Master chefs have been using some of the concepts of GLP, GMP, and PAT long before those TLA's existed. I consult master chef Julia Child a lot. Often, I find a recipe that looks deliciousand so easy! Then I discover that ingredient number three is "Sauce Brune, p. 67." Turning to page 67, I realize that the sauce requires another ten ingredients and an extra two hours!
Chefs understand the benefits of modularization, reuse, and repeatability. Perfecting a brown sauce or white sauce gives you the basis for many variations in different meals.
Designing Modular Control
Process control in a pharmaceutical facility should be designed in the same way. The components of a complex process can be thought of as "recipes." Control of a vessel's temperature during a batch run might be one recipe. Monitoring oxygen levels and controlling the flow of oxygen can be handled by a different recipe.
Stirring a mixture faster or slower based on the timing and amount of ingredients fed into the mix could be a third recipe. The different recipes can interact, cooperate, or be separate in the processit's all up to the design and the nature of the process.
FermWorks from WireWorks West (www.wireworkswest.com) helps automate fermentation and cell-culture pharmaceutical processes at labs and pilot plants.
However, FermWorks' concept of modular recipes for the pharmaceutical industry can be generalized to control of any process. These ideas are not new and are not limited to any one sciencechefs, writers, and software developers all know about breaking a problem into parts. Breaking a control strategy into separate recipes has many benefits:
Modular recipes make it easier to develop efficient strategies for each part of the process.
Recipes make complex process control more manageable.
Each recipe is a tested, verified, repeatable component of the overall process, making the entire process easier to validate and repeat.
The recipes can be reused, combined, and recombined, saving development time.
A recipe can be reused with variations in parameters.
One part of the process can be refined, without affecting the rest.
The recipes may interact, even though they are developed separately.
Different recipes can run in parallel, sequentially, or cooperatively by handing over control from one to another when certain conditions are met.
A set of recipes can integrate different instruments and sensors into a unified control strategy.
The recipes can be scaled up, taken from the research and process development labs to larger-scale production.
There are many examples of modular recipes from the pharmaceutical cell-culture process. Both continuous and batch control can benefit from modular recipes.
The growth of cells to secrete certain products may require either continuous or batch control. During continuous control (such as perfusion) process variables, such as oxygen flow, temperature, and pH, are monitored and controlled to keep the cells healthy. Each of these process variables may have a separate control algorithm.
During a batch run or experiment, these and other variables are controlled in different ways. The temperature of one fermentor might be kept higher than normal for the purposes of the experiment, or the agitation may be ramped for varying dissolved oxygen levels. Each variable or related group of variables is controlled by an individual recipe.
Traditional Choices for Control
How you achieve a modular design in your real-world process depends on how the process control is implemented. At the low end of the spectrum, an operator adjusts pumps and valves by hand, perhaps following written instructions. This is crude control, because the operator cannot respond as quickly or effectively to changes in the process as automatic control. Because this manual control is more variable, it is more difficult to achieve repeatable results.
In many situations, control is implemented by a programmable hardware device, such as a biocontroller or PLC. These devices allow some flexibility in their control strategies. Sometimes, operators interact directly with a biocontroller's front panel. This is the same as manual control if process variables are affected individually.
However, some biocontrollers allow sequences or programs to be entered at the front panel. These panel interfaces are cumbersome for nontrivial tasks.
At the next level, a control strategy can be downloaded to the biocontroller (or PLC) from a PC. The user interface on the PC might just mimic the biocontroller's panel, or it could be a more sophisticated application. However, software applications supplied with a particular instrument do not give comprehensive control of the larger process.
Some systems employ a sequence editor in which you can express your desired control strategy. PLC applications use ladder logic. Some systems follow the ISA SP88 standard for batch control language, and others define their own sequence language.
Sequence editors provide a predefined, finite set of building blocks that you may select and then arrange in a certain order. A good system lets you create and add your own building blocks to that set, gives you control over execution of the sequence, and manages multiple sequences running in parallel.
A last-resort implementation option is custom software and/or hardware that you develop and maintain in-house. Few process-focused facilities would want to maintain the engineering resources to do this.
Automated Process Control
The best systems do not just control one instrument, but automate control and manage data for the entire process or facility. These systems provide the full benefit of recipe modularization, using an implementation method that meets certain criteria. You must be able to:
Express your control strategy with the building blocks provided, or be able to add your own blocks.
Divide your control strategy into self-contained parts, as recipes.
Implement and test each recipe individually.
Choose how multiple recipes are executed, whether in parallel, sequentially, or cooperatively.
Share or pass information between recipes.
Reuse a recipe in a different run.
Define and modify options to a recipe when it is reused.
Change one recipe without affecting others.
Track changes to recipes for repeatability and accountability.
Integrate biocontrollers with other instruments and sensors.
Scale up, using the same recipes for larger volumes.
WireWorks West works with pharmaceutical companies to help them implement modular recipes and other aspects of process automation and data management. They use the bioreactor control system called FermWorks. FermWorks is a unified, scalable solution for comprehensive management of cell culture and fermentation processes. It meets the criteria listed above.
Using FermWorks, process development researchers develop their own recipes, mixing the building blocks provided by FermWorks with some of their own design.
They decide how the recipes are executed, how they interact, and what factors an operator may vary. They can quickly refine their control strategies and develop new experiments based on existing, verified recipes. The recipes cooperate with the biocontrollers and integrate other instruments as well, regardless of the manufacturer.
These recipes scale up, from the 2-L fermentors in the researcher's lab, through process development's midsize 10-L vessels, to hundred- or thousand-liter vessels.
The complexity of the recipes is hidden (remember the Sauce Brune?) from the technicians who run experiments on a daily basis. But the master chef, the researcher, has all the tools he needs for flexibility and innovation.
Effective Control Means Higher Quality at Lower Cost
Control of any process is extremely important, especially in the process development phase. Effective control determines both the quality of the resulting product and the efficiency of the process. Poor control means you have a poor product as a result, or no product at all, and you wasted all the time and resources that went into the process. Adequate control might give you the product you want, but you might have expended time and resources unnecessarily.
Tight, efficient control gives high-quality results with low expenditure. Add flexibility, innovation, and automation to that efficient control strategy, and you can drive that expenditure of time and resources lower than was thought possible.
When process control is designed in a modular way using FermWorks, you get a unified, scalable solution that automates control and data management from small-scale process development labs up through production facilities. But the ultimate benefit is this: high-quality results for minimum expense that accelerate the commercialization of new biotherapeutics.