Degree Type

Dissertation

Date of Award

2021

Degree Name

Doctor of Philosophy

Department

Industrial and Manufacturing Systems Engineering

Major

Industrial Engineering

First Advisor

Frank Peters

Abstract

This dissertation will present two methods of how 3D point cloud data can be used to significantly advance two important operations within the metalcasting process. The first is the inspection of the casting surface, and the other is an automation method to replace the current reliance on manual grinding. Currently, surface roughness inspection is performed manually by an operator who compares the surface of a casting with comparators and determines if they are acceptable based on the casting design specifications. The comparators are pictures or physical replicas of representative casting surfaces. As the inspection process is manual, it is very subjective. The low repeatability and reproducibility of the inspection process cause communication problems between foundries and customers as well as within the foundry. This could cause unacceptable castings to be sent to customers, customers falsely identifying acceptable castings as unacceptable, and excessive rework iterations. The dissertation will present an objective method to inspect castings repeatably and reliably. The method will use 3D scan data in the form of point clouds. The point clouds will be used to determine the underlying geometry of castings and then use the distance between point clouds and the mesh representing the underlying geometry to calculate the surface roughness. The second operation covered in this dissertation is grinding of casting surfaces in foundries. Much of the steel casting industry is made up of companies that produce a high mix in low production quantities. This environment precludes currently available automation solutions. Hence there is a heavy reliance on manual grinding. Manual grinding is one of the operations in foundries that has the most ergonomic issues as well as most safety incidents. Currently, the automation of grinding operations is mainly done through fixed automation, where the robot performs the same operation for every part. This requires expensive fixturing for repeatable orientation and programming for every new part. Large production quantities are required to justify the fixturing and process planning tasks. This dissertation will present a semiautomatic grinding solution. In this method, the operator identifies the excess material that needs to be ground and marks it with colored markers on the casting. The casting is then placed in a robotic cell that only requires the casting to be secured and but not the need for exact fixturing. A 3D scanner with a color camera is used to scan the casting, identify the markings, and segment the surface based on the markings. A removal strategy is automatically determined and executed. Overall this dissertation will present two methods to utilize 3D information to improve foundry operations.

DOI

https://doi.org/10.31274/etd-20210609-168

Copyright Owner

Daniel Wilhelm Schimpf

Language

en

File Format

application/pdf

File Size

127 pages

Share

COinS