- In Customer Stories
- Posted 08/01/2003
AstraZeneca uses WITNESS simulation software from Lanner to optimize supply chain performance
For AstraZeneca, the ability to bring new drugs to market quickly and to meet expected demand from day one will be critical to the future profitability of the company. One of their new products, an important class-leading drug in its field, which for many reasons must remain nameless, was anticipated to be an international bestseller and the company was gearing up for its launch in 2002. The performance of the drug’s supply chain was critical to enabling AstraZeneca to achieve its launch targets and satisfy long-term demand for the new drug. Bulk manufacture of the drug involved a multi-stage process carried out at brand new state-of-the-art manufacturing facilities. Lanner was commissioned to undertake a detailed analysis of the manufacturing site to underpin supply chain performance.
Lanner consultants carried out a structured program to optimise the performance of the site. The goals were to:
- Ensure timely movement of material at each stage of the manufacturing process through the warehouse.
- Assess whether proposed equipment levels were sufficient to meet throughput.
- Optimise operator numbers, shifts and skill levels.
- Identify “smart” scheduling rules to increase capacity.
By constructing WITNESS computer models of the site, Lanner effectively enabled the plant to be run before it has been built. The work has provided AstraZeneca with insight into plant performance, its capacity and its constraints. Significant changes in the planned operation of the plant have been made as a result, and Lanner has developed an operational “process board” to ensure its smooth operation.
The Internal Supply Chain
The bulk manufacturing process is in three stages. Raw materials are converted into crude, before purification and then milling. Initially, production is at a Technology Transfer Plant (TTP), which is used to test the conversion of small-scale manufacturing trials to large-scale bulk manufacture. Two new production plants are then coming into operation in 2001-2002 to complement and eventually supersede the TTP. Adjacent to the plants is a mill.
Lanner adopted a modular approach to the work, starting with the warehouse and then looking at each of the production facilities. Within each module, a common process was followed:
Static analysis of workload
Process documentation was analysed to develop detailed process maps identifying the main direct and indirect tasks and corresponding cycle times.
A data manager was developed to calculate the workload under the range of throughputs detailed in the five-year product plan. Utilisation of resources and processes was also calculated.
Dynamic analysis of workload
The data manager was linked to a WITNESS simulation model of the process. The model enables the effects of shared resource, variability and interaction to be quantified.
Equipment levels, production scheduling and control rules, operator levels and skill mix are varied in the simulation model to optimise throughput.
Lanner consultants were onsite working alongside production staff. A key feature of the project was the use of “lunch-time progress meetings” to report progress and share understanding.
- Development and testing of a two-footprint kanban system to control movement of batches of material between the warehouse and production plants. This enabled continuation of the five-day week in the warehouse while supporting seven-day working of the manufacturing process.
- Identified production loss that would occur without additional operators prior to go-live equal to millions of dollars of lost production. These operators have been recruited and trained.
- 10% capacity increase through smarter production scheduling rules
- Avoidance of duplication of effort and ensuring the deployment of best practice across the project teams linked to each plant.
- Identified weaknesses in communication, particularly on shift hand-over. A physical production board was developed and located in the control room to overcome these. This is described below.
The long cycle times, number of batches in the system and increased number of operators mean that the plant is more complex than many of AstraZeneca’s bulk plants. In order to realise the theoretical capacity, timely carrying out of operator tasks is vital. The process board was developed to provide an instant view of plant status, including location of batches and the time and nature of the forthcoming tasks.
The process board was developed in conjunction with the process technicians. An initial scoping study was carried out to ensure clear definition of the scope of the board and its purpose. During the project, a number of alternative designs were evaluated by producing a number of mock-ups. The design optimises the trade-offs between ease and speed of use, immediate visual impact and level of detail.
The board is now in continuous use, delivering the expected benefits in terms of shift workload planning, and enabling a smooth hand over between shifts.
A key feature of this project has been the quantitative and visual contribution made by computer simulation. By running the virtual plant, changes to resources, equipment and operating rules have been made to ensure that the design meets the capacity requirements from the start. The visual nature has enabled the sharing of information and best practice between the separate project teams, and ensured rapid buy-in to proposed changes. The improvements have been embedded into a practical process board for ongoing management of production.