Systems model introduces stability and predictability to manufacturing

Jason Adams

Once a component manufacturing company has gone through the process of getting product yields to desired levels, the primary issue becomes how to maintain manufacturing stability. This stability is a basic prerequisite for the efficiency, predictability, and agility needed to support the volume production required by a company such as Lightwave Microsystems (San Jose, CA), which makes planar lightwave circuits and integrated devices for WDM systems.

The system being built at Lightwave blends three application components-manufacturing execution (MES), statistical process control (SPC), and product data management (PDM)-into a self-regulating manufacturing infrastructure that is intended to keep manufacturing processes in their "sweet spot," while maintaining consistency of line yields in the face of change. Manufacturing consistency is critical-without it, there is no foundation on which to build the kinds of customer-responsive scheduling and supply-chain management systems necessary to remain competitive.

The foundation

The execution system serves as the system of record and control for all plant floor operations. It tracks product movement through each production step, collecting quality and parametric data in real time. The addition of SPC ensures that deviations from processing norms are detected and corrected quickly. Finally, PDM is brought into play to maintain stability in the face of change (see Fig. 1).

The role and purpose of SPC in manufacturing is fairly obvious-maintain an overview of process and product metrics and invoke the execution of out-of-control action plans (OCAPs) when out-of-spec conditions are detected. Statistical process control, when integrated with the MES, eliminates the effort involved in tracking process parameters manually, removes human error from the SPC process, and allows response to variant conditions on a real-time basis.

Not so obvious, however, is the role of PDM in this context. The job of PDM, whether implemented as a manual or an automated system, is to keep line production stable whenever changes are introduced into the manufacturing process.

Management of change

Changes always occur in manufacturing, and are either product-related, process-related, or a combination of the two. Product-related changes take place when new products are introduced or existing products are modified. Process-related changes occur when processes are modified to improve yields, reduce cycle times, or reduce cost. Product-related changes naturally induce changes to processes. Process-related changes normally do not introduce changes to the product unless such change is intentional.

The reality is, however, that all changes to production, whether process-related or product-related, have the potential for creating mis-processing-resulting in costly drops in line yield, increased levels of scrap, and other erosion of full production capability. This potential arises because changes inevitably cause confusion unless such changes are managed effectively. That`s where PDM comes in.

The goal of PDM/MES integration is to realize no increase in line loss as the result of change. To make it work, all configuration changes must be smoothly integrated into manufacturing to minimize or eliminate disruptions. Baselines must be established and monitors put in place to track line yield. Deviations from full production potential can then be monitored to detect occurrences of miscommunication between the agents of change and the MES-driven production lines.

Implementation

At Lightwave, the integrated system is a work in progress. The MES has been installed in all three production areas of the company-fabrication, dice and test, and assembly and packaging; a new, integrated online SPC module, provided as an optional module of the MES, is being deployed as well. Once the MES/SPC implementation is complete, the manual PDM system will be replaced with a commercially available product, and then an automated PDM will be integrated with the MES/SPC infrastructure.

The advantages of an automated PDM system, when used in conjunction with integrated MES/SPC, are substantial. Manual change- management processes are fraught with communication errors, which can have negative effects on line production as changes are introduced.

An automated PDM system, on the other hand, offers a new front-end business process for the disciplined management of change. This process ensures that the MES version control-which allows new process and data definitions to be built up within the execution system before being released to production-is being fed with accurate engineering-change-notice information, and in a timely way.

Another benefit of PDM/MES/SPC integration is that the PDM system can manage change to all aspects of the manufacturing system, from MES-defined processes to the charts that control the operation of the online SPC. Statistical process control charts are configurable objects and can be managed by the PDM: when a change is to be put into production, the SPC charts corresponding to the change are released to production at the same time as the manufacturing process steps. In short, a stable manufacturing environment requires disciplined PDM to maintain production line yields, SPC to keep product yields stable, and the MES to tie it all together as system of record and production manager (see Fig. 2).

Statistical process control

Online SPC is a powerful multidimensional tool for controlling and improving manufacturing predictability. It can be used to monitor production on a process-by-process, pure process, manufacturing-tool-performance-relative-to-process, or manufacturing-tool-by-itself basis. It can also be used quite effectively as an overseer of preventive maintenance activities, ensuring that manufacturing equipment is serviced properly and that sophisticated equipment performs within specified limits before being put back into production (see "PM: The hidden process in manufacturing," this page).

Wherever and however it is used, online SPC is driven by user-defined charts, fed by data collected by the execution system, and looks for deviations within product-specification, control, and pattern limits. When such violations are detected, the logic behind the charts triggers OCAP activities that are stored within the MES. These OCAP scripts are user-defined, stored within the MES database, and consist of combinations of automated and operator-driven actions that, when executed, are intended to bring the violating process back within acceptable operating limits.

