(465d) Optimized Startup and Shutdown Strategies for Continuous Pharmaceutical Manufacturing

Continuous manufacturing is a developing area in the pharmaceutical industry with increasing efforts focused on transitioning from the batch to the continuous mode both for drug substance and drug product manufacture. Some of the key requirements for this transition are improved process models, robust on-line sensing technologies, as well as effective supervisory and regulatory control strategies. This research focuses on one aspect of process control, dynamic startup and shutdown control with application to continuous tablet manufacture using a dry granulation process. Being able to optimally control dynamic states of the continuous process is essential for ensuring efficient use of materials, especially the drug substance.

Startup and shutdown control for the process has two overall goals: to minimize the amount of wasted material during the transient phase, and to limit the magnitude of any step changes made, which reduces the strain on equipment and controllers. To achieve these goals, an optimal control strategy is developed, using model-based dynamic optimization techniques to choose optimal setpoints.

The dry granulation line consists of two feeder units, one for Active Pharmaceutical Ingredient (API) and one for excipient, a blender, a roller compactor with attached granulator, a second feeder and blender for adding Magnesium Stearate (MgSt) lubricant, and a tablet press. Instrumentation on the pilot scale equipment is used to both assess the fidelity of the process models and the efficacy of the optimized strategy itself. Each unit op has an associated process model, developed with gPROMS modeling software. These models are used with gPROMS’ Optimization package to determine setpoints in the optimal strategies.

The optimal startup strategy begins by determining the optimal final state. Steady state optimization is performed on the process models, maximizing production rate while keeping key process parameters, such as ribbon density, blend composition, and tablet weight, at desired specifications. Dynamic optimization is then performed, which minimizes the mass of “off-spec” material generated during startup. Additionally, constraints are placed to limit the difference between successive set points. The system is brought online in three phases. First, the first two feeders and blenders are turned on and allowed to fill the hopper on the roller compactor. Next the roller compactor is brought online with a series of step changes from the minimum operating speed to the final speed determined. Intermediate set points are determined from the optimization. The granulator and lubricant feeder are also turned on at this point. Finally the tablet press is activated after its hopper is sufficiently full.

The optimal shutdown strategy operates in a similar way. First, the feeders are shut off to prevent more powder from entering the system. While the blender is emptying, the roller compactor is slowed down to its minimum operating speed, and stays at that point until its hopper is empty. The tablet press lowers its production rate to match that of the roller compactor, then shuts down after its hopper is empty. All intermediate setpoints are determined from the dynamic optimization.

These techniques can be compared with other strategies, including a base case where the unit ops are brought online sequentially, to determine their overall effectiveness. This can be done by performing the optimal strategies on the testbed equipment and measuring the amount of offspec material generated for each case.