First of all, clear the d**n windshield!

Shortly after my sixteenth birthday, I asked my father if he would – please, please – teach me how to drive. After giving it an hour or so of thought, he ultimately put down his newspaper, grudgingly raised himself from his well-worn La-Z-Boy recliner, and motioned for me to follow him. We put on our hats, gloves, and winter jackets before opening the door leading to the driveway. Parked there, underneath an inch or two of snow and ice, was my father’s beloved 1953 Sherwood Green Mercury sedan – outfitted, courtesy of our local auto parts store, with pricey after-market clear plastic seat covers. (Ice-cold in the winter and blazing hot in the summer, clear plastic seat covers were the “in thing;” i.e., the latest fad that every middle-class automobile owner just had to have in the fifties and sixties).

I was so excited that I raced to the car, opened a door, sat down in the driver’s seat, and hurriedly placed the key in the ignition. I was ready to rumble. My father’s reaction was less enthusiastic. He opened the driver’s side door, removed the key from the ignition, and – through clenched teeth – said: “First of all, clear the d**n windshield!”

My father and I were never close. He did, however, teach me a few useful things. One was how to (properly) mix mortar and lay bricks. Another was how to swear, rather splendidly, in five or six different languages. It was not until somewhat later in life that I realized that quite possibly the most important thing he had taught me was to “first of all, clear the d**n windshield!”

That windshield life-lesson has helped me solve, among other things, a multi-million-dollar problem that a large, international oil company was facing. They had represented their production and inventory problem by means of a massively large linear programming (i.e., optimization) model. The solution of that model would – they were confident – provide them with the answer to how to minimize their costs and maximize profit. After all, that’s what all the contemporary newspaper articles promised.

The answers that were cranked out, after many hours on a then state-of-the art mainframe computer, were not what they had hoped for. Most solutions were, in fact, infeasible. Others were obviously ridiculous. Having read some of my books and papers on mathematical programming/optimization, they contacted me. They asked for my recommendation for a “better” linear programming software package – one that would generate “rational” solutions. They were convinced that the inane results they were being provided were, “naturally,” the fault of the software they were using.

Rather than recommend an alternative linear programming software package, I suggested that they permit me to examine the linear programming MODEL they were using. They were bewildered. Their model, they noted, was massive in size – thousands of variables and constraints – and had been, they insisted, meticulously and painstakingly constructed by a carefully selected team of their top researchers. Those researchers, they explained, were not only graduates of the top Operations Research and Computer Science programs in the country, they were also intimately familiar with the business of oil and gas production.

A few weeks later I received another call from the same firm – this one was considerably more frantic than the first. I was offered a first-class airplane ticket, reservations for a suite in a luxury hotel, and they even volunteered to triple my usual consulting fee. Being a huge fan of airline food, I agreed. I arrived at their facility the next day. (By the way, the over-cooked, rubbery, and tasteless airline meals that were served were just as fabulous as I had remembered.)

It took less than an hour to discover the source of their problem. There was nothing at all wrong with their linear programming software package. What was wrong was that, when they “meticulously and painstakingly” constructed the model, they failed to perform a dimensional analysis. When I conducted that crucial aspect of mathematical model building, I discovered several instances in which the dimensions on the left-hand side of a constraint were expressed in gallons while those on the right were in barrels. Once the model was made dimensionally correct, the results were both credible and optimal. It was as simple as that.

In other words, the firm’s “carefully selected team of their top researchers” had failed to “first of all, clear the d**n windshield.”

In another instance, I was hired as a consultant for a highly respected national “think tank,” one that dealt with matters of military readiness, defense planning, supply chains, and environmental issues. I happened, by chance, to sit in on a meeting in which one of the speakers reported on the progress he (I’ll call him Bob) and his group had made on a problem related to the targeting of enemy missile sites. Their team, Bob claimed, had developed a model that “guaranteed an optimal targeting scheme.”

That afternoon Bob gave me a more detailed briefing of the missile site targeting model he and his team had developed. One portion of the model consisted of a series of simultaneous linear equations. The number of equations were equal to the number of variables (i.e., unknowns). Instead of, however, using readily available simultaneous linear equation solvers (i.e., commercial software), they had “solved” the problem via simulation! More specifically, they had used trial-and-error to input random values for the unknowns and then checked to see how “closely” they satisfied the series of equations. Once they had exhausted their allotted time on their mainframe computer, they selected the set of randomly generated variables whose sum of deviations (i.e., for the set of equations) had been minimal. Rather than mentioning just how absurd this was, I recommended that they use a readily available statistics package that was more than capable of solving their problem – and providing its actual, CORRECT solution.

