Over the past year, the accessories division of Black & Decker Corp. has changed its entire approach to quality control. This division, which makes blades for power saws and other tools at its Tampa, FL, plant, no longer relies on inspectors to check each blade before it is shipped. Instead, production workers must ensure that all blades meet corporate quality standards before they leave the production line.

"The mindset in the plant was that manufacturing was separate from quality," says Simon Barrett, the division's engineering manager. "In this day and age, we can't operate in that manner. We must have the mindset of getting things right the first time, and that means carrying out in-process inspections."

Since adopting this approach, Barrett says Black & Decker has seen a significant reduction in product defects, and its quality assurance staff is starting to think about long-range quality improvement strategies rather than focusing primarily on uncovering existing product defects.

Barrett calls these changes "a cultural shift." He also says the new culture is built around the practice of a long-established quality control technique called SPC--or statistical process control.

Manufacturers have used statistics to analyze production processes for 50 years or longer. But SPC enjoyed its greatest popularity in the late 1980s, during what was known as the total quality movement.

Industry experts have long agreed that SPC is a reliable method of determining whether a company is meeting its own standards for product quality. Considerable debate has occurred over the years, however, about its value as a means for improving quality in an in-process scenario.

In simple terms, SPC involves collecting product samples at various stages of a manufacturing process and comparing those samples against a predetermined standard. The results are plotted on sheets called control charts. Samples falling within a certain range are deemed acceptable. If too many samples fall outside of the range, the manufacturing process is considered "out of statistical control." Products made during an out-of-control phase typically have defects.

Those who questioned SPC's effectiveness at improving product quality generally argued that it took so long to collect and interpret the data that defects could not be identified until long after a production run was complete. That meant the factory knew it had made substandard product, but the only possible remedy was to rework or scrap it and try to avoid making the same mistakes in the future.

Because it proved to have little, if any, immediate impact on product quality, industry experts say that many manufacturers gave up using SPC as a tool for continuous improvement. "It got to the point that companies would collect SPC data and basically just let it sit on a hard drive somewhere," says Kevin Prouty, an analyst with AMR Research (Boston), who tracks the use of quality management tools in the automotive industry. "Every now and then, a quality problem would crop up and someone would go back and review the SPC data to see if it would help uncover the cause of problem." In recent years, Prouty adds, some companies have admitted to using SPC data as little more than supporting documentation for their ISO or QS-9000 certification programs.

To foster continuous improvement, Prouty says, "manufacturers want more real-time information. They want to know that things are breaking before they are actually broken. It is no longer enough to tell them that a bolt is too long to meet specifications. They want to know who worked on the bolt, what product it goes into, and what customer is supposed to receive that product. They need that type of contextual information in order to make business decisions."

Recently, software developers have been responding to these issues by creating tools that make it easier to collect, interpret and disseminate SPC data. As a result, some manufacturers--such as Black & Decker--are viewing this old technique in a new light.

"Our customers cite a variety of benefits from using our system, including reduced raw materials and labor costs," says Warren Ha, president of Zontec Inc. (Cincinnati), which sells an SPC package called Synergy 2000. "One customer was able to completely eliminate end-item inspection, which allowed the company to transfer eight inspectors to other jobs. Customers also tell us they are getting new products to market more quickly, because they are able to make their production processes meet quality specifications much faster."

Teeth problems
Barrett attributes Black & Decker's success with SPC directly to its use of a software package called WinSPC that was developed by DataNet Quality Systems (Southfield, MI). "We were having problems with the positioning of the teeth in some our blades. So we decided to go back to basics," Barrett says in explaining the catalyst for Black & Decker's SPC initiative.

Initially, going back to basics meant practicing SPC the old fashioned way. Production workers used hand gages to take measurements on every fifth blade that moved through their work cells. They typed those measurements into a computer and later used the data to generate control charts that were supposed to show whether the cells were producing blades that met specifications.

This didn't work very well because the workers weren't sure of how to read the control charts, and they quickly grew tired of collecting the data. "We had no way of knowing if workers were taking all of the required readings," Barrett concedes. "So we really never did anything with the data. It was reviewed every once in a while when someone was studying our manufacturing capabilities. But we never used it to drive process improvements."

Barrett says that changed in January 2001, after he discovered WinSPC. The first thing that struck him was how easy the program was to use. "I taught myself to use it before introducing it on the factory floor," he says. "The workers started using it almost instantly because it was so much easier than what they were doing before."

With WinSPC, the same gages are used to take product measurements, but those gages are connected to a computer that holds the WinSPC package. Data flows directly from the gages to the WinSPC database.

"We have immediate access to that information," Barrett says. "Three seconds after they enter new data, the operators know if their process is outside of control limits, and they can stop the process and make adjustments. You can set the system to flash an on-screen warning, sound an alarm or send an e-mail message."

Richard Slanksy, an analyst with consulting firm ARC Advisory Group (Dedham, MA), says two specific forms of technology--plant-floor devices such as programmable logic controllers (PLCs) with embedded software and Internet-based networks--are spawning this new generation of SPC systems. He also confirms that these systems can help manufacturers do more than identify product defects. "We are moving into an era of what I call collaborative manufacturing," Slanksy says. "It is based on the old vision of having systems that provide a direct connection between the shop floor and the boardroom, allowing executives to see everything going on in the company.

