Mathematical model to predict coat weight variability in a pan coating process

The variability in drug release from extended-release products is strongly dependant on the tablet coat weight variability. A mechanistic model to predict the coefficient of variation (CV) of the tablet coat weight is proposed. Although the main assumption is complete random mixing, the model also assumes that each tablet spends a fixed amount of time in the coating zone and receives a fixed amount of coating in each coating event. The number of coating events that each tablet undergoes is given by the binomial distribution. The model predicts that the coat weight CV will depend on the projected area of the tablet (a), velocity of the tablet in the spray zone (V), the number of spray guns (N(spray)), the length of the spray zone (L), the total number of tablets (N), and the total spray time (t). The CV is estimated to be equal to 100 square root of (an)/VLtN(spray). The overall R2 of the model for a side vented pan was 86% with a prediction error standard deviation of 1.3%. Two empirical correction factors were identified to explain the offset.