The Goal

succeeded in institutionalizing the usage of the know-how written in The Goal. I mean institu- tionalizing it across the board.

EG: Given that General Motors is the largest manufacturing company in the world, you should get an outstanding proof by interviewing

347

Kevin Kohls. (Eli Goldratt interview to be continued.)

Interview with Kevin Kohls General Motors

Director of Throughput Analysis and Simulation for North American Assembly Plants.

DW: What drove you to seek help from The Goal?

KK: It goes back almost 15 years, when I was starting off as a controls engineer at the Cadillac Detroit- Hamtramck assembly plant, just re- turning from Purdue University after completing a masters degree in electrical engineering. When I left a year and half earlier, the plant was just starting production. When I returned, they had yet to hit their production targets; in fact they were far short. As you might imagine, everyone was frustrated about not hitting these targets, and there was a lot of effort being expended to improve the system, with minimal results.

I was frustrated as well. The solutions I was putting in place rarely had a significant impact on the production of the plant, and it wasn't clear why. About that same time, Dave VanderVeen from GM Research made a presentation to Larry Tibbetts, who was then plant manager. Dave was promoting a research tool that he said would help improve throughput in the plant. Larry was very impressed, and asked me to go see Dave to find out if we could use this tool at Hamtramck. When I went down to the Research Building at the GM Tech Center in Warren, Dave explained what a bottleneck was and how his tool identified it. He handed me a copy of The Goal and said if you want to understand bottlenecks and how to improve throughput, this is the book to read.

I took the book home and started to read it right away. The first thing that surprised me was that it was written in novel format. The second was how much I could identify with what was happening in Alex's plant. I finally had to put it down at 2 A.M. so I could get some sleep, but I finished it the next day. I wanted to apply the concepts immedi- ately, so I began collecting data from the systems we had, and putting it into the bottleneck program. After about a week of effort, I was fairly certain I had found the bottleneck. The scary part is that it was not 20 feet away, on the production line right outside my office!

348

DW: What was the problem?

KK: It was an operation where they were installing the fuzzy, felt-like material that goes in the ceiling of the car-very big and very clunky. Our data said that the mean cycles between failures was about five minutes, and the mean time to repair was about a minute. I was amazed that the line was stopping that often, and thought maybe the data was wrong, so we went and looked for ourselves. Sure enough, we watched the operator run for five cycles, stop the line, walk away, pick up five more of these big, bulky items-they weren't heavy but they were big- drag them back, restart the line, and continue to install them. Every five cycles she would stop the line. Was it considered a major problem before we looked at it? No. It's not like we were losing an hour straight of production because something had broken down. We were only losing one minute. But it was happening every five cycles.

We could see immediately why the material wasn't closer to the line. There was a supervisor's office in the way. We found out there had been a request made some time ago to move the office, but it was considered very low priority and it wasn't getting done. So I got the office moved, and lo and behold, throughput of the entire plant went up, which was a surprise, because my experience told me that I couldn't expect that. Then we used the software to find the next bottleneck and continued on with that process until we were making our throughput goals very steadily, every day. That was a real change in the way that plant operated.

DW: Did you take your insights to other GM plants?

KK: Yes. We demonstrated the process when central office manage- ment visited the plant, and it became apparent a lot of plants in GM weren't hitting their throughput targets. Eventually, I left Detroit- Hamtramck and went to a central office position to help start a divi- sional group to implement this solution. Seventeen years later, I'm an executive at GM who owns the process for all of the North American plants, and it has been expanded to include the simulation of future manufacturing designs.

DW: And this is all TOC related?

349

KK: Yes, but there are other disciplines involved. You have to un- derstand simulation, and how it predicts throughput, and why it's important to understand where the bottleneck will be for a future design. But TOC is the basis for what we do. I still teach a two-day course. We might go to a plant and train the whole staff in how to use TOC concepts. I always give out copies of The Goal ahead of time and ask them to read it before the training. It's gotten to the point in manufacturing, however, where there are not that many people left to go through the training. My internal customers are usually very savvy now about TOC, bottlenecks, data collection and analysis. So I rarely have to sell the concept anymore. Demand for data collection imple- mentation to drive the bottleneck software, for example, exceeds our ability to install. And while I'm responsible for GM North America, this week alone I have people in China and in Europe working on these kinds of issues.

DW: How has your use of TOC concepts changed over the years?

KK: What we found when we first started out is that we were dealing with the low-hanging fruit. You look at that first example I told you about, and it was very obvious that the office was in the way, and the solution was just to move it. Over time, the solutions to the problems have become a lot more difficult to find. This doesn't mean you can't solve them, it just means you might have to use more scientific tech- niques. Now I might have to apply statistical methods as opposed to simple observation to understand what's driving the problem at a work station.

Another thing we're doing lately is applying what we've learned from The Goal to the design of new plants and production lines. In -effect, we're solving problems before they arise. Eli Goldratt hasn't spent a lot of time talking about using TOC in that way, but we've taken his concepts and adopted them to our needs. That's been the beauty of it for me. If you understand the logic and the reason behind the methodology, then you can apply that stuff continuously.

DW: It's interesting that a way of thinking about production problems that you found useful 15 years ago you still find useful today. Does that surprise you?

350

KK: Yes and no. The Theory of Constraints is a very scientific, logical process. And because of that, when the game changes you can always go back to the logic. Originally we just had to find the bottleneck, walk out there, ask three or four questions, and we knew what to go and do. Now we can change the way we design whole manufactur- ing processes to make sure they're better from the start. But the logic behind TO C-the conflict clouds, the current reality trees, the way we ask questions to uncover the constraint-all that still applies.

I think the problem with too many other approaches is that once the first layer of problems goes away, and the crisis no longer exists, then it's, 'Phew! We're done!' In the TOC world, you find yourself asking, 'Where has the constraint gone, and what can I do to help break it?' So you're never done.

I'd like to be able to tell you that as soon as I started telling people about these concepts, the whole organization immediately changed to the new paradigm. The fact is that it has taken years to get the process going, and the leverage to make improvements is still significant, es- pecially in a company as large as General