Print this article

Comment

1 April 2002

MES: Lower risk (than ERP), higher reward

by Charles Gifford

Beware of oversimplification by vendors.

Burned by promises gone sour on how investments in enterprisewide information technology (IT) networks would bring a rapid return on investment (ROI), manufacturing companies today are turning instead to plant-side IT projects that seem certain to lower production costs.

Simply stated, IT initiatives that focus on the factory floor today are being viewed as a lower risk than open-ended, enterprise IT transformations. It's about time. A plant optimization project typically produces a <12-month ROI, which is also easier to quantify and verify to real plant problems.

While many enterprise projects were oversold during the past five years, plant-side IT solutions, also called manufacturing execution systems (MESs), have had their own shortcomings. To be successful with MES, an end user must first get past oversimplification by many industrial automation and enterprise resource planning (ERP) vendors.

The MES challenge can be high risk if not properly planned. Done well, however, it can result in a high reward. The goal is to optimize manufacturing while staying flexible enough to easily incorporate, and take advantage of, new manufacturing technologies.

Although MES technology has matured significantly during the past 10 years, it is still nontrivial and is not a silver bullet in any form. It is a process, not an event, and it requires organizational change and support. This is a challenge. Long-term success requires transforming the manufacturing environment.

MES not understood

During its early industrial IT evolution, MES software was misunderstood and ill defined. This contributed to a high failure rate and slow acceptance. To reduce this fragmentation, standards organizations-including the American National Standards Institute (ANSI), ISA, the Open Applications Group, OLE for Process Control, the Institute of Printer Circuit, and the Electronics Industries Association-began work in the early 1990s to build terminology and reference models for the field of computer-integrated manufacturing (CIM).

ISA standards committees, for example, created what are now the ANSI/ISA-95.00.01-2000, Enterprise-Control System Integration and ANSI/ISA-88.01-1995, Batch Control standards sets.

Standards have matured into object models and extensible markup language schema, which vendor and user communities are now adopting. "Best practices" have matured to become the Microsoft Solution Framework and Universal Modeling Language, among others.

The Manufacturing Execution Systems Association (now MESA International), formed in 1992, helps educate businesses in MES transformation. Properly installed and maintained, these intelligent systems continually show dramatic results.

According to MESA International, industry averages show that for high-volume manufacturing, an MES will yield the following results:

• Reduce work in process by 20%
• Reduce plant efficiency/downtime by 30%
• Reduce manual data entry time by 75%
• Reduce waste by 15%

Many MES challenges are interrelated and weighted differently, based on individual circumstances. Challenges are typically 80% cultural and 20% technical.

Challenge 1: Aligning the plant and enterprise on a win-win approach.

A primary challenge of MES is to define a project that has simultaneous buy in on both the plant and the enterprise sides. The best MES application will fail if plant personnel reject its introduction and the automation transformation-or if the enterprise side does not understand or support the plant transformation.

A lesson learned the HARD WAY: You cannot force IT solutions into manufacturing plants. A Phase 1 MES project needs to solve a real plant-side problem while also fixing some enterprise data requirement in the area of capacity, capability, or order visibility for planning and scheduling applications.

The best applications usually start with a plant-side point solution that solves an immediate need such as lowering cycle time, waste, and rework or increasing yield and throughput.

Challenge 2: MES differs from vertical to vertical.

Unfortunately, the term MES is too broadly defined and simplified by analysts and standards. MES software varies significantly for each vertical industry and product line, depending on type of process, organization, and lean manufacturing practices. Batch, discrete, continuous, and hybrid MESs and their approaches differ.

For instance, in discrete electronic assembly, the solution set centers on the product genealogy of the work order, where a user can verify and archive a bill of material, route, work instructions, and quality specs. A different form is required in a high-speed consumer packages goods/food process, with the solution set aiming to improve efficiency through using equipment better, tracking waste, and reducing downtime. The MES form is supposed to control the basic intrinsic pains of each manufacturing environment.

Challenge 3: The maintenance paradox.

MES is a dynamic IT solution, where change is continually driven by product, process, and scaling of new and old products. Database applications run off a data model derived from process requirements and transaction loads. When a process or product changes, the data model has to be "migrated," which involves normalizing table structures and relationships and adding new data loads. If a user appends tables onto the old model, process performance will degrade quickly to a point where the line will slow or will not stop. Then the MES application will create bottlenecks.

