April 1998
Practice Lesson:
OR: Out of Closet, on to the Factory Floor
By Nancy Bistritz
Imagine a manufacturing environment where mistakes can be caught instantaneously and progress can be monitored constantly, thus enabling productivity of new products to move more smoothly. Sound impossible? Thanks in part to a new use of operations research, it's the standard operating procedure at the Nortel (Northern Telecom) Wireless Networks manufacturing plant in Research Triangle Park, N.C.
In 1994, Nortel approved plans for a manufacturing facility - Wireless Networks Raleigh (WNR). The plant would be dedicated to manufacturing advanced cell-site base station and antenna equipment based on a personal communications services (PCS) design protocol known as global system for mobile communications (GSM). The business units that designed products manufactured by WNR are based in France and the United Kingdom, thus communication (specifically, feedback) from these locations would be critical to the success of the new facility [Brinkley et al.].
In order to provide customers with reliable, high-quality products at the lowest possible cost, WNR needed state-of-the-art manufacturing practices including just-in-time, demand-flow production control philosophies and statistical quality control. In order to support each of these practices, management realized it needed an "information technology backbone capable of tracking and routing product through production and providing real-time feedback regarding factory inventory and quality metrics" [Brinkley et al.].
A factory (regardless of the product being made) typically uses information technology to automate processes, to track products and to streamline operations. The use of information technology often mandates writing code or developing an automated software solution that is specific to that factory's needs - all of which can cost a company a lot of money. Traditionally, an operations researcher's role in the factory consists of consulting and performing analyses. Nortel was about to change that.
Paul Brinkley is currently a senior manager of information technology strategy with Nortel. In 1995 as manager of Operations Research with Wireless Networks, he believed that existing technology (existing 3rd party software) coupled with OR techniques could drive a factory's production cycle and make it more efficient in the long run.
"We wanted to prove that a better approach was to take OR specialists and use them to implement systems inside a factory with a focus on buying software whenever possible and writing as little code as possible," Brinkley said. "We focused on integrating systems and using them to deliver immediate manufacturing value, in addition to enabling future operations analysis and optimization."
Selling the Idea
In April 1995, Nortel recruited a team of recent graduates and experienced OR practitioners and engineers. Nortel gave the team responsibility for "implementing factory data collection, decision support and product-and-process control systems" [Brinkley et al.].
According to Kristopher Haag, one of the recruited OR specialists, the idea was to empower the workforce with the tools to do their jobs more efficiently, while also providing process engineers with the ability to conduct OR-type analyses. "Each system that was to be implemented was assigned to an OR project specialist. The OR project specialists were in charge of planning and executing the implementation of a quality tracking system, a statistical process control system and a decision support system so that we could manufacture products within a period of months instead of a years."
"We also wanted the workers to be able to understand the data and quickly analyze it. We wanted the engineers and analysts to be able to quickly perform OR-type analysis based on real-time data."
Louis Clement has been with Nortel for 17 years. In 1995, he was the plant manager for the Wireless Division in the Nortel plant in Raleigh. Clement admits that while he was open to new ideas, he was somewhat skeptical about using OR techniques in a manufacturing environment.
"There was this tradition of having a little OR group upstairs in the corner that would propose something or do a report or indicate direction, but we hadn't put them on the front lines," he said.
The OR team implemented systems that were designed to achieve three primary objectives: to control production and inventory; to ensure quality and reliability; and to provide real-time decision support. In addition, the team wanted to provide accurate information to their product designers located around the world. In order to do the latter, the team decided that using the Intranet would be the most effective method.
"There were people all over the world who were going to be looking at this data," Brinkley said. "I didn't want them to have to go buy special hardware or software to access our data. Basically, anyone, anywhere could see the data. We didn't have to sell them really hard on using the tools because they were already immediately available to them."
The team admits that at the time, using the Intranet in this fashion was a relatively new concept. However, doing so not only allowed product designers to quickly spot design flaws, it enabled factory floor workers to pinpoint problems. Located all around the plant were Unix terminals that linked all the testing sets and assembly machinery that provided feedback to employees in real-time. According to Folger, every mission-critical step of a product had a terminal.
