Smart Production Scheduling
of Wafer Fabrication
“Fabrication of integrated circuits (ICs) on silicon ‘wafer’ discs is argued the most complex manufacturing process in existence”
Mönch, L. (2013). Production Planning and Control for Semiconductor Wafer Fabrication Facilities.
Introducing Flexciton FabPro.
The 21st-century solution to optimising wafer fab scheduling.
Off the back of 10 years of academic research of operations, Flexciton engineers have developed a solution that can globally optimise the schedule of all wafers in a wafer foundry using state-of-the-art A.I.
Built in collaboration with Seagate Technologies in the UK:
Flexciton combines cutting edge Mixed-Integer Optimisation techniques with A.I. to solve what was previously an unsolvable scheduling problem - global wafer fab production scheduling.
Increase in wafer throughput and machine utilisation
demonstrated through early trials at Seagate Technologies. To learn more about about our solution and why this is an industry breakthrough, book a free consultation and software demo.
Wafer Fabs and Foundries have the hardest scheduling problem in manufacturing.
Each wafer starts it’s journey as a blank silicon disk. To become a finished wafer, it must endure a long, complex and rigorous manufacturing process.
700+ PROCESS STEPS
100+ MACHINES & TOOLS
Visualisation of a wafer journey through its fabrication process:
For decades, semiconductor manufacturers and foundries have been searching for a production scheduling solution that can globally schedule a factory.
Due to the huge complexity and mathematical intensity of the wafer scheduling problem, traditional scheduling tools do not work well.
Instead, dispatch tools are increasingly being used to control the local scheduling decisions for many modern facilities.
However, dispatch systems have limitations and are not a suitable fit for global scheduling and optimisation:
Use basic 'rules' or 'decision tree' logic to prioritise wafer moves for
the current moment in time.
Can only make local decisions, thus cannot consider consequences outside of local area/zone.
Dispatch is unable to predict dynamic bottlenecks.
How Flexciton A.I. works
Data science and mathematical optimisation is used to simulate millions of schedules at once, selecting the most optimal every 15 minutes and feeding back to your dispatch system.
The optimal schedule will direct wafer traffic around dynamic bottlenecks and can streamline production to increase throughput. The model incorporates all constraints such as Kanbans, Batch Sizes, Reticle Availability and Maintenance.
Key Benefits of Flexciton FabPro:
Improved capacity utilisation
Dynamic bottlenecks avoidance
Flexciton FabPro has been built in collaboration with Seagate Technologies in the UK, with proof-of-concept results showing a throughput increase of 7-10% over current dispatch methods.
For more information about Flexciton A.I. scheduling technology click below.