Lessons and the value in making lean manufacturing work

05/01/2002

by Rob Mionis
Axcelis Technologies Inc.

Several years ago, I invited James Womack — the renowned thinker on lean manufacturing — to tour our manufacturing operations. At the time, we were pulling ourselves out of a market nosedive and realized there had to be a better way to serve our customers, rather than operating in a fire-fighting mode.

Womack looked around, saw a typical chaotic day in the life of a semiconductor equipment manufacturer, and came back with surprisingly simple advice: start somewhere and start now.

Womack's advice applies to almost any semiconductor equipment manufacturer today. For most of us, the lean philosophy is counterintuitive; we're all accustomed to a job-shop mentality. Lean manufacturing challenges all those ingrained and long-accepted practices; it requires a new way of viewing production and administrative activities, and it doesn't happen overnight.

Lean manufacturing has been around for about six decades, but the semiconductor equipment sector has yet to catch on to the concept. Maybe this is because we're too preoccupied with scrambling up and down market cycles to make real operational improvements. Whatever the reason, my experience suggests that it isn't just for mass-manufacturing lines. It can work in the high-tech, relatively low-volume world of IC equipment making.

While we are still pursuing leanness, here is an example of what we've been able to achieve during the last ramp:

Increased systems revenue 86% without adding infrastructure. This was accomplished by shipping 40% of systems under a ship-from-cell concept, where final assembly and test are done at the customer's site, allowing us to deliver tools four to six weeks faster than conventional methods.

Reduced customer lead times by 25-40%. We achieved faster product delivery by trimming cycle times 17-22%.

Strengthened the supply chain. As a byproduct of waste elimination, we reduced inventory days on hand by 31% and reduced our supply base by 20% (45% over two years). This allows us to respond more quickly to changes in customer demand.

Added value for customer. We have cut transaction costs and simplified the business-to-business interface for customers by launching e-commerce programs that allow spare parts procurement and uninterrupted Web-based supply of critical consumables.

The Toyota example
Toyota pioneered the lean manufacturing model. This model instills a series of somewhat cerebral principles across production and administrative activities: specify value; identify value streams; eliminate waste so a value stream flows without interruption; let customers pull value from the manufacturer; and pursue perfection. In action, these principles not only revolutionize conventional mass manufacturing, but also lead to a better understanding of customer needs and demands.

It's not that the semiconductor equipment community doesn't strive to improve manufacturing operations. Most of us recognize and embrace the virtues of outsourcing, for example. From robotic subassemblies to entire tools, outsourcing allows manufacturers to focus on core technologies and hand off noncore work to outside partners. When done right, this usually results in shorter time-to-market, lower production costs, and incremental improvements to a company's overall competitiveness.

These are great benefits, but equipment makers are still held captive to semiconductor industry market cycles. The current downturn is a sure sign we can do better. To be truly competitive, equipment firms need to operate in sync with cycles, rather than react to market sways. Some lessons from Womack's book Lean Thinking, in which he and co-author Daniel Jones provide common-sense lessons on making the transformation to a lean organization, apply directly to semiconductor equipment manufacturing.

Deliver customer value
Arguably, the most important lesson is to define value, those products or services your customer must have, and deliver it. The largest obstacle to delivering value is waste. Once you start looking for it, you realize waste is everywhere, and its elimination is not always easy or straightforward.

For example, a typical semiconductor equipment manufacturing line will include a test-and-assembly area where a tool could spend 8 hours to 45 days for assembly and run-through. The test step is designed to ensure that there were no blunders in the manufacturing process.

The waste in this scenario is that test typically adds no value to the customer, who already expects the tool will work as requested. Worse, this can delay shipment of a tool if test becomes a routine catchall for reworks. Although we can't stop all test activities, we can rethink the manufacturing product flow to test modules on the assembly line vs. integrating the modules for test in a separate bay. Final assembly and test can be performed right at the fab vs. in-house.

At a minimum, this eliminates duplication of efforts and wasted time of test and moving product between manufacturing areas. At a maximum, product-manufacturing costs drop, available capacity increases, and the customer is thrilled because you've delivered the product sooner to support the fab's manufacturing needs.

Make no tool before its time
To be of any use to customers, toolmakers must be able to provide the right technology at the right time in the right quantity — all paced by the customer's demand. To make our operations a lean enterprise, I began to take notice of suppliers' operations. Every time we consumed material (i.e., frames, pumps, circuit boards, or electrical assemblies), it triggered our supplier to replenish the just-used item on our manufacturing floor. At the same time, this triggered our supplier to manufacture another part at its factory. So, for every part we pulled, the supplier produced another. This means that we never had to wait for a part from the stockroom. The needed part was always there, without question.

Imagine the effect of this across the entire supply chain, where a consumer's purchase of a cell phone would trigger all suppliers along the value chain to replenish only the consumed product — the phone, its ICs, and, eventually, the equipment used to make the ICs.

It's the difference between a push-driven manufacturing system paced by a pre-determined demand forecast and a pull-driven one paced by real customer demand. The problem with the push system is that forecast demand and real demand will inevitably fall out of sync, leaving either too much inventory or not enough. In the first case, an oversupply of unneeded parts is pure waste; in the second, the lack of parts creates waste in the form of delays because production cannot progress until the needed part arrives.

We now have several product lines on a pull-based system and others making the transition. We've also been able to implement this kind of flow on our manufacturing floor. On one production line, tools are manufactured in five stages, with the production of each new tool triggered only when the tool frame moves from stage one to two. This has resulted in a 30% drop in cycle time, a 52% drop in operators/shift, and a 14% drop in floor space.

We're still finding new ways to introduce lean principles into operations. The move to lean manufacturing is not an end-goal, but a business philosophy. Lean companies, in turn, are strong, flexible enterprises, better able to manage downturns and respond quickly in upturns. Ultimately, I believe lean companies can become a lean industry, where market cycles dissolve under the flow of a waste-free value chain. It's a utopian vision, but maybe not completely out of our reach — if we start now.

Rob Mionis is senior VP of worldwide operations at Axcelis Technologies Inc., 55 Cherry Hill Dr., Beverly, MA 01915; ph 978/787-4000, e-mail [email protected].