Selection of the Optimal Critical Velocity for Sand Transport at Low Concentrations for Near-Horizontal Flow

Abstract

This manuscript presents a systematic methodology to determine the optimal critical velocity for sand transport on-the-fly for a given field operating condition. Using publicly-available experimental data on sand transport and sand transport models, the methodology combines data clustering and optimization approaches with statistical analysis. The data clustering algorithm is used to select the representative data points that lie closest to the operating condition, and then, the parameters of the sand transport models are fine-tuned using unconstrained optimization with the representative data. Using statistical analysis, the fine-tuned models are compared to identify the ones that are most applicable and provide accurate velocity predictions for the given operating condition. Upon applying the methodology to a field operating condition, the sand transport velocity for the operating condition, as suggested by the methodology, are consistent.

Quantification and Reduction of Uncertainties for Solids Transport Velocity Predictions at Low Concentrations in Near-Horizontal Flow

“…A methodology was developed by Soepyan et al 3 to identify the appropriate models for a given operating condition. The methodology consists of a data clustering component, a model parameter fine-tuning module, and a model ranking protocol.…”

“…Despite the improvements made by the methodology presented in Soepyan et al, 3 the velocity predictions produced by the five highest-ranked models can differ. In addition, the methodology did not take into account the uncertainties of the operating condition and experimental data, which we have identified as the main sources of uncertainties.…”

“…To overcome these shortcomings, as a first step, this paper presents a modified approach to the methodology described by Soepyan et al, 3 which takes into account the uncertainties of the experimental data. In what follows, the modifications to the methodology presented by Soepyan et al 3 will be described in detail. The velocity predictions suggested by the methodology were compared against the experimental velocity.…”

Confidence in Critical Velocity Predictions for Solids Transport

“…A review of the literature reveals that there are more than 50 solids transport models that can be used to predict the critical velocity (these models are described in the "Methodology" section). For the same input field condition, the critical velocity predictions of these models may differ by orders of magnitude (Soepyan et al, 2012), which makes the selection of the appropriate models for the given field condition difficult. Furthermore, none of the models include information regarding the confidence in their velocity predictions.…”

“…The methodology contains three steps: (1) a data clustering component creates a reduced database, which consists of experimental data points that are representative of the input condition; (2) a model parameter fine-tuning module adjusts the parameters of the models using the reduced database to remove the bias in the models' predictions; and (3) a model screening and ranking protocol uses statistical analysis to determine the most accurate models for the input condition. This methodology is incorporated into the computer program TUSTORM (Tulsa University Sand Transport -Optimization and Ranking Methodology) (Soepyan et al, 2012;Soepyan et al, 2013b), and was implemented using Microsoft Excel and Visual Basic. At the 95% confidence level, the critical velocity predictions of the three highest-ranked models suggested by the methodology produce errors that range from 6.6% to 16% (Soepyan et al, 2013a) when the uncertainties from all sources are ignored.…”