With the introduction of a new prescription medication, the pharmaceutical company made the lives of many arthritis sufferers a lot easier by reducing their pain, inflammation, and stiffness. This article is about how they are taking the pain out of production using simulation.
Taking the Pain out of Production
Imagine, you have just been given the task of producing the greatest pharmaceutical-drug launch of all time, and you have a short period of time to pull it off successfully. This is not new for the Pharma’s Plant in Caguas, Puerto Rico where over 20 brands of capsules and tablets are produced
Because this was a new drug with such big expectations, The company assembled an ad-hoc task force, “Major-Pain-Relief Readiness Team” (MRT), to manage the introduction and commercial production of the new product. And here is where the challenges just get started.
The most important goal of the Readiness Team was to have the product ready to ship on time for the launch. The production of the pharmaceutical company would double the production of the plant from approximately 2 billion to 4 billion units. Production challenges included determining the best delivery schedule for raw materials; determining the cycle time of the whole process and needed equipment in order to satisfy cycle time; and determining how many employees and shifts were needed at the manufacturing, packaging and laboratory areas in order to support their vigorous manufacturing schedule. Very important to these goals was the accurate calculation of resources in a timely manner in order to allow purchasing of any additional equipment and training of the personnel.
The Virtual Factory – Digital Twin
The entire manufacturing process was to be input into ProcessModel. Due to the short timeline for the project, the Process Engineer recommended a top-down approach, by developing a macro level simulation model of the whole process using existing data and estimates to model each major operation as a black box. Then, sub-models were to be developed adding more detail as needed, in order to investigate or improve the bottleneck operations at a particular stage of the Major-Pain-Relief process. This approach could facilitate handling the simulation effort by breaking the process in manageable segments and obtain quick results at each stage of the process.
The first area to be modeled was the receiving operation at the warehouse, and Esteban Perez, the Material Agent in charge of procurement, had a job to do. He developed static inventory models which were then tested in ProcessModel to determine their adequacy to guarantee the material supply for the new product. At this stage ProcessModel proved to be invaluable in planning for the receipt of goods. A model was built which allowed determining how long it would take the warehouse personnel for a full receiving process of the different components and raw materials and determining the quantity of resources needed. With this information, detailed receiving schedules were prepared and validated with the vendors.
With the output statistics from the warehouse simulation the scene was ready to integrate the whole Major-Pain-Relief process in a single simulation model. When the full production schedule was input into the model it was noticed that the capsule samples taken from the encapsulated lots started to accumulate in a long queue just at the entrance of the analytical laboratory, while its utilization was at its maximum.
The laboratory is the last step to the process before the drug can be shipped and the last thing the manufacturing plant needed was Major-Pain-Relief to sit on the shipping warehouse shelves waiting for approval. The Readiness Team did additional model verifications, but everything looked fine. So why was this queue growing so large?
The analytical laboratory is critical in the production of any drug, including Major-Pain-Relief. Laboratory calculations had been done using static Excel spreadsheet capabilities. The laboratory operation had been scheduled to work two shifts six days a week. Human resources, estimated cycle time and equipment calculations had been set accordingly and input to ProcessModel. It seemed to the Readiness Team that this was good enough based on the static calculations performed in Excel. However, after a thorough analysis of both capacity calculation tools the reason for the discrepancy was apparent. The static Excel calculations assumed a fixed daily average arrival rate of samples to the laboratory coming from the multiple encapsulation machines. The simulation model considered random variability in the arrival pattern of the samples at the laboratory, which is caused by the “unexpected” but common events experienced in any manufacturing process such as random downtimes, delays in schedules, absenteeism, etc, which cause variations in the arrival pattern of samples at the laboratory. “If capacity calculations and resource needs estimates are based on averages you are missing the point since averages just mean that you will not get your average number close to 50% of the time and it proved to be crucial in this analysis invalidating static analyses”.