University of Twente Student Theses

Login

Solving the dynamic container relocation problem to minimize transportation costs using approximate dynamic programming

Bosch, R.R. (2022) Solving the dynamic container relocation problem to minimize transportation costs using approximate dynamic programming.

[img] PDF
2MB
Abstract:This research is done on container logistics in terminals. Container terminals are busy interfaces between modes of transport with uncertainty in the arrival and departure of containers. Retrieving a container with other containers above it requires the removal of all containers on top, called reshuffles. These introduce inefficiency and should be avoided. This problem is complex as the number of available locations is too large to evaluate. This problem is solved by formulating the terminal and container schedule as a Markov Decision Problem, which is then solved using Approximate Dynamic Programming. The solution method consists of two parts: an approximation of future costs in any given state of the terminal, and an optimization method for finding the best decision at a single stage of the Markov Decision Problem. The value approximation is done using a Basis Feature Function and a Neural Network. Both methods outperformed heuristics from literature, but the Basis Feature Function provided the best results. For the optimization at a single stage two methods are developed. These are a Partial Search Tree and a Mixed Integer Program. The Mixed Integer Program provided the best results but was prohibitively time-consuming to solve.
Item Type:Essay (Master)
Clients:
Cofano, Enschede, Netherlands
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:31 mathematics, 54 computer science, 55 traffic technology, transport technology
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:https://purl.utwente.nl/essays/93315
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page