Real Quality Enforcement

My company built 8 very large tankers in Korea in the very early 2000's. Here's what I learned about quality enforcement.

The Contract

Everything depends on a strong contract. Unless the contract is rock solid, all the enforcement in the world will not result in a solid product. To the extent possible, the contract should be functional rather than prescriptive. This maximizes the vendor's responsibility while giving him freedom to innovate and come up with better or cheaper ways of providing the required functionality.

The contract must require stringent physical tests of all critical components. Those tests must be delineated in an unambiguous fashion, no wiggle room. The contract should say as little as possible about how the vendor produces the component. For example, standards that specify how welds shall be tested are essential. Standards that specify who can do the welds are anathema.

Bid everybody, trust nobody

In procurement, the most important weapon is competition. Quality requires doing everything possible to maximize competition among vendors. There is always somebody who will do it cheaper and better. Our job is to find that guy; and all the vendors must know we are searching for him. Often that somebody is the new guy on the block. Sometimes he has discovered a better way of providing the function. He tends to have low overhead. And he's always the hungriest.

There is no greater motivation than survival. If the vendors know that in order to survive, they must come up with the cheapest product that will meet the spec, that's what they will do.

The antithesis is pre-qualification, a procedure which constrains ``competition" to a small, well-established handful, who know each other well. Hard to imagine a more counter-productive policy. Much the same thing can be accomplished by requiring burdensome, paperwork intensive certificates.

Don't show me your certificates. Show me your guarantee.

So if we are going to base everything on price, what's to prevent the vendors from producing a shoddy product? Fear. Fear of production delays. Fear of rejected products. Fear of penalties. Fear of warranty claims.

When asked about quality, vendors will offer a long list of references and extol all their QA certificates. The proper response is ``Wow! That's really impressive. With such great quality, you should have no problem giving us a ten year guarantee with substantial penalties if the product fails." It is amazing how many vendors offer wonderful quality which they won't guarantee, especially if they know the competition is limited.

References are pretty much useless. No one wants to admit that his ship or whatever is sub-standard. He has to claim that he has bought a wonderful product. And even if the quality is so lousy, he's prepared to say something uncomplimentary, his lawyers will tell him to keep his mouth shut, for fear of being embroiled in a legal dispute. Usually the only way to get the real story is to get the reference drunk.

A good model here is the commercial aircraft industry. Here are the typical guarantee terms for a commercial aircraft purchase.

1) Full warranty for five years.

2) Rewarranty of two years.

3) Service Life Policy. Primary structure including landing gear and movable surfaces are guaranteed for 12 years in the sense that cost of replacement is shared between builder and buyer with the builder proportion decreasing linearly from the end of the full warranty period to zero at 12 years.

4) Similar terms are provided by the engine manufacturers.

Such guarantees generate another benefit. Since the aircraft and engine builders are on the hook, they take a real interest in how the airplanes are maintained. For example, the airplane engine builder is involved in every major inspection and overhaul of his engines. He has a strong pecuniary interest in calling out poor maintenance and operating policies in a way that a regulator does not. If he can prove that the maintenance/operation is not per manual, he is off the hook. This in turn puts real pressure on the aircraft owner to do his maintenance correctly. The system is self-policing.

Buyers tend to treat guarantees as an add-on. Maybe we can get another six months with no change in price. The guarantee is at least as important as any other requirement in the contract. A strong guarantee may not come cheaply; but, if that's the case, then that is the real cost of obtaining a quality product.

Test the Weld, not the Welder.

Real quality enforcement focuses on the results, not procedure. This is summed up in the mantra ``test the weld, not the welder". All shipyard welders are trained and duly certified. But that does not mean they are equally competent, or even competent. Our newbuilding contracts had weld specifications that were considerably more stringent than normal shipbuilding practice. It did not take long for the yards to figure out that it was in their best interest to put their best welders on our ships. But even the best welders have bad days. All we care about is the weld itself. This substance-not-process rule applies across the board.

Hands On, Test Enforcement

It is not enough to require a thorough, rigorous set of tests. All acceptance tests must be witnessed by our guys. Accept no vendor paperwork. Our inspectors must be aware that acceptance tests are carefully choreographed, as much stagecraft as test. The contract must allow them to force a repeat test if anything is questionable. Most importantly, the inspector must know that we want him to reject the test if he's the least bit unhappy, regardless of the effect on the production schedule.

Our inspectors must spend most of their time randomly patrolling, getting to know the real workers, explaining our standards and why, appreciating a job well done, and politely but firmly calling out defects and bad practice. This avoids the stagecraft, and properly done motivates the work force. Most people would rather do good work than bad work. Patrolling is at least as important as witnessing tests in enforcing quality. We caught far more problems patrolling than we did during acceptance tests, in part because if the patrols had revealed no problem, the component was unlikely to fail the test.

All this implies a large, expensive inspection force. Most shipowners assign a half dozen inspectors to a large newbuilding project. For our project, we had 24. This too is part of the cost of real quality.

Admit Mistakes and Correct.

It is inevitable that as the job proceeds, we will uncover mistakes on our part, and design features that can and should be improved. This means Change Orders. In negotiating a Change Order, we are at a big disadvantage since the vendor will have a monopoly on us. The tendency is to cover up the mistake and avoid pointing out the improvement.

Our team must know that if you go down that route you will be fired. We must catch our mistakes as early as possible. The owner welcomes ideas for improvement and is willing to pay for them. If you made a mistake and admit it, you are likely to get a fatherly lecture, and a mental note that this guy is the real thing. If you made a mistake and attempt to cover it up, you are gone. The goal is to build a web of trust within which each level can be completely honest with the next level up, and expect the next level up to react in a way that is consistent with getting the job done right.

Quality is hard work.

Quality enforcement is a lot of hard work. It starts with drafting an iron clad contract. It forces buyers to turn over every stone. It means a lot of on-site inspectors, working 60 hour weeks, almost none of which is sitting at a desk. It means managers must spend most of their time patrolling with the inspectors. It is the only way they can know which of our guys are doing their job and which are not. There is no easy way to quality.

Nuclear QA violates just about everyone of these rules. It focuses on process, rather than substance. It erects expensive barriers to entry and stifles competition. It elevates certificates and ignores guarantees. It assumes paperwork represents results. It discourages fixes and improvements. Nuclear quality should be a pejorative.