Discrete event simulation based study of a PLM software network

Chaudhry, Shashank

Permalink

https://hdl.handle.net/2142/49829 Copy

Description

Title

Discrete event simulation based study of a PLM software network

Author(s)

Chaudhry, Shashank

Issue Date

2014-05-30T17:19:55Z

Director of Research (if dissertation) or Advisor (if thesis)

• Dutta, Debasish
• Patil, Lalit

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Discrete Event Simulation (DES)
• Product Lifecycle Management (PLM)
• Simulation
• Network Simulation

Abstract

A model for describing the data storage and data flow in a Product Lifecycle Management (PLM) software network has been developed in this study. Discrete event simulation (DES) has been used to represent this flow of data and queuing models have been used to represent the hardware components of the network. The aim of the study has been to accurately represent the true behavior of the network during heavy operational load and to develop and run scenarios to increase software user productivity. A flexible tool has been developed for this purpose, which makes it simple to test numerous scenarios and carry out bottleneck analyses. The tool uses three tiers of code to accurately model the PLM software network, the hardware tier, the software tier and the application tier, which includes business insights that are being simulated. The studies show that increasing the cache size in the network can improve the response times only to a certain extent despite significant increase in the number of cache hits and that the response time can be negatively impacted if the number of non-critical users in the network increase. We observe that rather than expensive hardware upgrades, which show modest improvement in response time, to the tune of 5 - 8%, other more innovative business oriented approaches such as allowing the users with deadlines to start work earlier or allowing the cache to be preloaded show much more promising results, with as much as 50% decrease in response time. Statistical validation of the model and the output results has also been carried out.

Graduation Semester

2014-05

Permalink

http://hdl.handle.net/2142/49829

Copyright and License Information

Copyright 2014 Shashank Chaudhry

Owning Collections