern, mass quality, void fraction and/or fluids properties. Instead, many correlations have been published, based on experimental and/or field data, but the use of these correlations outside multiphase range conditions is doubtful. This study proposes a new methodology that uses genetic algorithms to find correlations that better fit a set of data, which allow determining the mass flow of a two-phase mix through a Venturi tube. For that purpose, binary trees and Prüfer encoding are used to accomplish this implementation. The correlations found in this new methodology provide lower values of RMS error, 1–3%, against correlations proposed by previous authors that show an RMS error range of 5–10%. This technique allows finding further correlations, regardless the number of parameters to be used, at a low computational cost, and it does not require previous information on the behaviour of the data.
Validated Analytical Modeling of Diesel Engines Intake Manifold with a Flexible Crankshaft
