Optimization

How do we choose the set of dimensions that optimize the performance of a component? To carry the load with the smallest possible self-weight, to have the highest possible eigenfrequency, to have the smallest risk of failure, to allow the fastest possible rotational speed, to minimize the cost of welding, to minimize the cost of material...?

Design (or synthesis) is actually the opposite of analysis. It is also infinitely more difficult. How close is a typical, simple, well-designed mechanical component to its optimum design? Our experience tells us that, when optimization is employed, it is not unusual to enhance the performance of a component by 50%, even after years of continuous development based on manual design and traditional analysis.

Consider this very simple example: the frame of a bicycle has five different tubes. Let us describe each tube by its diameter and wall thickness. This gives us ten parameters to choose from. How many different designs can we make? Each parameter can vary continuously, so the answer is technically "infinity". But to limit our search, we decide to test only five values of each parameter, e.g., small, medium small, medium, medium large, and large. To exhaust the design space, we must analyze the performance of 5^10 = 9,765,625 different designs. Almost ten million! And this is just a simple case. Very few, even among the simplest, components can be described in full by ten parameters. There is a very good chance that what you think is a good design has not nearly exploited its potential.

Don't be sorry about this point. It is the reason for the wonderful and infinite diversity of the world. It is also the reason why you can continue to improve our products. Rational Engineering merely accelerates the process.

sprocket | Slide gate | Turbine wheel | Truck floor

Services | References | Contact | Links

Last modified: november 15, 2001