The PharmaMV platfom is used to apply process response tests to a mechanistic process model, to generate a statistically rich dataset. This data is used to create a reduced order model, which is incorporated into a Model Predictive Controller (MPC), then simulated and tuned in-silco. The MPC is subsequently commissioned on the physical process, achieving equivalent performance to the in-silico system, demonstrating a reduction in material and energy consumption, as well as minimised experimental effort in comparison to a step-by-step empirical approach on the physical plant.
Perceptive Engineering, in collaboration with CPI, PSE and AZ, have developed an APC solution for a Twin Screw Wet Granulation (TSWG) process using the digital design approach. The solution, developed as part of the ADDoPT project, used the digital twin for process response testing and APC design in-silico. The tuned APC scheme was then applied to the physical TSWG process to control the particle size (D50) to target with the real-time D50 predictions coming from the Digital Twin. This digital design workflow resulted in 60% fewer trials and substantially reduced API consumption.