Case Study: Scheduling, Rescheduling & Design
This case study will take a virtual tour of the highlights of the unique scheduling capabilities provided by the VirtECS Scheduler. as an example we'll look at scheduling a fermentation operation, starting with a very simple model, then adding more realistic and complex conditions. Next we'll consider some important rescheduling considerations, and finish by addressing some challenging scenarios for process modifications or redesign.
1. Basic Fermentation Process
This is a typical piece of a pharmaceutical process. While this system is relatively small, it illustrates a level of complexity that is non-trivial to schedule.
- 1000 lb of material is processed in a fermenter with a batch time of 1 hr
- When fermentation is complete, 500 lb of broth may be transferred to a feed tank, requiring 20 mins
- When the transfer is completed, the Fermenter is empty and available to start another batch
- Broth in the Tank is fed into the Column (10 mins) and the broth is refined, requiring 83 mins to produce 410 lb of product
- One fermenter batch requires two batches of Refining on one Column
2. Process Vessel Storage
Note that the second batch of Fermentation cannot begin until the entire first batch has been fed to the Tank. This requires accounting for the time to feed the Broth into the Tank, start the Refine task, complete the Refine task, and feed the second half the Broth into the Tank just in time to begin the second Refine task. Only after the second transfer into the Tank is completed can we begin the second Fermentation task.
3. Unexpected Equipment Shutdown
Of course, if something happens which requires shutting down the Column (C1) for an hour, the schedule must be recomputed. For this reason, the timing of the tasks on one piece of equipment interacts strongly with those other pieces of equipment.
It is easy to implement a drag and drop interface allowing boxes on the chart to be moved about willy-nilly. But such moves will, in the absence of very strong model-based underpinning, produce infeasible schedules which are useless. Suffice it to say that any scheduling system which cannot automatically produce a feasible schedule for this simple process (without user intervention) will also lack the necessary modeling support to ensure feasibility under drag and drop manipulation.
4. New Equipment Considered
Now we begin to see the full power of VirtECS - the ability to automatically produce fully detailed feasible schedules for a realistic process. We have added a second fermenter and two additional columns with all three columns sharing a single feed tank. We can see here that the columns running the refining task are the rate limiting piece of equipment.
5. Rework Within Process
Now we add one further complicating factor which is common in pharmaceutical processes. The Fermentation task not only produces good product, but a small amount of material that must be fed to a rework tank. This material accumulates until at some point there is enough available to feed a special rework task on one of the fermenters.
The Rework task requires three hours to complete, but converts all the off-spec material into broth suitable for refining. The Rework task must be run with some regularity because of the limited storage in Rework Tank. Note that VirtECS handles these complications automatically, requiring less than two seconds of CPU time to produce the schedule shown. Because VirtECS does not have to be "taught" or "configured" to generate these schedules, it can also easily and automatically respond to contingencies which arise in daily operation, as shown next.
6. Equipment Maintenance
Here we see how effortlessly VirtECS is able to automatically generate schedules that walk around four different equipment outages.
Process Improvement Scenarios
7. Advanced Scheduling
Here we begin to use VirtECS to determine the value of various proposed improvements to the process. This is the base case where we set the demand to require six days to complete.
Even this small process contains constraints which most other schedulers on the market cannot handle properly. The addition of a larger demand requiring more batches illustrates this point. Note the storage activities are shown as pink boxes.
8. Process Time: Fermenting
Here we evaluate the capacity increase derived from reducing the process time on the ferment process from 60 to 50 minutes. The makespan is reduced by 0.14%.
9. Process Time: Rework
When we reduce the process time for the Rework task from 3 hours to 2 hours 40 minutes, the makespan is reduced by 0.27%, twice the capacity increase achieved by the previous change. This result is surprising and counter-intuitive. The capacity increase to be had by a 12% reduction (20 minutes) in process time for the Rework task, which is run relatively infrequently, is worth more than a 17% reduction in process time for the common Ferment task.
These results highlight that you need not necessarily make large changes to a process time to see a significant improvement in results. This highlights the need for a rapid automated scheduling engine, to allow you to consider a wide range of improvements and specific changes in process parameters.
10. Adding a New Column
Here we use VirtECS to predict the capacity of the plant if we add a fourth column (C4). The capacity increase from the fourth column is a full 25%. The complex pattern of processing, storage, accumulation and Rework-Broth and reworking would be very difficult to work out using conventional methods. With VirtECS it took less than two minutes to make the changes to the model input data and three seconds to resolve and get results for the new test case.
11. Process Time: Columns
Another way to improve refining capacity is to reduce the process time in the columns. This case shows the result of reducing the process time for the Refine task by 10 minutes. This produces a significant capacity increase of 6.9%.
12. Fill Time on Feed Tank
Even though the feed tank is not busy all the time, it may be a significant bottleneck in this process. Again, it is not a simple problem to address without the ability to automatically generate quality schedules. In this case we investigate the performance of the process if the time required to feed the material from Fermenters to the feed tank is reduced from twenty minutes to ten minutes. This modest change in the process generates a 2% improvement in capacity.
This case study has examined a basic fermentation process, simplified to be able to visualize what's going on, but with enough complexity and realistic process constraints to highlight some of the key features of Version 7 of VirtECS Scheduler. The case study began with a simple process and added complexity including process vessel storage, unexpected equipment downtime, scheduled maintenance, consideration of new equipment, and a number of process improvements.
Conclusions to the Scenarios Considered:
- Decreasing fermenting process time from 60 to 50 minutes had a relatively small impact, decreasing makespan by 0.14%
- Decreasing the rework process time from 180 to 160 minutes had twice the effect on makespan, reducing it by 0.27%
- Adding a fourth column increased total capacity by 25% (this may not have been the case if other process steps were rate limiting)
- Observing that the columns were an important determinant of capacity, a new scenario reducing refining process time by ten minutes gave a 6.9% increase
- Decreasing the time required to fill the feed tank from the fermenter improved total capacity by 2%
Problems Other Scheduling Engines May Encounter:
- Process vessel storage and limited or time-sensitive storage
- Frequent rescheduling to respond quickly to equipment downtime
- Rework of material and 'cycles' in the process network
- Adding new pieces of equipment, especially shared equipment
- Automatic enforcement of firm constraints which may not be violated
- VirtECS Scheduler was seen to perform admirably under these conditions
APC is committed to helping you succeed and providing outstanding business value. The point of having superior technology is not for the sake of technology, but to allow our product to handle the details that give you a competitive advantage. If you have questions on what aspects of your process might make scheduling difficult, or to see VirtECS Scheduler in action with a demo on your process, please contact us.