|Customer||Major steel manufacturer|
|Need||Steel manufacturers produce batches of products for each order that must satisfy customer-specified requirements such as the steel’s physical dimensions, tensile strength, yield point, and chemical composition To fulfill orders profitably, steel companies must select process settings and operating conditions for each batch that satisfy the customer’s specifications while optimizing manufacturing efficiency. Selecting the best process settings and operating conditions is very complex because as many as 100 input and 100 output variables may require consideration.|
|Solution||Stottler Henke developed a case-based reasoning system that enabled a major steel company to design the manufacturing process for each new order quickly and automatically. The system generated an initial manufacturing process for each order by searching a database to select process information and composition data about batches with product specifications that were similar to those of the current order. Users could then request that the software adjust these process conditions and composition to satisfy the current order’s specifications by using known relationships between manufacturing inputs and outputs and a set of priorities for choosing among the alternative adjustments.
By encoding process knowledge within this knowledge-based system, the steel manufacturer was able to retain expert knowledge of operators who had left the job, train inexperienced operators more quickly, reduce the number of mistakes by both inexperienced and experienced operators, set-up of batches more quickly, reduce the frequency of off-spec batches; and make better cost estimates and quotations for new orders.
|Status||Stottler Henke developed and delivered the process advisor to the steel manufacturer.|
|Related Applications||Case-based reasoning can be used in batch-oriented manufacturing areas (chemicals, semiconductors, food, rubber, textiles) where many input variables and process conditions must be set to produce batches of product efficiently that are within spec.|