For Bioprocess Success: Simulate!

I’ve been using simulation techniques for almost 15 years in various areas, including scale up, technology transfer, and equipment design and coordination – and I passionately believe that the approach has many benefits. Right now, the pharma and biopharma industries are moving into the next generation of manufacturing. Many advances are hitting manufacturing both in terms of equipment and processes – consider the move to more cost-effective production and in some cases continuous manufacturing. And let’s not forget the fast-moving world of cell and gene therapy, where many new companies are jumping into commercial manufacturing for the first time. Particularly when it comes to cell and gene therapies, many companies are struggling to select the best equipment and processes because they are pioneering new approaches. Often, questions about a project simply can’t be answered with pen and paper, or an Excel spreadsheet – ultimately, you are just guessing. Modeling tools, however, can provide more accurate insight and ensure that you are designing your facility to work the way you want it to.

A process model is a computerized representation of a real world process and can be used to either reproduce the past for model validation purposes – or to predict the future. Models can be employed in a range of applications, including evaluating cost of goods, debottlenecking processes, planning clean-in-place, utility sizing, equipment selection, architectural planning, and even warehouse planning to support supply chain management. The beauty of modeling? It’s highly customizable nature. You can use it for a high-level activity or intricate production details – and the findings are often fascinating. (For example, in warehousing, you’d be amazed how much time savings and congestion reduction can be made by simply evaluating the pathways chosen by forklift truck operators!)

Divination in practice

In biopharma, modeling is often used to quickly evaluate different production scenarios.  When properly performed, a simulation can alleviate project unknowns and be used in tandem with traditional engineering design to efficiently design facilities suited for both current and future production. Notably, a process used today will not stay the same in the future – titers may increase, and there may be changes in technology. A model can be used to predict some of those changes and how they will affect support equipment. For example, a process today may have a titer of 5 g/L, but what if, in two or three years, this rises to 7 g/L or higher? If you need to make twice as much buffer, can your buffer handling operations cope as they are designed now? Or would it be better to design the facility with flexibility so that changes can be made more easily in the future?

In my experience, design projects always go more smoothly when simulation has been used upfront. Today, many companies are keen to move from stainless steel to single use, but I often find that those companies don’t have a great deal of data on how much single use is really going to cost. Single use may have a lower cost in terms of facility design and initial outlay, but the consumables cost is usually much more than companies expect. Here again, simulation can help by setting out exactly what the consumables cost will be, allowing a company to examine all the different ancillary pieces that they may not have thought about at first. When moving from a stainless-steel vessel to a single-use mixer, it’s clear that you will need single-use bags, but companies can overlook all of the tubing needed to connect the single-use mixer to other pieces of equipment. Modeling and simulation can reveal that information upfront, giving you more data on which to base your decision – rather than just finding out after you’ve made the switch. In one facility design project I worked on, the client was keen to have a primarily single-use facility, but once cost-of-goods modeling was performed and the data presented to the client, stainless steel proved to be the more economical long-term solution given their planned production rate and scale. The resulting data caused a complete change in design, showcasing the true value of simulation: its ability to drive a project forward in the best direction.

I’ve also used modeling for clean-in-place debottlenecking in a stainless steel facility. We had to model the pathways of flow in incredible detail, including going right down to individual valves. This couldn’t have been done without simulation software.

The model maker

Before building a model you need to know what you want to achieve. Too often, I’ve seen companies with a “fuzzy” picture who plan to build a model and “see what comes out of it”. If you don’t know what you want your model to do, you won’t be able to collect the right data – and to make a good model, good data is paramount. The better your data, the better your model! Getting the right data is usually the biggest challenge in making a model and will involve walking about the facility, talking to operators and examining batch records. The models I have worked with have been very accurate when it comes to comparisons against real production data. There will always be variables in real world production, but a good model should give you a good understanding and appreciation of the overall feel of the facility and what throughput is going to be like. As the model evolves, however, you may find new areas to explore as the model brings information to light about production processes.

There are dozens of commercially available simulation and modeling tools on the market – each has their own pros and cons and application areas. I advise using a specific software platform that is best suited for the intended application – don’t just assume you can perform production debottlenecking and cost of goods analysis with the same software tool! Once you have the data, you should typically build the model as the facility is currently functioning. If the model is able to mimic the process correctly, you know it works and you’re ready to run “what if” scenarios.

One final piece of advice: when embarking on your modeling journey, it is very important to remove bias. Some people distrust models or believe that they know their own processes so well that there is no need for them. But modeling can often bring to light bottlenecks or room for improvement that were not even on the radar. In many cases, particularly with cost-of-good models, people have assumptions upfront about what would be cheaper, and so use these biases to influence the data entered into the model. Other times, people may not use the model effectively or only run a single brief simulation. With modeling, you must be objective. Choose the right software, collect the right data, and be thorough with your simulations. Leave your biases at the door and let the data speak for itself!

Emily Thompson is a Process Engineer at CRB.