We have to get service delivery right in South Africa!

The rapid urbanisation in South Africa means that the make-up of cities, in terms of size and distribution, changes continuously. The socio-demographic diversity in South Africa also means that households vary significantly, also in the amount of collectable solid waste they generate. Consequently, estimating the demand for waste collection services is very hard. In response, the (re)design of service delivery, specifically waste collection, needs to be more dynamic so that it maintains high efficiency by minimising cost. 

Residential solid waste collection is an essential but expensive service delivered to South Africans by local authorities. Municipalities only recover a small portion of the collection costs via rates and taxes, and then only for a small segment of its customer base. Optimally designing the waste collection service holds much potential for authorities. Essentially, reducing their costs, both capital in the form of fleet sizing/composition and running expenses in terms of reliable waste collection beats (routes and crew). This project contributes to the cities’ challenge by solving the “waste collection service design problem” as a simulation-based optimisation variant of the Capacitated Arc Routing Problem (CARP). As input, these algorithms require the location and size/amount of waste collected. So, as a first step, this project estimates waste at a disaggregate, household-level using state-of-the-art synthetic populations that are relevant in the South African context. The waste demand then acts as input into the simulation-based optimisation of the service design.

Video 1: Inferring waste collection vehicles' service routes from GPS trace data.

This project forms part of the Waste Research, Development and Innovation (RDI) Roadmap under Grant CSIR/BEI/WRIU/2019/028. There are three fulltime bursaries available:

• two M.Eng (Industrial engineering) or M.Sc (Applied Science) bursaries (valued at ZAR90,000 per bursary per year); and
• one PhD (Engineering) bursary (valued at ZAR150,000 per year).

All three positions are fulltime only and require competence in programming, simulation and optimisation. Knowledge of the statistical programming language R is essential. For the programming, we require knowledge of Java (primarily) and Python. If you are interested, please send your Curriculum Vitae, along with a brief motivation in the form of a cover letter, to [email protected] by 28 February 2020.