Just-in-time Lot Dispatching for Semiconductor Tool Utilization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, existing systems face inefficiencies in assigning and managing lots of workpieces between tools, leading to potential idle times and reduced throughput due to immediate assignment to the next available tool without considering priority changes or delays, resulting in long waiting times for higher-priority lots.
Innovation Solution
Implementing a just-in-time dispatching and pickup system that uses predictive models to determine the expected dispatch time for lots, allowing for dynamic assignment and reordering based on tool availability, transport times, and changing priorities, with data mining to adapt to real-time conditions and optimize lot flow through buffer and bulk storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If lots are assigned to tools immediately when tools become available, then tool utilization is improved, but higher-priority lots experience long waiting times
Solution Approach 1:
The system performs preliminary actions by determining expected dispatch times in advance and using predictive models to forecast tool availability. Lots are held in buffer storage and only dispatched when the predicted dispatch time aligns with actual tool availability, preventing premature assignment and ensuring high-priority lots are processed promptly without excessive waiting.
Solution Approach 2:
The system dynamically adjusts lot assignment based on changing priorities and real-time conditions. The predictive models are continuously updated with actual dispatch times, allowing the system to adapt its dispatching strategy. This dynamic approach enables the system to optimize both tool utilization and waiting times by flexibly responding to changing manufacturing priorities.
2Loss of time
If predictive models are used to determine expected dispatch times, then waiting times for high-priority lots are reduced, but system complexity increases
Solution Approach 1:
The system implements feedback mechanisms where actual dispatch times are fed back into the predictive models for continuous refinement. This feedback loop allows the models to learn from real system behavior and improve their predictions over time. The feedback-based approach reduces waiting times for high-priority lots while managing system complexity through iterative optimization rather than complex real-time calculations.
Solution Approach 2:
The predictive models serve themselves by automatically updating with actual dispatch data without requiring manual intervention. The system self-optimizes by comparing predicted versus actual dispatch times and adjusting its predictions accordingly. This self-service capability reduces the need for complex external control mechanisms while effectively reducing waiting times for high-priority lots.
Data Source
AI summary
A method includes processing each of a plurality of lots with at least one first equipment and moving some of the plurality of lots to a first storage. For each of a plurality of second equipments, an expected dispatch time of one or more next lots for processing by the second equipment is determined. Each of the lots in the first storage is assigned to one of the plurality of second equipments on the basis of at least the determined expected dispatch times and moved to one of a plurality of second storages that is associated with one of the plurality of second equipments to which the respective lot was assigned. For each of the plurality of second equipments, each of the lots in the second storage associated with the second equipment is moved to the second equipment and are processed with the second equipment.