In addition to these types of summary metrics, discrete simulation allows the design teams to see every meaningful detail
about equipment and materials flowing through the process. For example, Table 2 shows that a batch of material waits an average
of 3.2 hours for a cart to be available in this scenario. There are detailed metrics and graphics to help understand the
reasons. Figure 3 shows a graphic of utilization of the lines and lyos on a time line over the duration of the simulation
run. It is one example of why a cart is not available, that would be included in the summary results.
A similar example of resource utilization for a fixed conveyor configuration is shown in Figure 4. The sequence of preprocessing
decontamination steps for each zone is depicted, rather than the cart activity.
RESULTS FROM SIMULATION
For this project, with the product mix and volumes projected, and for the constraints of the facility design the team was
working with, there were several findings that resulted from the modeling that would not have been possible, or would have
been unlikely uncovered without it. These included:
- Staffing levels would have to be 24/7 even at lowest anticipated volume levels in order to have skills available to perform
all the required operations.
- The cart single track constraint could be a limiting factor even with the 3.5 day lyo time, given the sensitivity that existed
with wait times for the carts.
- In the fixed conveyor system configuration, there was less buffer time available to make the required volumes due to the requirement
for VHP decontamination of the conveyor zones.
- For the 3 & 3 configuration, the conveyor cleaning time after unloading became a constraint at about 8 hours total time.
- For the 4 & 2 configuration, the time after unloading did not have an effect until 12 hours.
- The variability in times after unloading for manual cleaning and changeover activities could have a significant effect on
throughput, and were also managed in an ongoing Six Sigma program.
- Alternative schedules to process batches on regular repetitive cycles should be simulated as the project went forward to attain
the level of throughput consistency required.
Given the cost of developing and bringing a new drug to market, and the investment required for manufacturing capacity in
the commercial and earlier stages, discrete simulation technology should be considered an essential tool. The examples shown
above were meant to demonstrate that the tools are robust enough to handle even the most complex rules we face in designing
facilities, processes, and schedules to run an operation. This project used the Extend simulation program from Imagine That,
Companies use spreadsheets with averages for the processing times involved, and project planning systems with probabilistic
ranges of times. Neither of these is as robust as the latest discrete simulation tools. Simulation proves the point very specifically
that "idle time can never be made up" when there is a constrained resource in the system. In other words, things do not even
out when using average times.
To successfully implement these types of simulation tools requires training programs and commitment. The programs are already
being taught in both undergraduate and graduate programs. Therefore, there are engineers and analysts who are well-versed
with these tools. A key to success is giving people the time to learn and work with them.
Jim Curry is the CEO of OpStat Group, Inc., Ridgefield, CT, 203.431.3905, jimcurry@OpStat.com