TOC, Lean Six Sigma, Continued; Part 2

Like many of my contemporaries in the late 80s and early 90s, I had read Eli Goldratt's "The Goal" and enjoyed it immensely. Also, like many of those same contemporaries, implementing the lessons of "The Goal" seemed difficult if not impossible without Jonah, the philosopher-industrial engineer, being present.

My thinking changed in the fall of 1996. I assume the reader is familiar with the half-day seminars that seem to serve to get people out of the office in exchange for continuing education credits. I attended one of a different caliber; Debra Smith of Constraints Management Group presented a useful several hours of Drum-Buffer-Rope, or DBR, which was the named system for scheduling work that "The Goal" featured. The scales fell from the eyes, and I was able to envision how the wirebond area, which was the production line's bottleneck, could use the TOC philosophy to improve throughput.

I proceeded to set up training sessions for all of the assemblers and technicians using "The Goal" videotape. Training centered on how the wirebond area was the process bottleneck and what steps we would take from that day forth to subordinate decisions towards improving that process.

The basic concept underlying "The Goal" is this process:

- Identify the constraint
- Exploit the constraint
- Subordinate all other decisions to exploiting the constraint
- Elevate the Constraint
- Start over

In the video (and in real life at Interpoint), the huge pile of work-in-process was a dead giveaway as to the location of the constraint. Constraint identified!

The second step requires a bit more work. Exploiting the constraint requires looking at the input to it and how well the available process time at the constraint is used. Essentially, in Six Sigma terms, performing a Value Stream analysis through wirebond revealed many significant details that would assist in exploiting the constraint.

Each wirebond machine is subject to a certification process that lasts for 4 hours of production time. Scheduling assemblers around that certification time was the first subordination step. The first rule applied is that, except for the mandated break, certification means 4 hours of wirebonding.

The astute reader will note that the process of exploiting the constraint involves subordinating many decisions. The decision to exploit the constraint is treated separately as a "line in the sand" for management to communicate to other team members what the focus is.

The second subordination step was to examine batch sizes in wirebond. The concept behind this thought process was to ensure that 4 hours or some even fraction of that amount was transferred to the wirebond machine.

Implementing these two steps increased throughput in wirebond but did not result in first test yield improvements. The logical conclusion we reached is that while the throughput through wirebond is better, some of the product coming through was not capable of passing first step or even pre-cap (before hermetic sealing) inspection.

In my next post, the next subordination step will be presented and dissected.

Using TOC, Lean, and Six Sigma Tools

A writer at a LinkedIn blog asked a very good question: has anyone used the Theory of Constraints, Lean, and Six Sigma tools together to solve a problem??

This post will be the first in a series reviewing how we utilized a variety of tools to improve throughput and quality in a systematic fashion at Interpoint. In hindsight, we did successfully mix the tools of TOC, Lean, and Six Sigma. What was missing from our application was the lingo. More later on why that may not have mattered.

BACKGROUND

Interpoint was a small, publicly-traded company with a long history serving the defense and aerospace markets. The company designed proprietary DC-DC converters and custom-designed hybrid microcircuits using a customer circuit. The hybrid featured silicon die (FETs, transistors, diodes, etc.) that were mounted on a multi-layer silicone substrate. After die attach, the substrate was epoxied into a metal package. The circuits were joined by electro-mechanical wirebonding, tested, and then sealed and burned-in using MIL-STD 883 criteria.

A good visual of hybrid circuits is here.

A barcoding system was used to track trays of parts. A group of parts, say the 11424-002, would have a work order associated with it and have several trays of parts associated with the assemblies. Manufacturing was done in a clean room with many static prevention tools and processes. Hence the trays were metal.

Our first improvement was to digitally download the barcode into Excel and use this spreadsheet to track product progress. Predicting scheduled completion and the quantity (yield) of a work order was at best an educated guess. We then developed notional durations for each process in manufacturing for each custom product that would normally be manufactured in a given year.

These steps, while an improvement over guesswork, left me wondering what else we could implement to make yields better. My next post will describe how we started down that path.

Process Friction

Process friction is, simply put, the accumulated negative effect towards achieving a goal due to rules, tribal knowledge, stovepipes, and ignorance. Process friction makes completing a task much more difficult than your intuition allows; the nagging feeling of being held back feeds frustration in even the most resilient people. Identifying key processes and drilling down into the major steps tends to put comfortable people on edge - what are "They" questioning and is my job being threatened are not unreasonable reactions.

The best literature on process friction comes from the online marketing and web sales folks. They understand that any online process which takes too many keystrokes and mouse clicks will drive page views elsewhere.

A good starting point to understand the issue comes from Payson Hall.