All Eyes on Execution

Once a rarity in the IT infrastructure, manufacturing execution systems (MES) are seeing increasing take-up by plant owners looking for improved shop floor visibility and an effective enterprise link. By Kumar Pisharath.

The 1990s saw the evolution of the manufacturing resource planning (MRP) System into the enterprise resource planning (ERP) system and its adoption by a large number of organizations, especially those having multiple manufacturing companies with facilities scattered around the globe.

Implementing ERP systems, however, has required significant investments and often been the cause of great pain. In addition, only after installation, did many companies realize that a great deal of real-time data input from manufacturing operations was needed in order to be able to make full use of ERP capabilities. Overlooking this requirement resulted in poor return on investment and, often, the necessity to add manual data entry and other processes into manufacturing operations, thereby impacting productivity.

Organizations are gradually discovering that to achieve the benefits of their ERP implementations, they need to obtain accurate and detailed information from the factory floor in real-time. This is the domain of the manufacturing execution system (MES), which analysts cite as a key IT direction in which forward-thinking manufacturers are moving. The main benefits of such a system are to increase visibility of production, allowing the smoothing out of bottlenecks and reduction of waste.

Evolving from the need to have accurate and timely work-in-progress (WIP) information for production planning, scheduling and tracking purposes, the main functions of an MES now typically include:

• Resource Allocation and Status

• Operations Scheduling

• Dispatching Production Units

• Document Control

• Data Collection/Acquisition

• Labor Management

• Quality Management

• Process Management

• Maintenance Management

• Product Tracking & Genealogy

• Performance Analysis & Reporting

Driving demand

A key trend over the last decade is increasing competition forcing manufactures to change production operations from a “make-to-forecast” to a “make-to-order” model. This implies (and requires) closer co-operation with supply-chain partners, an ability to react faster to changes in supply and demand, and major improvements in production tracking.

Since 2001, following the massive rollout of ERP systems, an increasing number of companies have begun to use MES to make the connection between their enterprise systems and manufacturing operations.

MES solutions can be applied to a wide variety of manufacturing operations and are acknowledged to have brought tremendous overall benefits: 45 percent reduction in manufacturing cycle time, a 75 percent reduction in manual data collection processes, a 50 percent reduction in paperwork and on average, a Return on Investment (ROI) in 14 months (source: Manufacturing Enterprise Solutions Association, MESA, www.mesa.org).

Previously, the perception of MES was one of “nice-to-have”, expensive systems that only large semiconductor companies with big budgets could afford. But increasingly, MES systems are a mandatory requirement for semiconductor foundries and back-end manufacturers, as they are unable to secure customer contracts unless they have one. In some industries, it has also become a requirement for ISO and other certifications.

And as competition increases, more and more contract manufacturers are discovering that having an MES system is becoming essential if they wish to remain competitive.

This increasing demand is reflected in industry analyst reports. Frost & Sullivan forecast the MES market size to be around US$3.5 billion by the year 2010, with a CAGR of 20-25 percent. AMR Research (the first to coin the term MES) reported that global MES market revenue reached an estimated $1.06 billion in 2004, a 50 percent increase since 2001, and cited “compliance initiatives” and “ERP maturity” as the two major drivers.

Market dynamics

There are estimated to be around 70 companies worldwide supplying MES software products. These range from multinational corporations (Tier 1) operating in global markets to small companies operating in a specific geographic region (Tier 3). Tier 1 companies include ABB, Aspen Tech, Emerson Process Management, Honeywell, Rockwell Automation, Invensys and Siemens. Tier 3 companies include lesser known names such as Visiprise, Apriso, iBaset etc.

The market has been dominated by Tier 1 vendors. However newer, innovative and more competitive product offerings from Tier 2 and Tier 3 vendors have started to erode their market share. This trend is expected to continue. The solutions offered by Tier 1 vendors also tend to be expensive and unaffordable to an increasing number of manufacturers faced with daunting cost cutting pressures. In recent years, there has been a noticeable trend of Tier 1 vendors acquiring Tier 2 and Tier 3 vendors.

The MES systems market for the discrete industries is immature and vertically fragmented with many small suppliers. Few companies have major market positions in multiple vertical industry segments. Opportunities abound for MES system suppliers as awareness and value propositions become clearer to users, new technologies emerge, products become easier to compare, and justifications become easier to analyze.

As mentioned, MES systems have been mandatory in the semiconductor industry, particularly foundry fabs and assembly and test houses due to the complex nature of the manufacturing process and customer demands for accurate real-time data on their manufacturing orders. In the pharmaceutical and health care industries, MES systems are required to satisfy the requirements of regulatory bodies such as the FDA.

These markets are dominated by big MES suppliers such as Brooks Automation, Applied Materials, Camstar, Honeywell etc. However, these cover only about 20 percent of the manufacturing segment. The remaining 80 percent consist of manufacturers, large and small, who are coming under increasing pressure to implement MES solutions to streamline their manufacturing processes, improve efficiency and cut operational costs.

