January 1, 2006 (Vol. 26, No. 1)

Find the Best Operating Scenario with Process Modeling

Can our existing labor force handle a 20% increase in production? If the ultrafiltration step takes four hours longer than usual, how can the remaining process steps best be rescheduled to avoid interfering with the next batch? Does the flexibility of using disposable bags justify the additional operating costs? If we CIP the tank and the lyophilizer at the same time, are we going to run out of water-for-injection (WFI) or overload the pH neutralization waste system?

These are questions encountered daily by operating management at biopharmaceutical companies seeking to either maximize utilization of their facilities, lower production costs, or resolve conflicts in production schedules.

Contract manufacturers look for windows in their schedule to take on new customers or to fit in that extra batch per week, facility engineers try to prevent the USP tank from running low half way through the process and with the increasing commoditization of biopharmaceutical products, success for manufacturing companies will likely rely on minimizing costs to maintain a competitive edge. All strive to succeed with the resources available.

Meeting production challenges is often complicated given the complex system characteristics involving relationships between process demands and resources availability. Because resources are shared throughout the facility, engineers must consider how a change in one operation can affect concurrent or subsequent operations. In addition, there are usually tradeoffs to consider.

Imagine, for example, the demand for your product increases and requires you to increase your production output, but the demand on raw materials, labor, and utilities increases as well. Can your facility sustain this increase in production or are modifications required? Or better, how can you increase production with the smallest impact on your resources?

For labor, perhaps there is a way to modify the production schedule to alleviate the demand on operators, but does the new schedule overload your utilities? If your operators are busy with a filtration step that is taking too long, will it delay the fermentation step for the next batch?

Modeling can provide quantitative answers to help determine what you need to increase production. A model of your process and facility allows you to simulate modifications to your process so you can study the differences between several operating scenarios. Modeling can also help develop system characteristics for different aspects of your facility so you can understand limitations and capabilities not immediately evident. To support our consulting work, Biometics (www.biometicsma.com) uses modeling as a facility design tool and to help clients compare alternative production strategies.

This article discusses a couple of illustrative examples. The first uses a simple, hypothetical model to convey the use of the model. The second example uses results from a model developed for an actual fermentation process. Both examples look at the affect of the operating scenarios on a facility’s water resources. The example models were developed using Intelligen’s (www.intelligen.com) SuperPro Designer. This program is specifically designed for modeling bioprocesses and uses a graphical user-interface to collect process and facility resource information. There are other comparable modeling software packages available to the bioprocess industry.


Fig. 1: WFI usage profiles and average usage for a simplified hypothetical process.

Adding a Second Production Train

The first example is to show the impact on a facility’s WFI system when the demand from a second, identical production train is added. This example also demonstrates how decisions made with respect to the production schedule can be adjusted to reduce this impact. WFI is one of the more expensive utilities required for bioprocesses and, as such, is usually one of the first to run out and become a bottleneck when the demand on a facility increases.

Two basic properties define the capacity of a WFI system: generation rate and operating capacity in the storage tank. The operating capacity is not the working volume of the tank, but the range between the normal maximum and normal minimum levels during facility operation. In determining the generation rate and operating capacity, you can trade one off for the other and still meet the facility’s WFI demand. A small operating capacity can be combined with a large generation system, or a larger operating capacity can be used with a smaller generation system.

There may be several other considerations that go into sizing the generation system and operating capacity, such as emergency reserve volumes or floor footprint limitations, but you can typically explore a set of generation rate and operating capacity combinations. The model considers the WFI system characteristics or the combinations of generation system size and operating capacity that can support the WFI demand.

Consider the base case, a single production train of a hypothetical process. In Figure 1a, the red line shows the WFI usage profile for a single production train over four, 8-hour shifts. The average WFI usage rate over this time period is shown in blue. Note that the maximum instantaneous draw is 20 kg/min and the average draw is approximately 6.5 kg/min.

Now, consider the combinations of generation rate and operating capacity that can be used. Assuming the generation rate is fixed, we used the model to calculate the WFI system characteristic curve depicted as line A on Figure 2, which shows the relationship between the generation rate and tank operating capacity in the context of meeting the demand profile in Figure 1a. This line represents the minimum requirement or boundary for the WFI system that can support the single production train.

In this single train scenario, if the generation system is producing WFI at 10 kg/min, the storage tank must have an operating capacity of at least 1,000 kg. From the graph you can also see that the smallest generation system that can support the single train is one that produces WFI at about 6.5 kg/min, equivalent to the average WFI draw. The accompanying operating capacity would have to be above 1,350 kg. As the generation rate approaches the maximum facility draw rate (~20 kg/min), the operating capacity needed approaches zero.

