SPC/SQC Closes the Loop
Statistical process control/statistical quality control software is closing the loop by working online, identifying and fixing problems before an operator is aware any problem exists.Software for statistical process control and statistical quality control (SPC/SQC)—once used exclusively offline for analysis—now integrates with control and human-machine interface software.
Statistical process control/statistical quality control software is closing the loop by working online, identifying and fixing problems before an operator is aware any problem exists.
Software for statistical process control and statistical quality control (SPC/SQC)—once used exclusively offline for analysis—now integrates with control and human-machine interface software. Online responsibilities include providing alarms or even fixing problems before an operator knows any problems exist.
"Traditionally, manufacturing and quality control functions have been separate," says Ralph Rio, Cimplicity Manager, GE Fanuc Automation (Charlottesville, Va.). "Operators ran the machines that make products and, hours or days later, an inspector looked for flaws in the products. This lag time between production and inspection allowed bad products to get through the process, causing rework, scrap, and/or custom returns. This wreaked havoc on production schedules and costs."
Greater integration means "operators can control and monitor both production and quality processes automatically," he says.
Busy in the background
Ken Dingman, ABB Automation (Rochester, N.Y.) global manager, Information Management Products, agrees. "An SPC package should be able to analyze data and perform the statistical computations in the background. When an 'alarm' condition occurs (such as a run alarm), an event should be generated in the control system so that the operator is aware of it. This will allow for immediate attention" to avoid deviation in product quality.
An example of how SPC charts can prevent errors includes compensating for tool wear on a metal-cutting machine. The SPC software detects when parts grow in size. Before the upper control limit is reached, says Mr. Rio, software identifies the trend, creates an alarm to notify an operator, then before the part gets out of "spec," automatically adjusts the cutting tool's position to compensate for the drift.
Applications extend far beyond machine tools into almost any industry.
Daryl Walther, of Rockwell Software's (West Allis, Wis.) HMI Business Unit, Rockwell Automation, says, "Controlling a process is essential for producing quality products, so virtually all processes can benefit from SPC." Rockwell Software once charged for the SPC module, but views it as "core functionality" for HMI so includes it at no additional cost. The software "collects samples from the process, analyzes the data for SPC trends, and displays SPC charts, all in real time," Mr. Walther says.
Speaking in multiples
It's not just integration that makes this class of software functionality more valued, it's more power.
ABB's Mr. Dingman says, "One of the important areas of SPC is multivariable SPC [MSPC]. Currently, typical Shewhart analysis only considers one variable at a time and whether it is within its control limits. What it does not address is the effect that multiple variables in combination can have on product quality." For example, two variables might be in control individually, but interact in such a way that moves product outset specifications. "Multivariant SPC would identify these interactions and indicate when product is made out of specification," Mr. Dingman says.
Important requirements for implementing online SPC/SQC applications, according to David McCaffrey, product development manager, MDC Technology Ltd. (Teesside, U.K.), are:
A server to perform MSPC model calculations;
Easy interfacing to plant data;
An internal historian for logging trends of model inputs, outputs and alarms;
Distributed client software to allow monitoring of model output and alarms either in real time or retrospectively; and
Embedding of MSPC client software and alarms into the same operator environment as the factory control and alarm system.
Fisher-Rosemount Systems (Austin, Tex.) embeds MDC products into the DeltaV automation system. In addition to interfaces with more than 200 distributed control and SCADA systems, MDC touts 10% reduction in batch cycle times, 4% improvement in product quality, and up to 5% in increased profitability to customer processes.
Real-time operation and MSPC aren't the only functionality users have demanded from SPC/SQC software vendors. Other major trends include:
Easier interfaces (set-up, alarming, and correction) for non-statistically inclined users, with re-emphasis on solving problems, rather than analyzing;
More sophisticated modeling tools so problems can be identified and corrected earlier, before product moves outside spec;
Better data storage and retrieval, in part, using standard tools, such as web-enabled software, ActiveX, and Object Linking and Embedding (OLE) for Process Control; and
Integration and information exchange with other software, throughout the enterprise.
Brad Klenz, senior quality technology specialist in the Quality Center at SAS Institute (Cary, N.C.), points to SPC software set-up as a potential source of grief when implementing these systems.
Building on Mr. McCaffrey's suggestion of a central server, Mr. Klenz recommends a "Quality Data Warehouse" for enterprise quality improvement. Such a warehouse collects data from and makes it available to systems such as SPC, enterprise resource planning, manufacturing execution systems, manufacturing resource planning, and laboratory information management systems.