Closed loop
"We are at the point where you can set parameters for a production process and then embed a software device inside a PLC that will monitor the process and send out an alarm when the process moves out of tolerance," Slanksy explains. "Because this information can travel in real time, you can create a closed loop to send instructions back to the production machines to recalibrate their settings on the fly. And with Web-based systems and multitier network architectures, you can pass that information up to your supply chain management (SCM) or enterprise resource planning (ERP) system, where it can be shared throughout the entire value chain--with remote sites within your own enterprise or even with customers and suppliers."

Zontec's Ha says his company's Synergy 2000 package has a three-level architecture that supports what amount to three separate systems--one for production workers, another for manufacturing engineers and a third for managers and executives. "The engineer level is for the quality manager who sets up and manages the system," Ha says. "This is where you map your production processes and establish the specific points at which you want to collect data." Engineers also use this level to set production control limits and to determine if and when operators will be warned that they need to adjust specific machine settings because a process is moving out of statistical control.

Production workers use the operator level to input data from production machines and to monitor processes. Data relating to each process being monitored goes to its own separate database, or what Zontec calls a data bank, where it is accessible to anyone using the system--operators, engineers or managers.

"Information can be funneled to control charts," Ha says, "and the system can be programmed to update those charts whenever new information is input. Meanwhile, managers can monitor multiple processes, through numerous different data banks, on a single screen. They can see colors change in real time as process conditions change. They can click a button to get a detailed view of charts for specific processes. Some companies use this for real-time monitoring of production activity at multiple plants."

The spread of Internet-based networking also is enabling companies to blend SPC data with information from other plant-floor applications such as manufacturing execution systems (MES) to create broad-based quality management programs. "Because most SPC tools, and a lot of the infrastructure in plants have a Web component, companies are able to aggregate data into what amount to quality management portals," says AMR's Prouty. "We call this enterprise manufacturing intelligence. These portals don't just show SPC data. They grab information from various systems around the plant and present it in a form that it useful for making business decisions. Three years ago, these types of portals would not have been possible," Prouty adds, "because it was too difficult to get all of this data."

Stephen Wise, director of statistical applications for Infinity QS International Inc. (Manassas, VA), says his company's SPC package, called Infinity, offers a method of organizing data that gives users control over how they will use the data collected. "We also let users hold data in any database they wish, as long at it complies with open database connectivity standards. Essentially, our customers design their own methods of collecting and storing the data, and they use our system as a statistical engine to organize and analyze the data."

Wise says Infinity tags data according to multiple parameters--such as part numbers, production process numbers, machine numbers or test numbers. Once data on a process is entered into a database, users can design queries to get information based on any of these parameters. For instance, a typical search could be, "Find all out-of-control conditions for production process 24 for the past five days."

Wise says this type of search capability gives users a better way to analyze their operations than previous generations of SPC systems, which typically only allowed for examining the condition of parts.

Molecular Devices Corp. (Sunnyvale, CA), a manufacturer of equipment used in pharmaceutical and biotechnology research, is using an SPC system embedded inside an MES package as part of an enterprise-wide quality management program. "SPC is really a subset of what we are trying to accomplish," says Robert Murray, Molecular Devices' vice president of operations. "We have been moving toward a flow-based production operation, primarily to help us get orders to our customers more quickly. To do that, we need to be able to make more real-time decisions."

To accomplish its goals, Molecular Devices is depending on an MES package it purchased from Datasweep Inc. (San Jose, CA). With this system, Molecular Devices aims to collect and distribute information from its plant floor fast enough to allow anyone within the enterprise--from production workers to executives--to make decisions that will help move products out the door faster. The SPC component of the Datasweep package is part of a quality assurance module that allows manufacturers to capture information on product defects and measure how manufacturing processes are performing against the company's own predetermined performance metrics.

Follow the wizard
Julie La, a Datasweep manager, says this component contains an "SPC wizard" that helps users quickly define data collection points and control limits. In addition, the wizard leads users through the process of establishing rules for notifying the appropriate parties when out-of-control conditions occur.

Murray says Molecular Devices is using this wizard to set up an SPC program on one of its production lines. The first step is establishing the points at which data will be collected. "We want to make that process friendly to the operators," Murray says. "A lot of the data we collect will be imported automatically from production machines into a central data file. Operators will then be able to click a button and generate SPC charts based on data from that file." If one of those charts shows an out-of-control condition, it will trigger an alert, telling the operator to take some action to correct the condition.

The system also has a workflow feature that allows for setting rules for sending these alerts to people outside of the manufacturing area--such as manufacturing engineers--when the out-of-control condition warrants their attention.

Murray says Molecular Devices also plans to use the Datasweep package to conduct quality checks on its completed products. The idea is to compile a complete history of each product--including all of the parts it contains and which suppliers provided each of the parts. "That will allow us to manage the total production process," Murray says, "from the supplier who provides raw material or subassemblies all the way until our customer gets the product. That will help us in resolving any problems with our key suppliers. If we can collect and analyze data related to their components, we can go back and pinpoint the problems for them. We also can show them whether a particular problem represents a slight variation or a longer-term trend."

La, the Datasweep manager, says some users are feeding this type of SPC data to their suppliers in real time through Web sites that are connected to the Datasweep systems. But Murray says, "that is a little further than we want to go right now. We will be passing our information to our own production operators and engineers in real time so that they can intervene in our own process to make corrections before we send anything out."

In terms of working with suppliers, