RESEARCH: THINK OUT OF THE BOX!
Motivations for Research
The fundamental motivations for my research are to understand the analytical structure of industrial problems, to conduct research that seeks to solve these problems analytically as best possible and to explore how research in other domains such as VLSI CAD, Architectural Design, Molecular Chemistry, Genetics and Pattern Recognition could suggest novel solution techniques in my domains of research. Special emphasis is placed on studying problems that extend conventional results and thinking such as:
Production Flow Analysis (PFA) is a systematic method for the design of Cellular Manufacturing Systems. It consists of four nested (or hierarchical) steps - Factory Flow Analysis, Group Analysis, Line Analysis and Tooling Analysis. Unfortunately, since its introduction into the literature in the 1960’s, it has been implemented manually, thereby making it cumbersome to use in practice. I am leading a project to develop a library of algorithms – PFAST (Production Flow Analysis and Simplification Toolkit) - to automate the manual steps that have impeded the industrial deployment and effective use of PFA. PFAST is at the core of FACTORY2000, a proposed software toolkit for the automated analysis, design and evaluation of the next generation of factory layouts. This toolkit will be a primary mechanism for seeking funds to support my research.
There has been a mindset in the area of Cellular Manufacturing Systems that cells must be designed with machine duplication to eliminate all intercell flows. I have studied the bottleneck problem - duplication of machines among several cells - and proposed a new idea viz. Hybrid Cellular Layouts. These are novel layout configurations that reduce the number of machines that need to be physically duplicated in non-adjacent locations a facility.
The vast majority of papers on Facility Layout have focused on models and algorithms to design Functional Layouts. Since these algorithms use the From-To chart as input, they are unable to guide machine duplication decisions. I have shown that, if we use the operation sequences as input i.e. we do not aggregate them into a From-To Chart, then we can use string clustering and matching algorithms to duplicate machines in a layout for increased routing flexibility while adopting partial cellularization in the same layout.
The traditional thinking in Facility Layout has been that a single type of layout must be designed for a manufacturing facility. Based on a case study done for Motorola Inc., I have developed a novel idea – Layout Modules – whereby any facility layout can now be decomposed into a network of modules. Usually, a module has one of the three traditional layouts discussed in textbooks – Functional, Cellular and Flowline.
To date, the three areas of research - Group Technology (GT), Production Flow Analysis (PFA) and Computer Aided Process Planning (CAPP) - have not been integrated in order to incorporate process and manufacturing-related knowledge in the layout design process. I am investigating a new approach to facility layout - Feature-based Facility Layout (FFL) – that integrates GT and CAPP.
Research Contributions - I: Software for Production Flow Analysis
Research to develop PFAST, a library of algorithms to automate the classical manual method of Production Flow Analysis (PFA) for cell formation, began with the paper by Irani & Khator (1987). The three papers - Chen & Irani (1993), Chandra, Irani & Arora (1993) and Arvindh & Irani (1994) - explored various versions of a Machine-Part Matrix Clustering (MPMC) algorithm to automate the second stage in PFA - Group Analysis. A Simulated Annealing version of this permutation generation algorithm was subjected to exhaustive testing in Shargal, Shekhar & Irani (1995). A first attempt at demonstrating the commercial viability of this algorithm using a commercial package, STORM, was demonstrated in the book chapter by Irani & Ramakrishnan (1995). The NSF GOALI grant (DMI-9521278) supported further development and industrial testing of the algorithm. Preliminary results from this software using industry data have yielded excellent results, as reported in Daita, Irani & Kotamraju (To appear in IJPR).
Summary of Research Contributions:
Proposed a comprehensive library of algorithms to automate Production Flow Analysis (PFA)
Sponsored by Engineering Animation Inc. to develop a matrix clustering algorithm for cell formation as a potential add-on to their VisFactory product
Research Contributions - II: New Concepts in Cellular Manufacturing
My doctoral dissertation and the two papers - Irani, Cohen & Cavalier (1992) and Irani, Cavalier & Cohen (1993) - proposed the idea of designing Virtual Cellular Layouts instead of the age-old concept of independent cells. The subsequent paper (Arvindh & Irani, 1994) provides a unique classification of the cell design sub-problems. The NSF Grants (DMI-9523809 and DMI-9734815) have supported research in this area.
Summary of Research Contributions:
Developed the idea of Hybrid Cellular Layouts to simplify the machine duplication problem
Developed pattern recognition methods for Facility Layout that use operation sequences
Explored the use of Subgraph Isomorphism for Facility Layout
Developed the idea of Layout Modules to custom design factory layouts
Research Contributions - III: A Manufacturing Engineering Approach for Facility Layout
I am hoping to further develop this research idea. Significant research publications in this area are Irani & Kimbler (1986), Irani et al (1988), Smith et al (1992) and Irani, Koo & Raman (1995).
Research Contributions - IV: Quantitative Models in Manufacturing Engineering
Several papers represent my interest in developing quantitative models to automate the manual and/or ad-hoc methods used in Manufacturing Engineering. The two papers on Tolerance Charting - Irani, Mittal & Lehtihet (1989) and Mittal, Irani & Lehtihet (1990) - as well as the paper on feature-based machining sequence planning - Lakkaraju, Raman & Irani (1992) - demonstrate this research interest. In particular, models are being developed for Variety Reduction, Design of Machining Centers and Setup Reduction.