A learning curve model for a multiple batch production process

Abstract

This study develops a mathematical model which describes the learning effect in a production process where batches of different sizes are produced. The model separates set-up time from production time, although for the estimation of its parameters these times do not have to be recorded separately. The data required to estimate the parameters of the model is commonly accumulated in organizations, and includes the size of consecutive batches and the respective times for their production.

A quasi-Newton search method is used for estimating the parameters of this model. This is carried out by minimizing the sum of squared deviations between observed and calculated batch times.

We also demonstrate the use of the model for the estimation of the parameters of a given production process, and the application of the estimators obtained in this way for the prediction of the total time required for future hatches