Introduction to Discrete Sizing

This section introduces the concept of discrete sizing, along with the possible approaches for it.

This page discusses:

Need for Discrete Sizing Methods

Standardization is a very important part of most manufacturing and fabrication processes. In most cases, fabricators use certain fixed sizes of sheets and thus the equipment and processes are calibrated to work efficiently for these set of sizes. Using a gradient-based optimization method like CSA (Convex Separable Approximations) the output from the optimizer is usually continuous and sometimes impossible to produce without precision equipment. Also a lot of time must be spent on designing new fixtures and calibrating the processes for the optimized sizes. A solution to the above problems is the discrete sizing optimization.

Possible Approaches

Discrete Optimization Heuristic Methods
These are a family of mathematically consistent methods that work toward finding discrete global solutions. However, they are not computationally efficient and impossible to use for industrial scale models. A heuristic is any approach to problem solving that uses techniques that are meaningful from an engineering point of view but are not guaranteed to be mathematically correct. In context of sizing optimization with discrete design variables heuristics are used to modify a gradient-based optimization algorithm that is based on continuous design variables to enforce their values to be discrete by the end of the optimization process.

A heuristic scheme is implemented to ensure computational efficiency and compatibility.