Seeking a solution

Established in October 2005 in Singapore, Nepes Pte Ltd is majority owned by the Nepes Corporation, which owns the only flip chip wafer bumping foundry in Korea.

“Wafer bumping” is a necessary step for enabling flip chip assembly, which is the direct electrical connection of face-down ("flipped") chips onto substrates by means of conductive bumps on the chip bond pads. In contrast, wire bonding, the older technology which flip chip is replacing, uses face-up chips with a wire connection to each pad. Replacing the wire bonds with bump interconnects reduces the package size and weight. Specifically, wafer bumping is the process of adding the conductive bumps onto the chip bond pads.

Nepes’ Singapore setup started with a bumping capacity of 8,000 wafers a month and is set to progressing to 15,000 wafers. The plant is designed to be flexible for both 8 and 12 inch wafers to meet customers' varying demands. It is Singapore first 12 inch bumping foundry.

Executive management at Nepes realized that MES would be a critical and necessary part of the new operation, as being a semiconductor foundry, customers demand full visibility to the manufacturing process – e.g. periodic up-to-date reports on product quality, process yields, production cycle times, conformance to specifications.

Such information can indeed only be provided by a full-functional MES with advanced tracking and reporting capabilities. While some ERP vendors do provide manufacturing modules, these do not provide the level of granularity required to meet the requirements of a semiconductor foundry business.

In addition to meeting the stringent requirements of its customers, an MES is also indispensable to Nepes, being used to measure and monitor its own production performance and ensure the quality and reliability of the manufacturing processes.

ASophia Asia Systems (ASophia) was selected by Nepes to implement the MES system for the wafer bumping facility. ASophia is a Singapore-based company founded in 2002 following the acquisition of a technology developed over a seven year period by a team of European software engineers. Since then the company has been actively developing and marketing the ASophia MES Software as a solution that maximizes the capabilities of the production process and the competitiveness of the manufacturing enterprise.

Factors cited by Nepes management in the vendor selection decision included ASophia’s willingness to adapt the core product to suit the unique requirements of the business; having its product development team located in Singapore – making easier to request and obtain specific customizations to the MES system; and flexibility in operating system (OS) and database platforms.

“Operating system independence was a big advantage. This meant we had the option to implement on a Windows platform to start with and migrate to the more robust UNIX platform at a later stage if required. No other MES vendor provided this flexibility”, says Jack Chua, in charge of the IT infrastructure architecture at Nepes.

System configuration

The ASophia MES is built on the J2EE platform to make it operating system independent. Porting to other platforms that support the Java Run-time Environment (JRE) is trivial. This is a first in the MES industry. Most other MES systems are platform dependant. Although ASophia is an IBM Business partner, and promotes IBM DB2 as their preferred database platform, the ASophia MES also supports Oracle and MS-SQL databases.

At the Nepes plant in Singapore, the ASophia MES configuration (see diagram) comprises a core MES engine implemented on an IBM xSeries server, with a second server used to form a two-member cluster for redundancy and load balancing functions. The Oracle Database, meanwhile, is installed on two further IBM xSeries servers clustered via Oracle’s Rapid Application Cluster (RAC).

Located on the production floor and office areas are three client applications that access the MES server:

MES Configurator: Mainly used by production and process engineers, this client creates and maintains the site model and process models, including the factory environment, manufacturing resources, product routes, bill of materials (BOM), user authorizations, and other information required to maintain the system.

MES Client: This is a thin client through which lot tracking, data entry, material consumption, defect data, etc, are entered into the system by production operators. It is also used to create, schedule and start lots in the system, and provides lot monitoring, lot hold, lot move and lot display functions.

Operator Station: A simplified and customized version of the MES Client deployed on the production floor with basic functionality required by production operators.

Keeping track

The project was scheduled to be divided into two phases: Phase I – basic MES functionality such as WIP tracking, lot and work order management, production process specification and control, engineering data collection; Phase II – additional functionality such wafer map, sales order management, inventory management, integration with accounting system etc.

As implementation progressed, the foundry found it necessary to rearrange priorities based on customer demands. As a result some Phase II functionality had to be brought forward and implemented in Phase I. Wafer map was one such functionality.

“Wafer maps” are critical to the wafer bumping process, as they keep track of the defects on each wafer as they are received from the customer, processed by the foundry and delivered back to the customer. Most semiconductor companies have been slow to adopt a common standard file format for wafer maps.

As a result, it was necessary to convert these customer specific formats into a common format for internal processing and then convert them back to customer specific format before sending them back to their customers after processing.

Business benefits

The implementation of the ASophia MES system has enabled the company to comply with the stringent and varying requirements of its customers with regards to quality and traceability of the wafer bumping manufacturing process. And the management reveals that the MES has helped in securing valuable contracts from the most demanding of customers.

In addition, it has given the company a better insight into its own manufacturing process, enabling bottlenecks and areas of improvement in manufacturing operations to be identified.

