Production Planning Case Study Summary
A global semiconductor manufacturer with fabrication facilities across three continents faced the complex challenge of managing semiconductor production lines characterized by long lead times, interplant dependencies, frequent equipment breakdowns, and highly fluctuating demand patterns.
The company struggled with traditional planning methods that relied on historical fixed lead-times and pre-defined bottlenecks, resulting in increased work-in-process inventory, extended cycle times, and poor delivery performance. To address these critical operational challenges, the company implemented Eyelit Technologies’ integrated planning and scheduling system that connected with their MES and ERP systems to enable real-time capacity reservations, forecast commitments, and smart lot releases.
The results were extraordinary: delivery performance improved to nearly 100%, cycle times and WIP levels were significantly reduced, and planning cycle time was cut from weeks to hours, transforming the company’s ability to respond to customer demands in real-time.
The Challenge
One of the most difficult environments for planning and scheduling is semiconductor fabrication line. Characterized by long lead-times, interplant dependency, high occurrence of breakdowns and scraps, and notably expensive equipment, fabrication lines and back-end processes present a very challenging job-shop environment for planning and scheduling.
In addition, the industry faces highly fluctuating demand changes, variable mix of end-items, and vigorous competition internationally. In this paper, we describe deployment of Eyelit Technologies planning and scheduling system at the Company with semiconductor facilities in 3 continents and describe how Eyelit Technologies is used, in conjunction with their Eyelit Technologies MES solution and ERP systems, in order to make capacity reservations, commit to forecasts, release lots, and form dispatch lists for all equipment.
Planning Process and Order Commitment
Forecasts are received from each product group (customers), as unconstrained demand, on a monthly basis. Each product group is responsible for a number of products and technologies (i.e. family of products). The forecasts, by product, are generated by different product groups. The fabrication lines and sort areas, after examining the capacity and material requirement of the forecast, will make a commitment in terms of number of wafers that can be delivered in future periods. These commitments are made by Technology.
The implication is that, although the wafer quantity by product may change, the wafer quantity by technology may not be exceeded. Different products have different capacity requirements. Therefore, by changing the product mix, there is a chance that the capacity requirements may change from what was originally anticipated (based on the forecast).
After a commitment is made to each product group, by Technology or by Product, then the actual sales orders (work orders) will follow. Work orders arrive a number of times each day with a customer requested due date. The requested quantity is checked against committed quantity and a promised shipping date is then generated based on material and capacity availability. Traditionally, this is done based on the number of wafers and/or units (in the back-end processes).
However, this is not an adequate approach since units of different wafers or devices do not map into the same capacity requirements leading to erroneous commitments. This issue is particularly important where there is a high mix of end-items. There must be a mechanism for translating capacity and number of wafers.
Before Eyelit Technologies applications were deployed, the promised shipping dates were based on historical fixed lead-times. A measure that is insensitive to the product mix and WIP inventory. In addition, the time taken for collecting the right information in order to respond to the requested order was too long. The shop floor control system would then assign priorities to the lots (manufacturing orders) at each resource and try to push the execution of orders that seem to be more urgent than others.
In the meantime, more WIP was generated based on a pre-defined number of starts per day. The number of starts was based on capacity availability of certain bottleneck resources. As more WIP accumulated, the lead times increased causing more late lots (late orders). Pre-defining bottlenecks in fabrication lines is not a very effective approach because the bottlenecks are shifting based on product mix and availability of equipment.
The Strategy
The Company was committed to low cycle times and on-time delivery. A study was initiated in order to understand the major areas of improvement and focus on those activities where 80% gain can be obtained through 20% effort. Smart release of lots into the fabrication lines was considered to be one of the critical issues to be addressed.
Smart release implies starting lots at a point in time so that the wait-time of the lot is minimized and the bottleneck resources are utilized as highly as possible. The key to smart release is to have the ability to look ahead and decide when each lot should be released into the fab lines. The timing of release, if performed properly, will minimize the need for more reactive methods of sequencing and dispatch list generation.
Deployment of Eyelit Technologies
Most of the needed information is made available to Eyelit Technologies through the Eyelit Technologies MES solution system. This includes current WIP, current and expected status of equipment, and routings. With every new order arrival, they are compared against the quantities that are being made including scraps that may have occurred. Based on the difference between the demand and supply of wafers and inventory positions, the new quantity of wafers that need to be produced is decided.
At the same time, the new request is compared against the original commitment to the customer that was made previously. If the new order exceeds the contracted quantity (made at the time of forecast), then business rules may be invoked by the system as to what should be done.
After orders are entered into Eyelit Technologies system, it performs die-to-wafer conversion and lot sizing. The system will then generate a commitment date in real-time based on the current capacity & material availability as well as the original contractual agreements. By relating the actual orders coming in on a daily basis, to the original contracts made to the customers, Eyelit Technologies ensures the consistency of the actual sales orders and the capacity reservations made on a weekly or monthly basis.
This feature makes Eyelit Technologies an ideal real-time Available-to-Promise engine. Eyelit Technologies’ planning logic includes parameters such as availability of capacity, importance of the customer, and raw material (silicon) requirements, substitution rules, grading and binning.
Routing data, equipment maintenance, WIP status as well as shop floor model is downloaded from MES to Eyelit Technologies. Forecasts and customer orders are entered into Eyelit Technologies via the corporate ERP order entry system. Other data items include silicon availability and resource calendar.
System Architecture
WIP and equipment status → Wafer starts and dispatch lists
Order entry → Forecasts
ERP ↔ Eyelit Technologies ↔ Eyelit Technologies MES solution
Fab lines → Probe site
Order commitment
The frequency of data transfer is a function of how often the data is changed. For example, WIP status is read into Eyelit Technologies every time that new orders are entered into Eyelit Technologies. As an extension of integration with MES, the next step is event-driven operation of Eyelit Technologies, where Eyelit Technologies would send dispatch lists (via MES) to each equipment, or work cell, every time that a MVOU or machine breakdown occurs. Note that if the event was predicted by Eyelit Technologies, there is no need for a new dispatch list.
Benefits
Introduction of Eyelit Technologies at the Company has produced a number of unprecedented benefits and results. Through better visibility, Eyelit Technologies is enforcing the required discipline and it is institutionalizing the on-going process of data correction and practices that are defined by the management. Examples of the latter are on-going cycle time reduction and continuous improvement in delivery performance.
Eyelit Technologies has also helped to optimize a number of key parameters including cycle times, work-in-process volume and delivery performance. After a few months of Eyelit Technologies operation, delivery performance was improved drastically close to 100%. Moreover, cycle times and WIP levels were reduced significantly.
Equally important to reducing the manufacturing cycle times was the reduction of planning cycle time. The latter is a critical tool for being more responsive to the customers. Eyelit Technologies has helped to reduce the time from forecasts to commitment from weeks to hours. Furthermore, the time needed for promising a delivery for each work order can now be performed in real-time!
Other intangible benefits include real-time status checking of orders, what-if analysis for business and ramp up purposes, and the ability to perform the planning functions from forecasting and commitment to release of lots and machine sequencing all in ONE environment, viz. Eyelit Technologies. Such an integrated environment has helped the company to avoid the use of individual spreadsheets, hard-to-maintain simulation systems and many other home-made systems which have constituted the traditional planning functions in semiconductor and other industries.