GlaxoSmithKline (GSK) was the world’s sixth largest pharmaceutical company in 2014. The company was launching a new vaccine product on a new market that needed a distribution network different from what they had before. Therefore, the company needed to design a new supply chain and align manufacturing processes with it.
The supply chain of a vaccine is a complex system because it is geographically extended globally, and it uses a large amount of resources, including manufacturing plants and warehouses.
Production and distribution start with the preparation of “bulk”, where many different components are used. This stage is then followed by the “filling” step, where many containers, such as syringes and vials, have to be managed. Then the product goes to final packaging (with specific requirements for all different regions, in compliance with local laws). In addition, in the GSK case, the vaccine could only be produced and released for distribution in batches, which made further supply chain planning even more difficult.
A specific issue for the company was how to handle large amounts of components with specific expiration dates, while complying with quality control procedures (both internal and external at “single dose level”), and commercial rules. In addition, GSK implemented various inventory management policies using corporate transactional systems, including SAP. Above all, their supply chain faced a large amount of specific settings and balancing problems.
All of these special traits were the reason that GSK choose dynamic simulation as a decision-support tool for the vaccine supply chain design. Fair Dynamics Consulting developed the supply chain’s simulation model for the company using AnyLogic.
The model designed by Fair Dynamics simulated GSK’s supply chain, including the manufacturing and the distribution parts of it.
The manufacturing part was based on the discrete-event modeling method and it simulated business processes involved in the vaccine production. These included three levels of processes that could be intervened by each other:
- Manufacturing processes
- Quality control processes (product testing)
- Quality assurance processes (production process control)
The model had to take into account the different policies the company implemented and production constraints. For example, the model included shifts of human operators at the plant, their operation times depending on experience, as well as different sourcing policies that affected the production schedule.
The manufacturing part of the model was integrated with the distribution part that simulated the US market supply chain, orders to the manufacturing part, and received goods from it.
The supply chain in the model reflected the real-world supply chain design, and included warehouses replenished from the main European distribution center, wholesalers (product distributors), and hundreds of clients, all with defined geographical locations. Clients ordered goods from wholesalers, but were supplied directly from GSK’s warehouses, due to the sensitive nature of the product and to avoid delays caused by intermediaries’ participation.
One of the important metrics taken into account was maintaining high service levels, because the product’s selling proposition was guaranteed 24 hour delivery.
The model was able to simulate both the steady state of the supply chain, and situations with disruptions and emergencies, to see how “what-if” scenarios would affect supply chain performance.
The model allowed GlaxoSmithKline to determine the optimal vaccine supply chain design from the standpoint of costs and service level.
The model also served as a part of an operational decision-support tool for the supply chain planners. The tool allowed them to determine optimal production/distribution policies for the next week/month.
The uniqueness of this project was the combination of manufacturing and distribution processes in one simulation model, which was possible due to the multimethod modeling techniques available in AnyLogic. This approach gave GSK the ability to achieve more accuracy in simulation, thus allowing for more precise forecasting and more profitable decision making.