"Our workforce had access to everything - everyone had the same level of access," Folger said. "Initially, after we installed the terminals, it made their lives a little bit more complicated with the data entry they had to do, but after a while, they began to see it was working for them because they could see when mistakes were happening. We could trace everything."
Brinkley attributes much of the project's success to Clement and his willingness to accept a new way of doing things. Another factor that helped the OR team gain acceptance was the manner in which the team sold the idea - a critical step which Brinkley says is often difficult for the typical OR person to do.
"Typically, an OR guy comes into a factory and starts selling a method to a plant manager," Brinkley said. "A plant manager doesn't care about a method. He wants to see dollars and cents. An OR person, if they want to be successful, better start talking in terms of the bottom line. Go do it however you have to do it. But what I want to hear about is inventory, cycle time and how I can save money."
The people's perception
Selling OR techniques to the plant manager was one thing; selling it to the factory workers and training them how to use the system was a completely different story. "With any new system, you find resistance when you first introduce it. It's kind of hard to expect that everything would run smoothly and that everyone would embrace your new process and the work that they would have to do," Folger said. "For the manufacturing associates, sometimes it was a little foreign. The best part, though, was when people would actually get excited about all of the data they would input, and then instantly see a monthly summary or a weekly summary."
Brinkley believes that by using OR in this way, the workforce gained a new knowledge and a new appreciation for their jobs. Prior to the change, managers were used to preparing reports manually. Using the new system, data was gathered, and reports were prepared automatically and then made available over the Intranet.
As far as training was concerned, the OR teams set up workshops for the shop floor workers. Each workshop explained how to use the new systems. Workers were shown how to enter data, how to access reports and how to respond to changing conditions in the processes. Many employees took the computer skills they learned and went on to higher jobs within Nortel.
The benefits
It did not take long for the Nortel plant in Raleigh to see a significant savings in terms of implementation cost as well as annual support (staff, software license) cost. Using the OR approach, the plant saved approximately $1.6 million in implementation costs and approximately $2.2 million in annual support costs. Additionally, when doing new product testing, Nortel had real-time, online hookup to its testing equipment that provided statistical distribution information to the product designers all over the world. Feedback from product designers in England and France now occurs in a matter of minutes, not days.
Because of the success of the project within the factory, WNR has implemented OR methods elsewhere, specifically in materials planning and procurement. "I think we have managers at WNR who now understand and see the power of OR as a discipline," Brinkley said. "It's been a change in thinking. We brought science into manufacturing. Then we moved it, and brought science into materials and procurement. There's suddenly this awareness that this field (OR) has a lot to offer."
It works, so why isn't everyone doing it?
Brinkley credits the novelty of the idea of using OR in a manufacturing environment to the project becoming an Edelman Prize finalist. So why isn't it done more often? Because most factories don't have the data available to do the analyses, a fact that comes as a shock to new OR grads.
"Most students coming out of school think, 'I'm going to get to do OR!' " Brinkley said. "Instead, they come into a factory and find that most of their peers are off working simple spreadsheets, crunching numbers and doing no real OR work. OR has been shut out of the factory because many managers do not have a true understanding of OR or how it can work inside the factory."
For recent OR graduates - Folger and Haag - coming into a factory and being able to practice "true" OR was an experience unlike any other.
"For me, this project definitely underscored the importance of OR in the factory," Folger said. "In school I concentrated a lot on manufacturing. For me it was right on, it seemed like the natural extension. I think more OR people can move into those roles where they can provide a lot more value in a short amount of time. I definitely learned that OR's place can be carved out in a factory setting pretty easily."
References:
1. Brinkley, Paul; Stepto, David; Haag, Kristopher; Folger, John; Wang, Kui; Liou, Kuanlian; Carr, David; "Redefining Factory Information Technology : An OR-Driven Approach," Nortel Report, 1996.
Nancy Bistritz is managing editor of OR/MS Today.
Reader Service Form
OR/MS Today copyright © 1998 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com
Web Site © Copyright 1998 by Lionheart Publishing, Inc. All rights reserved.