Limits to concurrency

In the global race to bring new products to market, many firms have adopted concurrent engineering as a technique to shrink development lead time. Due to the many concurrent engineering success stories in the business and engineering literature, a common misconception has grown that more concurrency is always better. The major contribution of this paper is a rigorous demonstration that limits to concurrency exist even in the simplified situation in which concurrency is modeled as the number of design modules to be executed in parallel. As complexities such as communication linkages between modules are layered onto our basic model, we show that the expected project completion time is minimized at a finite number of modules, a number that decreases with increasing problem complexity. In general, the more complex the project, the stricter the limits to concurrency. This strongly suggests that project managers should be cognizant of the potentially adverse effects of pushing concurrency too far.