Quality Scan: Counterpoint: Lean Sigma Defended
In July's Quality Scan, Geoffrey Mika dismisses the concept of "Lean Sigma" as a misguided hybridization of two distinct methodologies. In doing so, he makes some contentious statements about six sigma, and so misunderstands the thrust of Lean Sigma.
First of all, although there is significant overlap in the tools used by Lean and Sigma methodologies, I contend that Lean Sigma is not a hybrid, but a two-pronged approach: bottom up (Lean) and top down (Sigma). As Mr. Mika says, Lean is a philosophy that can be applied on the shop floor, by shop-floor workers. While Mr. Mika suggests that six sigma is "a management tool for a very select few," the statistical theory behind six sigma is accessible to any manager or engineer. Certification of managers and engineers at Black Belt or Green Belt level is a desirable goal, because the underlying philosophy of six sigma is simply that management decisions should be driven by hard data, not by a manager's gut feel or an engineer's SWAG.
Mr. Mika also asserts that six sigma needs high volumes to work: actually six sigma copes better with low volumes than does Lean. What constitutes "high" volume is relative, but six sigma is statistics-based, and most statistics (at least for variables rather than attributes) will work with sample sizes of a few dozen.The volumes I work with are "low" by industry standards—injection molded components with annual quantities of a few hundred to a few tens of thousands (lower than many molders' run quantities)—but six sigma techniques are being successfully applied in production.
If Lean Manufacturing is understood to be synonymous with the Toyota production system (TPS), however, then Lean is difficult to apply outside of mass manufacturing, because the TPS is built around pull production and onepiece flow. To see why this can be a problem, consider an idealized pull system. Here, finished goods inventory (FGI) stands at the end of a linear sequence of workstations dedicated to a single product. FGI contains precisely the number of parts that the customer orders. A single part stands between each workstation and the next downstream workstation, and the operator of the upstream machine does not start working on a part until the next downstream operator takes the intermediate part from its location.
When the customer places an order, all the parts are shipped from FGI, and FGI takes the output part from the final workstation, so that a signal to start manufacturing is automatically passed upstream as each operator (in reverse sequence) reaches for a part and starts work.Work continues until the order quantity is reached, when FGI leaves the last part in place and production stops. If the time between orders is constant, and if the work content of each workstation is perfectly balanced to that time, then production is continuous (because an order quantity of parts will have been completed at exactly the time the customer places an order), lead time is zero, and work in process inventory (WIP) is minimized at one part per workstation. Even in this ideal system it is obvious that any irregularity, whether delayed raw material delivery, production of a defective part, or changes in order quantity or frequency will destroy the system's efficiency.
In the real world, workstations are rarely dedicated to a single product, resulting in set up changes, supermarkets (storage for fixed quantities of WIP), and Kanbans (fixed production quantities). As the environment moves from high-volume/low-variety to low-volume/high-variety, Kanbans and WIP proliferate, and parts compete for workstations, necessitating complex scheduling rules to maintain efficient throughput. Consequently, the academic manufacturing literature is replete with descriptions of modified Kanban and hybrid push/pull systems designed to extend Lean into higher variety environments.
The key point here though is not that Lean Manufacturing will not work in specialized job shops (although it won't), but that any Lean system is vulnerable to variability. With little WIP to buffer the system, defects, delayed deliveries, machine downtime, or product proliferation will disrupt production. Any enterprise embracing Lean needs an equally powerful set of tools to minimize variability, and six sigma provides exactly that. What results from combining Lean Manufacturing with six sigma, therefore, is not some misbegotten chimera, but a rigorous methodology for ensuring that manufacturing is both efficient and robust—in two words: Lean Sigma.
This article was first published in the October 2006 edition of Manufacturing Engineering magazine.