Meeting the analytical needs of the manufacturing enterprise requires proper data aggregation, correct subgrouping, allowing for multiple sources of variation, sufficient structure for traceability, and support for measures of uncertainty, Mr. Klenz says.
Central storage allows access and analysis, resulting in decisions that improve the quality of processes, products, and services.
What have you done for me?
Making SPC and SQC data available to all levels of the enterprise, brings a multiplicity of benefits, explains USDATA Corp. (Richardson, Tex.) chief software architect, Mitch Vaughn. Availability allows users to automate product routing; classify based on quality, production line, plant or enterprise analysis; and achieve ISO 9000 compliance. Data integration also allows for easier tracking of rework and repair and ability to provide electronic delivery of quality information to customers, Mr. Vaughn says.
Competition's tougher than ever. "Customers expect tighter quality, faster delivery, and more flexible production," says MDC's Mr. McCaffrey. SPC/SQC online tools—when fed with reliably sensed data in real-time—can help with ahead-of-time, in-spec, lower-cost deliveries that customers demand.
Before buying SPC software, answer these questions
Choosing SPC software requires up-front time and effort, but the investment will more than pay off through improved productivity, happier customers, and cost savings from process improvements.
The key to choosing the right SPC software is to carefully evaluate your needs and validate them against the SPC applications you are considering. The following questions are a good place to start in evaluating SPC applications.
Will it handle your data? Make a list of your manufacturing variables, including the number and mix of variables per data set. Will the software allow you to add more variables without modifying the data file? Can you capture and analyze all types of data in one file?
How easy is the software to learn and use? Ease-of-use is important, especially if user experience is diverse. Commonly used tasks should be obvious, and user navigation through the application should be intuitive.
Can you compare factors , such as suppliers, shifts, or operators? All SPC applications produce charts, but choose software that allows meaningful comparative analysis, including the specification of chart scales. For example, the software must be able to compare two shifts on the same production line, and place multiple charts on one screen or printed page.
Can routine tasks be automated? Will the software allow you to easily automate routine charts and reports? Must you record keystrokes, or is there an easy to use wizard?
How powerful are the output capabilities? Does the application provide sufficient flexibility and power to adequately communicate quality information to internal and external customers?
How does it connect to existing databases? To adequately mine and analyze SPC data from multiple corporate databases, ensure the application provides easy ways to connect to a wide variety of database formats.
How flexible is the software in accommodating changes? Will the software be easy to adapt to new products, manufacturing methods, or customers with different reporting requirements? Good software is easy to change and takes only a few minutes to create a new chart or report.
Can the software be network based? Network based applications make mining data from a variety of corporate databases easier. If you start with a single-user application, can it be upgraded to a network version?
Will it produce the needed charts? Whether you need x-bar and range, pareto, median/individual, or other reporting charts, validate the application to ensure it includes the necessary charts for process control and customer and government reporting requirements.
How accessible is the documentation? Manuals tend to get lost or shelved, the software should include comprehensive, context sensitive on-screen HELP.
Not all SPC applications are created equal. Spending the time to define requirements and conduct a search to find the one that best meets your requirements will pay big dividends.
Jeffery L. Cawley, vp of Northwest Analytical Inc. Comments? E-mail David Harrold of Control Engineering firstname.lastname@example.org.
Understanding SPC terms
While using current SPC/SQC software requires less specialized knowledge than previous versions, here are some traditional SPC/SQC terms that might help.
Cp: Process capability calculated by Tolerance/(6 sigma). Only measures dispersion.
Cpk: Process capability calculated by the smaller of (Mean-Upper Spec)/(3 sigma) or (Mean-Lower Spec)/(3 sigma). It measures both dispersion and central tendency. A 6 sigma program implies Cpk of 2 or greater.
EWMA chart: Exponential Weighted Moving Average. It detects small recent variations by assigning an exponentially declining weight to data points as they age.
Pareto chart: A means of classifying problems from important few to trivially many. Defects are assigned a cause and categories ranked in order of number. The most significant problem will be the first.
Run rules: Guidelines to detect nonrandom process shifts. Some rules use six or nine data points on one side of the mean in a row. They are based on the likelihood that so many successive readings will not occur randomly.
X bar, R: This most popular of Shewart charts plots average (X bar) and range (R) for a sample.
Source: Control Engineering