Predicting non-deposition sediment transport in sewer pipes using Random forest

Area of Science:

• Environmental Engineering
• Hydraulic Engineering
• Water Resource Management

Background:

• Sediment transport in sewer systems is a critical issue affecting infrastructure performance.
• Existing models for predicting self-cleansing velocity have limitations in accuracy and applicability.

Purpose of the Study:

• To develop and validate a novel Random Forest (RF) based model for predicting the self-cleansing velocity in sewer pipes.
• To compare the performance of the proposed RF model against ten existing literature models.

Main Methods:

• Implementation of a Random Forest model using experimental data from existing literature.
• Evaluation of model accuracy using multiple observed datasets.
• Comparative analysis against ten established models for sediment transport in sewers.

Main Results:

• The developed RF model demonstrated high prediction accuracy across the entire dataset.
• RF model predictions significantly outperformed those of other models, particularly for the non-deposition with deposited bed criterion.
• Volumetric sediment concentration was identified as the most influential parameter for self-cleansing velocity prediction.

Conclusions:

• The proposed Random Forest model offers a superior method for predicting self-cleansing velocities in sewer design.
• Accurate prediction of self-cleansing velocity is crucial for preventing sediment deposition and ensuring sewer pipe longevity.
• Understanding the impact of volumetric sediment concentration is vital for optimizing sewer system performance.