The Data Centric Architecture of a Factory Digital Twin
carrying more final product inventory , but this solution may be more attractive economically than plant expansion . While reducing inventory has become a watchword for plant efficiency , this example illustrates that tolerating additional inventory can sometimes make good economic sense . An FDT can be used to quickly explore alternatives that result in the best process behavior .
5.2.3 FDT ANALYSIS EXAMPLE 3
An example biologics process consists of nine processing steps that are divided into a front end and back end . The front end consists of Inoculation followed by three sequential stages of Bioreactors , 750 , 2500 and finally 10,000 liters . All stages in the front end have three parallel sets of equipment . The final Bioreactor stage ( 10,000 liter ) is the anticipated process bottleneck . The output from this reactor is stored in centrifuge feed tanks from which it enters the back end of the process . The back end consists of a centrifuge , and three chromatography columns , Anion Wash , Revo , and CatEx . The final step is Filtration .
The chromatography columns have cycle times ranging from 8 to 21 hours . Batch integrity must be preserved so a single batch from the centrifuge is fed to the Anion Wash column , then flows to the Revo Feed Tanks , Revo Column , CatEx Feed Tanks , CatEx Column and on to the Filter .
5.2.4 SPECIALIZED PROCESS CONSTRAINTS
In addition to the normal constraints implied by the physics and chemistry of the system , e . g . material balance , storage level limits , material availability , etc ., this process entails three classes of specialized constraints . First is batch integrity . Batches are required to be completely emptied into the next stage and the storage tanks washed out ( this takes one hour ) before the next batch may be introduced into the storage tank . The FDT must ensure that the requisite hour of time between sequential batches is present so that the schedule the system produces is actually executable in the plant .
The second class of special constraints involves the chromatography columns . Each of the three columns must be periodically repacked after a specified number of cycles . The maximum number of cycles between repacks are Anion Wash 20 cycles , Revo 17 cycles , and CatEx 15 cycles . The repack times vary from 15 to 24 hours . The FDT must track the number of cycles since the last repack on each column and must ensure that the repack tasks are executed at the required frequency .
The third class of specialized constraint results from random delays that can occur on the chromatography columns due to operational variability . A pattern and probability of such delays can be specified as part of the study input . This allows design engineers to experiment with the process before it is built and determine the expected resiliency with respect to expected levels of variability . This is an important tool for design engineers who can use it to ensure that the realworld performance of the plant will live up to nameplate expectations in the face of stochastic disturbances .
Journal of Innovation 115