A web based manufacturability agent framework for an E-manufacturing system

by Lungile Nyanga, Andre Francois Van der Merwe, Stephen Matope and Mncedisi Trinity Dewa

Abstract

E-manufacturing enables manufactures to share manufacturing resources hence increasing machine utilization, manufacturing capabilities and capacities. In the paper a manufacturability agent for an E-manufacturing system is proposed. A multi-mode part classification coding system developed from the Opitz coding system is used to create virtual manufacturing clusters using a two step methodology to group the parts and machines. Parts are grouped according to the form code using the similarity coefficients-based method (SCM) and then grouped according to manufacturing process sequence, batch size and processing times using Levenshtein distance measure. Decision criteria for machine selection Analytical Hierarchical Process (AHP) is then given.

1-minute pitch

 

            Paper presenter
Lungile Nyanga
Name: Lungile Nyanga
Organization: University of Stellenbosch, Private Bag X1, Matieland, Stellenbosch, 7602, South Africa
Email: ie@sun.ac.za
Web:

4 thoughts on “A web based manufacturability agent framework for an E-manufacturing system”

  1. In the proposed approach, parts are clustered (according to their characteristics and features) using the Similarity Coefficients-based Method (SCM). Could you explain why you have chosen this particular method?

    1. Thank you for the question. SCM is used because it is one of the the most basic methods for developing part and machine clusters and its flexibility when compared to the other methods. The method used produces a main cluster of parts which are similar due to their form and features, and a sub cluster with similar parts and manufacturing processes. To achieve this clustering should be done independently

      SCM uses three independent steps:
      1. Getting input data;
      2. Determination of the similarity coefficient;
      3. Selection of the clustering algorithm to get the part groups.
      Since the steps are independent we are able to use two different methodologies in determining the similarity coefficients and then use the same methodology to cluster the parts. For example the first stage uses part classification and the second stage uses the manufacturing process sequence to determine the similarity of the parts. Clustering is then done using Average Linkage Clustering (ACL).

  2. Interesting idea.

    You mention that the agent will be implemented at a tooling cluster. How did the tooling cluster originate? Was it centered around a single large company, or did several smaller companies wanted to join forces (or other, e.g. government)?

    1. Thank you for the question. The cluster was developed through a government initiative. The main objective of developing the agent is to increase visibility and resource utilization of the smaller companies in the the cluster which are some times not considered as options by the larger companies. The system seeks to make machinery from these smaller companies to appear as options for selection when larger companies are seeking a company to subcontract to machine a part.

Ask a Question