Typically, applications change 30% to 50% in a five-year period. MES systems also run 7/24 across networks with multiple system and equipment interfaces that need maintenance. Bottom line, companies must have manufacturing IT specialists who have a working knowledge of process control, IT, and business system terminologies.

These professionals are hard to find. Solving the maintenance paradox requires a systematic approach to integration by establishing common language, business processes, and protocols vertically across plant and enterprise applications. To reduce the level of required IT skill set for plant-side MES, MES is breaking into two interdependent layers: a centralized Web-based layer and a plant-side optimization layer.

The centralized Web-based layer consists of finite capacity scheduling and work-in-process/ work order management elements, where no equipment or system interfaces are necessary.

Plant side consists of plant efficiency (equipment utilization, downtime, waste tracking), statistical process control, product quality, compliance, and lot/batch genealogy functions. Theses functions typically require equipment interfaces of various types.

This may assist MES to have clear lines to aggregate data through vertical and horizontal processes. As MES software moves to the Web-based architectures of Microsoft.NET and the Java J2EE standard, centralized MES will become more frequent. However, IT support will always be required on the plant side.

Challenge 4: Establishing a common language.

MES integration is not trivial. Period. The problem lies in plant and enterprise applications having differing data sets. Enterprise and MES applications require common metrics for defining standard plant-side processes to effectively process the business changes from new product and process introductions. This is part of the maintenance paradox.

Change management is the key to a successful MES life cycle, low cost of ownership, and quick ROIs. Plant-side systems deal in a language based on processes and products, which in turn break down by the MES into plant capacity and capability terms for ERP and supply chain planning and scheduling processes. So as a company optimizes and streamlines manufacturing, its corporate IT systems and business processes are more directly affected by plant-side change. To simplify interfacing, companies must agree on numerous technical details across their many plant and corporate legacy systems:

• Common networking protocols
• Common data formats
• Common event and notification models
• Common device and application interfaces
• Common application integration methodologies
• Common business-to-business message formats

The resolution to these issues would be easy if every plant used the same operating system and the same middleware. However, multiple plants typically use multiple operating systems and multiple component architectures. Fortunately, common language has finally evolved for vertical corporate and horizontal plant integration.

A set of industry standards for CIM now adequately addresses vertical data aggregation. Standards evolved from reference models into object models and now into extensible markup language (XML) schema and data models. This metamorphosis dramatically simplifies the design, maintenance, and change management of applications and interfaces. Consequently, life-cycle costs reduce, and communication is established.

Challenge 5: Establishing a common protocol.

With the new generation of application-to-framework (A2F) XML-based enterprise application interface (EAI) software, IT infrastructures can now reduce the cost of enterprise and plant interfaces by 50%. As the technology matures, the MES's cost and maintenance will go down.

Many current integration practices use application-to-application (A2A) interfaces, which use many different protocols and languages across an enterprise. As applications change, users have to keep up on the maintenance. Many companies have installed A2A EAI software to establish an interface inventory and mapping point-to-point integration. With an A2A EAI, relevant data between any two applications becomes complex when multiple applications from inside or outside the plant and enterprise come into play.

In A2F EAI software, information moves independently among many types of systems and applications, while the framework acts as interpreter and guide. In the A2F approach, companies incrementally implement new functionality. This allows the integration framework to grow and change as business dictates, with a major benefit being that it allows legacy systems to transition in a much more cost-effective manner.

Another major benefit of A2F is that it abstracts plant and business processes from application-based limitations and puts data modeling first. Companies can therefore make the best decisions for growing revenue and expanding markets without regard to the limitations of individual software modules.

Challenge 6: Effective change management.

To effectively process the many changes MES addresses, a dedicated change management system (CMS) and steering committee needs to be incorporated into the organization. The MES steering committee should make "global" decisions and meet on a regular basis to prioritize change activities.

A typical steering committee divides into two teams: a business team and a technical team. The business team, not the technical team, owns the MES and the change management process. The steering committee ensures that "legacy systems" are not a barrier to progress. A fortress paradigm is not acceptable. The committee establishes and enforces common language and process standards throughout the plant. Committee members use a CMS to log all issues, problems, and changes.

