Project Durban Skhaftin: Container movement monitoring

Project Information

  • Stakeholder: Transnet Durban Container Terminal
  • Students: Lerato Langa (UP), Judith Mathebula (Wits), Vongani Maluleke (UCT), Takalani Mamafha (UJ)
  • Project Lead: Dr. Quentin Williams
  • Project Mentors: Pelonomi Moiloa, Linda Khumalo
  • Year: 2016/2017

Project Description

The Transport Durban Container terminal is responsible for loading and discharging of containers to and from ships which need to be stored in multiple yards within the terminal and from there they are moved onto trains or trucks and transported throughout South Africa and other countries. Project-Transnet Port Terminal (TPT) is a project that aims to better understand the movement of containers and their associated income/ expense and then identify key factors where improvements that can be made to minimize the costs. It has the objective of assisting Transnet Operational Management in gaining insight how containers movements can be optimised given financial and physical constraints, where the critical areas for improvements and how maintenance can be more effectively.

The TPT team focused developed and designed a dashboard visualising the financial model , using movement data and weather data from November 2015 to January 2016 and financial model (developed by Dr Daan Velthauz from CSIR). The TPT’s dashboard consisted of four components; the homepage indicating the general movement statistics and linked to a current weather meteogram as well as movement time curve; Weather implications indicating the effect that the weather has on the terminals activities; Cost analysis comparing cost and revenue at the berth level to the to the cost and revenue incurred throughout the terminal; and Port movement reporting on the activity occurred in the terminal and the contribution of each berth.

They further focused on Predictive Analysis performing exploratory data analysis for pattern recognition, significant features and corresponding targets that were identified. They build two models that perform predictions using machine learning techniques predicting the next location given move kind, category, time/ hour of the day, current location of the container, feeder type, unit type, operational line; and also predicting the cost of moving a container given move kind, weather conditions and time/ hour of the day.

Student Remarks

The students felt that being from different disciplinary fields the project final output was delivered with the help of one of the Students who has worked previously on the DSIDE Program and also the help of their mentors who were always available for assisting them. The TPT team learned to work together as a team and also learned to perform under pressure. They strongly believe the skills they have learned will come in handy in the near future.

Author: Team + Nolihle Gulwa, B Tech Journalism, Walter Sisulu University.