Once again, this was an example of a failure to “first of all, clear the d**n windshield.” Once again, it demonstrates that, before grabbing your favorite “tool,” or a management guru’s “latest and greatest” scheme, you really do need to clear the d**n windshield.

The third, and last example I’ll present of the need to “first of all, clear the d**n windshield” was one that occurred at a major semiconductor fabrication facility (aka, “fab”). The cycle time (i.e., average elapsed time between the introduction of a semiconductor wafer into the fab and its exit as a finished product) was far, far more than had been predicted by the firm’s analysts and their costly, time-consuming simulation programs. In fact, the average wafer spent more than 85 percent of its time in the production line simply waiting (i.e., in queues) to be processed. In other words, it was being worked on less than 15 percent of the weeks and weeks of its cycle time.

Senior management demanded a drastic reduction in cycle time. The “obvious” answer, it was claimed, was to purchase and install more “tools” (i.e., more of the multi-million-dollars per copy of the processors, such as etchers and lithography equipment) and hire and train more factory personnel. They were monumentally unimpressed when I recommended that, instead of adding more tools and human bodies, they should first consider identifying and then surmounting the 3 obstacles that hinder every decision. More specifically, I suggested that they look more closely at their maintenance program. The maintenance program that I observed had, in my opinion, the very same problems that I have observed in so many other manufacturing facilities or supply chains, or – in particular – in the maintenance of military systems.

I was informed that “the only significant problem” that had been identified in their maintenance program was the fact that their tools were spending a surprising amount of time in unscheduled maintenance (i.e., breakdowns). But they didn’t believe that matter really had much of an impact on their product cycle time. Their plan was to “remedy” the maintenance downtime by hiring more maintenance personnel and increasing both the number and extent of maintenance events. But, first of all, they were bound and determined to add more tools and factory personnel.

I recommended that, before pursuing that extremely costly and time-consuming course of action, they should determine if the fab was experiencing the Waddington Effect and, if so, identify the causes of that effect. They humored me by allowing me and a colleague to collect the data necessary to reveal whether or not there was a Waddington Effect (i.e., by constructing a graph of the scheduled and unscheduled downtimes plotted on a time axis). Once the “windshield had been cleared,” it was obvious – even to the most skeptical – that the Waddington Effect most certainly did exist. Given that revelation, we then looked for its causes. These were easily identified as:

(1)  Too many (rather than too few) and too frequent maintenance events,

(2)  Maintenance specifications that were not C4U-compliant (i.e., specifications that were NOT Correct, Complete, Concise, Clear and Unambiguous),

(3)  The allocation of maintenance techs to tools based mainly on the cost of the machines rather than their impact on fab cycle time.

Given that the Waddington Effect was first observed by the British in WW2, and that we rediscovered it during the Apollo manned moon-landing program in the early sixties, its crucial importance should be recognized in every maintenance program, both civilian and military. Unfortunately, that recognition appears to continue to be hindered by a failure to clear the d**n windshield. More precisely, the problems discussed above were a result of a failure to identify and then overcome the 3 Obstacles we face in any attempt to find the solution to any non-trivial problem we might encounter. Instead, and all too often, we focus on the SYMPTOMS of our problems and ignore their CAUSES.

So, just what are The 3 Obstacles? They are Unnecessary Complexity, Excessive Variability, and Intellectual Myopia. If we recognize and overcome them, we have – in effect – cleared the d**n windshield. A preview of The 3 Obstacles book may be reached by clicking on: https://amzn.to/2IRdjyq

About the author: Dr. James Ignizio began his professional career as an engineer on the Apollo manned moon-landing program. He has spent nearly 20 years in industry and 30 more in academia. He has served as a professor and chair at both the University of Virginia and the University of Houston, and as a professor at the Pennsylvania State University and the University of Texas. He is the author of 19 books, more than 300 peer-reviewed publications, and is a Fellow of the Institute of Industrial Engineers, the British Operational Research Society, and the World Academy of Productivity Science. He is also a Distinguished Alumnus of the Department of Industrial and Systems Engineering at the Virginia Tech university. His latest book, THE 3 OBSTACLES: How to Identify, Overcome, and Exploit Them, was released in March of 2018. It is intended to be used as a highly readable and yet effective and practical means to “first of all, clear the d**n windshield.”