The steering committee is responsible for prioritizing and quantifying change entries and assigning them to an accountable technical team member. Maintenance and use of this one single tool controls "scope creep" and application modification/release. The CMS contains a version control system and the production release procedures that encompass the control of the MES application, databases, graphical user interface (GUI) forms, GUI executables, software versions, data link libraries, etc.

Challenge 7: Correctly assessing the pain.

For a company beginning the automation odyssey, the hardest tasks are identifying the most inefficient processes for a Phase 1 MES implementation. Typically, plant personnel know where their bottlenecks and waste are but only in an arbitrary, qualitative way. There is no hard data to justify any expenditures. In order to justify the MES project and determine the ROI, a manufacturing assessment quantitatively analyzes, benchmarks, and prioritizes the scope of the MES initiative.

One word of caution. If circumstances allow, start small. Phase 1 and 2 projects should introduce only two or three areas of automation, with little or no enterprise integration. The compromise to consider is that small projects do provide a comprehensive data model for the completed integrated system. Consequently, it may be wise to select software that includes necessary functionality for Phases 3+. Basically, if economics allow, visualize the completed system when conducting the manufacturing assessment and design requirements.

These early projects allow a company to develop the internal framework necessary to handle and maintain full-blown integrated projects.

Challenge 8: Quantifying and profiling the transaction load and data model.

Once the manufacturing assessment is complete, all have to solidify and set requirements prior to selecting software. Software selection results in the data model set down in the analysis of the database transaction and data load. The analysis outlines worst-case size, number, and type of transactions that occur simultaneously during a production run so the MES does not slow down production cycle times.

Typically, bandwidth and response problems come from a large set of parametric data, test algorithms, work instructions, and overloaded queries on a single client. Once you characterize data model and transaction profiles, select the software technology architecture that best fits those profiles. Many users skip the data modeling process and purchase software based on apparent features. This may lead to a very abnormal and sluggish database.

Within 18 months, many of these end users are redoing the application or pulling the plug. Many of the MES-like modules contained in centralized ERPs have failed due to oversimplifying the production data model and underestimating the transaction load. The value of data obtained from the MES-like module is limited with respect to productivity improvements.

Challenge 9: Overcoming the OOB silver bullet mentality.

In general, people are brainwashed by the "plug and play" or "out-of-box" (OOB) marketing approach. MES and ERP software vendors sell to this weakness with rapid deployment and OOB features. Their products are much improved, but there are no shortcuts in the transformation process. Most rapid deployments fail within 18 months, and then you have to redo them. In the past, MES products consisted of a database transaction engine with limited configuration tools for designing and maintaining the data model, which has 80% custom/20% OOB set of business rules or modules.

As object-oriented and Web technologies mature, many MES products are building a very rich set of industry-specific OOB modules to increase the configurable portion of the solution to 50%, in many cases. It is still 50% custom, with the other 50% requiring custom interfaces among ERP, supply chain, or plant systems. The process and data flows of production must map into a set of business rules and a corresponding data model before you select any software or write any code. Otherwise, you are wasting time and money.

Challenge 10: Don't forget transformation training.

Transformation training of operator and line support personnel is crucial! Approval and acceptance of automation by plant personnel determines the success or failure. If the "best" solution deploys into a plant and the operators and line managers reject it, or if it initially reduces throughput by more than 20% for more than a week, the solution will go away.

The old manufacturing environment requires mapping so operators, line manager, production scheduler, process engineers, etc., understand their new standard operating procedures and deliverables and how the old procedures and deliverables are going to be handled automatically and not compromised. As well, operators need to understand that MES data collection is not going to add work to their jobs.

The challenges of MES are similar to most challenges associated with any technology renaissance. The basic point is to get educated and involved because all MES technologies have matured into a key enabling piece for plant optimization and corporate cost savings. IC

About the author

Charles H. "Charlie" Gifford is director of business development for ASECO Integrated Systems Ltd. He is chairman of Industrial Computing's Editorial Advisory Board and was director of ISA's Computer Technology Division (1996-1998). His e-mail is charlie.gifford@aseco.net.