Now that a base case has been established, look at what happens to the WFI system when we change the operations. When a second identical production train is added to the facility, the average WFI usage rate doubles. Figure 1b shows the WFI usage profile when both trains are running simultaneously. In this case, the maximum instantaneous draw also doubles. In Figure 2, the calculated WFI system characteristic curve labeled B depicts the generation rate and operating capacity relationship for this demand profile.

The system requirements for this scenario are much greater than that for the base case. Note, that the range in production rate for WFI system characteristic curve B is bound by the average draw (13 kg/min) and the maximum draw (40 kg/min).

Though the impact of the second train on the WFI system seems obvious, with the appropriate production strategy we could mitigate or even eliminate the increase in systems requirements. The model can also be used to look at how changing the schedule of the two trains changes the WFI usage profile and therefore the WFI system characteristics. Figure 1c shows the WFI usage profile for both production trains, but with a 50-minute difference between start times. The average and peak draws are the same as in the simultaneous production profile; however, the shapes of the profiles are different.

The WFI system characteristic curve labeled C in Figure 2 shows how this shift in schedule affects the requirements on the WFI system. Above a generation of 20 kg/min, the operating capacity requirement is cut in half as compared to line B. Below a 20 kg/min generation rate the operating capacity requirement is still lower, but converges with line B. Figure 1d and its corresponding WFI system characteristic curve D tell a similar story but for an 80-minute shift in start times.

When the start times of the two trains are shifted by 100 minutes, the WFI demand profiles cease to overlap and the maximum draw is back down to 20 kg/min (Figure 1e). The WFI system characteristic curve labeled E in Figure 2 is now bound by this maximum draw, so as the generation rate approaches 20 kg/min, the operating capacity goes to zero.

Finally, in Figure 1f and WFI system characteristic curve F, with a 150-minute shift the requirements of the WFI system are almost equivalent to what they were in the single train scenario. However, there is a difference between the A and F curves, because though the maximum draws for both scenarios are the same, the averages of the two WFI usage profiles are different. Curve F is limited between 13.5 and 20 kg/min, but A has a range between 6.5 and 20 kg/min.

By selecting the production schedule in the F scenario, the WFI system used for the single production train is sufficient to handle the second train as long as the WFI system has a generation rate greater than 13.5 kg/min. If the other resources can handle this schedule, no modifications to the WFI system are necessary in order to reach the production goal in this example.


WFI system characteristic curve.

Compressing the Schedule

Clearly facilities are more complex, so in this second example a more realistic process demand and one for a different utility (USP purified water) are considered. Similar to a WFI system, a USP system’s characteristic curve can be defined by its generation rate and operating capacity with respect to the process demand.

Figure 3 shows two USP usage profiles for an actual fermentation production train. The first profile is the base case and shows the demand over a 10-batch campaign. The peak usage rate is just under 70 kg/min and the average draw is about 12.5 kg/min. The second profile is for a more compressed schedule of the same process and the same 10 batches (Figure 3). This alternate case produces the same amount of product in 30% less time. The average USP draw in the alternate increases by almost 50% and the maximum draw increases by about 15%.

Figure 4 shows the USP system characteristic curves for these base and alternate cases. As in the first example, both curves have a generation rate range between their respective average and peak draws. With a generation system less than 35 kg/min, the operating capacity requirements between the two cases are different, but with a system that generates over 35 kg/min there is hardly a difference in operating capacity requirements between the two cases.

If the facility was originally equipped with a USP generation system greater than 35 kg/min, it would be able to handle the increase in production, at least with respect to USP. If the generation system was between about 20 and 35 kg/min, a larger operating capacity would be required. If the system was under 20 kg/min, no change in operating capacity would help and a larger generation system would be required.

Modeling can be used to reveal a facility’s limitations and find ways to work around these limitations while still meeting production goals. The examples focus on utility usage, but modeling can also look at other resources such as labor or process equipment occupancy. Facility resources can be studied individually or collectively using the same model.

Initially, facilities may be designed with certain products and production rates in mind, but it can be difficult to foresee facility needs a few years out. Facilities may be suitable for their defined purpose, but as demands grow or change, production can easily get utility-limited and equipment sizes may present bottlenecks. Models of the facility and its processes can prepare companies to make quick decisions about what is possible with existing resources and how to go about planning to meet production goals.


Fig. 3: USP usage profiles for an actual fermentation operation.

Previous articleTaiwan Industry Puts Strong Focus on Biotech
Next articleFAS Biological and Chemical Weapons Control