Phase I of the MES project was completed in December 2006, and Phase II is currently nearing completion. Both companies agree that the successful implementation was a result of not just of a solid technical solution but because of the close partnership that developed between ASophia and the wafer bumping foundry.

Kisoo Park, the IT Manager in charge of the MES implementation at the foundry said, “ASophia understood the importance of remaining flexible and adaptable to such changes in scope. This attitude reinforced the belief that my executives had chosen the right partner. I say ‘partner’ because this is what our relationship has evolved into. It is much stronger than a vendor-client relationship.”

S H Ansari, ASophia Solution Architect: “The give-and-take attitude in our relationship with the customer has helped us a great deal in ensuring the success of our implementation. They is willing to listen to our concerns and see things from our perspective.”

Kumar Pisharath is Senior Manager, ASophia Asia Systems (www.asophia.com)

---

Decreasing Downtime

By enabling rapid identification of downtime causes, MES helps to reduce ship unloading time at the world’s third largest alumina refinery.

The Queensland Alumina (QAL) Gladstone alumina refinery in Australia is the third largest in the world, with a maximum annual capacity of 3.8 million tonnes. In 2005, the plant actually produced 3,887,249 tonnes of alumina or 10,650 tonnes per day. With the worldwide demand for resources booming, QAL is striving to maximize the output of alumina.

The reddish pebbles of bauxite are mined at Weipa in North Queensland and shipped 2000 km to be refined using a chemical treatment known as the Bayer Process. Two tonnes of bauxite need to be refined to produce one tonne of alumina. The plant keeps more than 550 megaliters of process solution circulating through tanks, pressure vessels and pipes to achieve the scale of production required. There are more than 200 shipping movements at the wharf annually, involving some 11 million tonnes of imports and exports in bulk carriers.

Bauxite is unloaded from the ship at the wharf and travels via seven conveyors over 2 km to four stockpiles. The average unloading time for a bauxite ship is 32 hours.

With an aim to increase alumina production to meet demand, there was a need to reduce the time taken to unload incoming bauxite at the wharf. Also identified was an opportunity to reduce the need to hire extra ships by reducing the overall vessel time at the wharf.

The Equipment Care Reliability Engineer at QAL stated that, “We needed to analyze downtime across each discharge to determine opportunities for improvement. Our aim was to unload ships ideally in as little as 25 hours, an improvement of over 20 percent.”

Data deficiencies

Existing production reporting systems based on spreadsheets and manual free-form data entry needed to be replaced. Main concerns were problems surrounding data security, accuracy, consistency, and double handling. Reporting was slow and time consuming and it was tedious to compare and analyze downtime data across voyages.

To meet the production reporting requirements, a replacement for the existing manual and spreadsheet systems was required. The system had to cater for analyzing unloading information across voyages, across the four ships in QAL’s own fleet and external vessels hired for this purpose, and also by the two different types of bauxite received from the mine.

There were also some requirements for operational management from the system. Operators needed an indication on how the current voyage unloading was progressing and whether the ship would meet its scheduled departure time. If not, shipping agents needed to be advised of delays. In addition, it was required to know how long unloaders were digging in certain holds so that the ship unloading plan could be followed.

The Ampla MES system from Citect was chosen as the solution to meet the identified requirements. Ampla was already being used on some equipment in the Bayer Process and so it made sense to extend the system to the Raw Materials section.

Ampla is an integrated suite of modules for connecting, transforming and presenting information in real-time, throughout the enterprise, with the aim of revealing opportunities to increase performance, efficiency and profitability.

The implementation process involved integrating Ampla with the conveyor control systems and rolling out training to the plant operators. The first voyage recorded was in September 2005. Since then, 130 voyages have been recorded with 5101 downtime records created.

Revealing information

QAL uses the Six Sigma methodology for business improvement. Six Sigma requires that problems go through a cycle of Define, Measure, Analyze, Improve and Control, and Ampla is being used across the Measure, Analyze and Control phases. for 24 users across the Raw Materials area. Equipment downtime is now logged accurately and automatically. The information is available to be displayed and analyzed in various formats including Pareto Charts, Gantt Charts and Pie Charts.

According to the Area Mechanical Engineer – Raw Materials, “We were able to definitively rank and quantify the causes of downtime. Using Ampla’s Gantt chart view, we were able to show that the largest controllable source of downtime was clustered at the beginning of the unloading process. I consider the Ampla Downtime system to be an extremely effective and efficient tool in any Six Sigma reliability improvement project”

The use of Ampla is now part of the routine with the Top 5 downtimes reviewed weekly. Data is more accurate and the reporting system simplified and secure. Based on the information from Ampla, QAL authorized capital expenditure to eliminate one of the major causes of downtime. There are now plans to expand the use of Ampla to other areas in the plant – coal unloading, alumina loading and other Bayer process equipment.

Based on information from Citect (www.citect.com)

Typical MES architecture

The central position of MES in the manufacturing IT hierarchy

MES architecture at the Nepes semiconductor plant in Singapore

---