Quality: nearly perfect is good enough!
Article by Coenraad Esveld
With a moderate effort quality can often be brought to a reasonable level. Further improvement steps will require an exponential growing amount of energy and cost. This is illustrated in the graph above, clearly revealing that the effect of additional effort become less and less effective, supporting the statement that nearly perfect is good enough!
Quality has always played a central role in my work as railway engineer at Dutch railways, Delft University of Technology and in the consulting business. In practice the question is often raised how far you should go with perfection, as it is directly related to cost and effectiveness.
In practice the question is often raised how far you should go with perfection, as it is directly related to cost and effectiveness.
Anyone having experience with development of software and hardware knows that the actual time of the developing process is often a multiple of the predicted time.
In my experience at TU Delft factors of 2 to 3 were no exception. With for instance the development of software the basic functionalities mostly go rather quickly. But adding additional user requirements, better user interface and improvement of stability take an disproportional amount of time and associated cost. The same applies for the development of measuring equipment.
The efforts to make things better have a limited effect.
Let me give another example. An ideal railway track is smooth, without irregularities. Due to the passage of trains you see a gradual development of deviations in the track geometry, which then require maintenance.
Theory and practice have shown that the optimal solution is a high, but not perfect quality.
The key question is to what level of perfection the quality should be maintained, keeping in mind that absolute perfection – say an ideal straight line – is practically not feasible and would also lead to unacceptably high cost. The better the quality, the less deterioration, which means less maintenance cost. In fact this is an optimization problem with a target of minimal cost.
A salient detail is furthermore that a high initial quality is a prerequisite for low life cycle cost. In addition, limited maintenance budgets and restricted capacity of people and machines at the maintenance contractors are also practical conditions influencing the decision making process and the resulting track quality.
Last, but not least is the aspect of track availability for maintenance, often referred to as possession time. As revenue service has priority, most maintenance works are scheduled during the night and these working hours are relatively expensive.