What is SPC?


During my daily tour of the shop in a Detroit automotive parts plant, I noticed Doug was entering data into his Datamyte. The Datamyte was used by the shop operators to upload measurement data into the SPC program. SPC stands for statistical process control. During the set up of the machine for a production run, the operators would take measurements of the work pieces. They would make adjustments to their machines until all of the attributes on the work pieces conformed to the engineering specifications. 

Once the operator was satisfied with the setup, he would periodically take formal measurements of some of the parts attributes. The sample size for this was typically 30 pieces. They would typically do this at the beginning and end of production runs and a few more times in between. He would then enter them into his Datamyte, which was then uploaded into the SPC program. The SPC program would then assess if the operation was either "in" or "out" of control. Since the objective was six sigma, the operation would be deemed out of control if the Cpk value was less than 1.33. In this case, the operation would be red tagged and shut down until such time that 1.33 Cpk was achieved.

Doug's machine was a header machine which produced slugs for down stream closed die work. These slugs had several attributes. The most important one was the weight of the slug. In closed die work, the mass is very important. If it is too heavy or too light, the die could explode. Not only could this cause damage to expensive equipment and tools, it could be fatal to the press operators.

That particular morning Doug's machine was pumping out slugs at one every 1.5 seconds. This was not unusual. What was unusual was that Doug was rapidly entering information into his Datamyte. I stood there and watched him while he entered, 302, 304, 301, 302, 301, 302,......etc. I said, "Geese, I did not know you had a photographic memory." I had assumed Doug had previously weighed thirty sample pieces and had remembered the gram-weight of each of each piece in the sample. Doug had run that machine for twenty-years. He knew it better than anyone. Doug was very intelligent. I would not have been surprised if he had responded affirmatively to his having a photographic memory.

Rather than confirming his photographic abilities, Doug grinned at me. Then I knew something was up. Doug was well aware of the potential hazards if the slugs were out of tolerance. Doug then explained to me that his header machine cannot hold six sigma. He said management and the Maintenance Department had repeatedly ignored his request upgrade/repair deficiencies in his header machine. He also knew the parts were within engineering specification and that there would be no problems downstream at the presses. He then explained to me that several years ago when they first implemented SPC, most of the operations could not perform at six sigma. And when values of less than 1.33 Cpk were submitted to Ford, the several of the operations were shut down and some operators were sent home without pay. Management scurried around and got the operations back up to 1.33 Cpk. That took a lot of time, money, and energy.

So, this is what Doug and most of the operators in Detroit do. They reason that if Ford and management want 1.33 Cpk, then they enter data into their Datamytes so that 1.33 Cpk or better is computed from their uploads. Doug knows that if he were to enter the actual values of 301, 307, 302, 311, 308, 304,............etc. he would not get 1.33 Cpk. But if he entered a tighter range of say, 301, 303, 304, 301, 305, 302,....... etc. he would get the required 1.33 Cpk.

So rather than entering the actual weight of the slugs and endure the threat of being sent home without pay, Doug and several of the other operators entered what management and Ford wanted to see.

Doug is not evil here. The operators are not evil here. The management is evil here for putting the operators in this situation in the first place. As Deming once said, "management is responsible for 80% of the quality of the product." All the operators are doing are just delaying the inevitable. That is a loss of jobs.

Well, eight years later in 2002, Ford is laying off tens of thousands of people. Ford is shrinking. Why? Mysticism.


Quotes from Shingo.

"SQC (Statistical Quality Control) methods try to get around the problem of 100% compliance assurance by enlisting the aid of inductive statistics and cutting down on the task through the use of sampling."

"Active use should be made of statistics in the sense that the technique is extremely effective in the planning phase of management. But statistics is not always effective in control and execution phases. In fact, it can be surely said that an infatuation with statistics has impeded the progress of the management function itself."

See Shigeo Shingo, Zero Quality Control: Source Inspection and the Poka-yoke System. Productivity Press. 1986.

Click here to see how SPC can tell a story.



1995/2006 James Vick, Inc.

Contact: james@jamesvick.com    Phone: 817-279-3116    5102 Wedgefield, Granbury, Texas 76049