Project Description

Final product recipes usually involve a combination of many raw materials from various suppliers.

Setting specifications on each of these incoming materials separately is an unreliable metric since the final quality is affected by the simultaneous combination of all raw materials and their properties.

Project Details

Client:Eli Lilly

The Challenge

Tablet Manufacturing OptimizationTablets have very strict quality control and need to be manufactured with consistent and stable properties of weight, hardness, API uniformity, etc.

The goal of this work was to improve the tablet quality through the use of multivariate modelling and optimization techniques.

 

 

The Results

The multivariate modeling and optimization work was very successful. ProSensus demonstrated how this approach can be used to make small, optimal adjustments to the roller compactor and tablet press operatng conditions in order to compensate for everyday variations in formulations, raw material properties and other process disturbances. The methodology could also be used to prepare the line to manufacture new formulations.

The ProSensus Approach

The first step in this work is to build multivariate models that predict the product quality of the tablets. A multi-block model is built that includes two raw material information blocks (Z1 and Z2) and two block of process settings; one for the roller compaction process and one for the tabletting process settings. These 4 blocks are used to predict the tablet properties.

Optimization Pharmaceutical Manufacturing

A model with excellent prediction power was developed, that can not only be used for process optimization, but also to help with process understanding. The plot below shows the relationships between the various X data blocks and the Y variables (in red). Variables located near each other are positively correlated. For example, a higher than average value of cup depth will result in a higher than average tablet weight.

Optimization of Pharmaceuticals with MVA

Similarly, variables that are located opposite to one-another are negatively correlated. For example, API and Compaction Force are negatively correlated. These types of models provide a very powerful method to increase process understanding.

Applications for the Multivariate Models

ProSensus demonstrated how the model could be used for optimization in two ways:

1) Find the process settings for both the roller compaction and tabletting processes, given a desired Y and known raw material properties (API and excipient mass fraction). This can be used when a new formulation is to be used on the line.

2) The models can be used to make small adjustments to the tablet press after it has already been run through the roller compaction process. New optimal operating conditions are found for the tablet press to ensure that the desired product quality is met despite any disturbances that may have occurred in the raw material properties or roller compaction process.

These optimizations are performed using constrained optimization to ensure that the new process conditions specified are within physical limits and that the specified solution is not too different from past operation.

References

  1. Z. Liu, M-J. Bruwer, J.F. MacGregor, S.S.S. Rathore, D.E. Reed, M.J. Champagne,” Modeling and Optimization of a Tablet Manufacturing Line”, J. Pharma. Innov., 6, 170-180, 2011.