Implementation of an online SPC system requires the creation of a carefully crafted roadmap. During the implementation it is determined what to measure, how to measure it, how to control it, and how to establish OCAP scripts. At Lightwave, a phased, process-by-process approach is being used in which care is taken to mimic the manual SPC controls and OCAPs originally in place. This mimicking minimizes alterations to existing procedures during the phase-in stages, allowing the transition to online SPC to be done more quickly and with less operator retraining.

The persistence of change

Integrated PDM/MES/SPC provides the foundation required for operational stability, and supports the addition of automated scheduling software once acceptable levels of yield and cycle-time stability have been achieved. Such automated scheduling systems work with plant capacities to determine and then drive the product mix and process flows required to best satisfy demand.

Cycle times and yields must be constant in order for these scheduling systems to operate properly. In the past, much of the high failure rate of automated scheduling system implementations-particularly in complex manufacturing environments-could be attributed to the lack of stability in the supporting system infrastructure. With the application toolset now available to manufacturers, it is possible to create self-stabilizing systems to support automated scheduling, with all the attendant benefits that such scheduling can deliver, such as agility, flexibility, make-to-order capability, and shortened cycle times.

Automated scheduling itself constantly introduces change into the manufacturing process. An example of this is when a manufactured product, starting out as capable of being Product A or Product B, is made to go the A or B route at some midpoint in the process based on considerations of demand and flow conditions. In such cases, the scheduling system, redirecting the flow of product and using the MES as an intermediary, dynamically changes the processing characteristics. Such change needs to impact the production line as little as possible, providing another reason for tight PDM/MES/SPC integration.

Once the desired levels of predictability have been attained and automated scheduling has been implemented, then the next step is to examine ways to improve manufacturing. Continuous process improvement also introduces change on an ongoing basis for purposes such as increasing yield, reducing cycle times, improving quality, and reducing costs. Again, tight PDM/MES/SPC integration is required to prevent such change from causing large swings in line production, while allowing for continuous refinement to higher levels of operating efficiency (see Fig. 3).

All improvements to the manufacturing process will cause change. If properly managed, this change can then be disseminated throughout the enterprise, and then outward to the customer and vendor sides of the supply chain. Maintaining focus on manufacturing stability sets the standard for establishing predictable manufacturing operations so that commitments of both current and future stock availability can be made with high degrees of accuracy. This ability to provide available-to-promise and capable-to-promise information to customers is a basic requirement for new-economy manufacturing and supply-chain collaboration.

Because of the newness of its base technologies, the WDM component industry is blessed with the absence of a substantial manufacturing-software legacy. This fact, coupled with the availability of commercial, compliant, state-of-the-art manufacturing-systems software, makes it possible for companies to build ideal systems far more easily and economically than previously had been possible. Many component manufacturers are attempting to accomplish in months what took decades for their predecessors in semiconductors and electronics.

FIGURE 3. Changes generated by continuous process improvement and dynamic scheduling are propagated downward through the application stack.

PM: The hidden process in manufacturing

The in-house maintenance of complex manufacturing equipment (tools) is a process just like the other processes that take place in a manufacturing environment, even though it consists of work on internally used tools rather than on manufactured products to be sold. As such, preventive maintenance (PM) can be managed by MES, and kept within specifications by online SPC. This often overlooked part of the production process can yield major returns if it is monitored and executed with the same precision that is applied to the manufacturing of company product.

At Lightwave, equipment maintenance management extensions support full MES/SPC control over the PM process. This allows equipment service to be performed under the control of the MES, with the SPC module catching any out-of-spec conditions during the PM process.

Manufacturing execution and statistical process control over PM increases the predictability of the maintenance process so that the machines being put through PM will work better once they are put back into production. The benefits include the following:

- Machines operating more closely within normal specifications are likely to produce higher yields.

- Tracking and analysis of a tool`s performance history can recognize downward trending and allow for correction before the tool fails in production.

- Fewer failures in production mean more equipment uptime and less disruption on the plant floor, thereby increasing yields.

- MES/SPC management of the preventive maintenance process leads to shortened PM cycle time, resulting in more equipment uptime.

- Equipment parameters recorded during PM can be correlated to other measures of performance, which can lead to further process improvement.

With the PM process in control, and all equipment baselined and operating within specifications, equipment efficiency is maximized, supporting higher levels of product yield. When PM is not managed properly, it can produce more line, product, process, and yield hits than anything else, particularly on the fabrication side of manufacturing where equipment qualification and calibration are so critical.