Improve throughput by 5-10% using SmartFactory dispatching solutions

Increase productivity by another 3-5% by integrating SmartFactory scheduling and dispatching solutions

By Madhav Kidambi
Improve Throughput by 5-10 Percent Using SmartFactory Dispatching Solutions
Semiconductor factories often use the terms “scheduling” and “dispatching” loosely. In this blog, I define these terms, with greater emphasis on context, outline common misconceptions and challenges associated with these techniques, and highlight the value of deploying an integrated approach with scheduling and dispatching.

Scheduling: Lot Assignment

Scheduling in this blog refers to assigning lots to a given set of equipment in a specific area. Assigning lots is based on current and predicted lot arrivals, as well as current and future equipment conditions. Manufacturers typically create schedules for a 12-hour shift and generate them on 10–15-minute intervals. Typically, scheduling solutions are implemented for bottleneck tools or areas with complex processing requirements. Some examples include litho, wets/diffusion, chamber-based tools, testers, and so forth. The scheduling solution can be based on optimization or heuristics methods. Both these options enable better throughput. The optimization-based methods are known to provide an additional 3–5% improvement in throughput over heuristic methods. For details on this improvement, refer to Michael Förster of Infineon Technologies describe how they optimized their productivity, using SmartFactory integrated solutions.

Dispatching: Lot Sequence

Dispatching in this blog refers to the sequence of lots to process at specific equipment in real-time. Dispatching rules are implemented factory wide and are typically implemented as global and local rules. The global rule includes line balance logic that enables managing customer commits, and local rules that contain additional logic to optimize the throughput for a given area—like Litho.

Misconceptions

A common misconception is that dispatching rules don’t comprehend the upstream and downstream WIP/equipment conditions while creating a sequence of lots processed for specific equipment. Our SmartFactory dispatching solutions incorporate upstream and downstream WIP conditions, as well as current and future factory states to create dispatching lists in real-time. Based on the current capabilities, our dispatching solution can be considered as “real-time scheduling” and the scheduling solution as “near real-time scheduling.” Table 1 summarizes the differences between planning, scheduling and dispatching solutions.
Final Figure
Table 1: Planning, Scheduling and Dispatching Solution Scope

One more common misconception is that you only need a scheduling solution and not dispatching. Typically, schedules are generated every 5–15 minutes in a factory. During this time the factory floor encounters many changes. Equipment states might change, a lot might be put on hold or a process on qualified tools might become restricted. In their paper on near real-time scheduling and dispatching, Govind et al. present data on how often an area state changes1. Based on this data 50% of the changes occur in <5 minutes and 80% occur in <15 minutes. Due to this change a published schedule creates unclear and inconsistent decisions on the floor. Some of the challenges are:

  • Manual Adjustments. An operator working from a 5 to 15-minute old schedule attempts to select a lot for the tool, but the lot is now on hold. The next lot on the screen is scheduled for a tool that just went down. In this case the operator must now adjust manually. This adjustment creates more idle equipment with WIP on hand, impacting throughput and cycle time.
  • Schedule Responsiveness. Published schedules are unable to react in real-time for critical yield excursion recovery. In this case engineering needs to immediately restrict certain tools and funnel work to a golden tool “where possible.”
  • Automated Transportation. In factories where transportation is automated using an overhead track (OHT) system or automated guided vehicles (AGVs) or other types or robots, they often fail to deliver lots to the correct equipment, resulting in numerous exception handling scenarios and creating white space on equipment.
To overcome these challenges an integrated dispatching solution is required. In areas where manufacturers implement scheduling, dispatching rules try to follow scheduling results and due to some of the challenges mentioned previously, it adjusts the scheduling results based on real-time factory status. Figure 1 shows how a SmartFactory integrated and dispatching solution is implemented.
Final Table1
Figure 1: Executing Integrated Dispatching and Scheduling Solution

Deploying an integrated scheduling and dispatching solution ensures that:

  • Real-time adjustments to dynamic events in a factory occur with a consistent, heuristic-based approach.
  • Rules incorporate both global and local parameters at a fast logic execution time (in seconds) for the lowest hanging fruit in productivity.
  • Scheduling in highly constrained areas provide additional value when rules become too complex to provide an optimum path.
  • Each domain is used to converge consistently and tightly with direct links toward factory objectives and key metrics.

Customers using our SmartFactory dispatching solutions improve throughput by 5-10% and increase productivity by another 3-5% by integrating these solutions.[1]

REFERENCE

[1] Operations Management in Automated Semiconductor Manufacturing with Integrated Targeting, Near Real-Time Scheduling, and Dispatching, Nirmal Govind, Eric W. Bullock, Linling He, Bala Iyer, Murali Krishna, and Charles S. Lockwood, IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 21, NO. 3, AUGUST 2008

Share / Subscribe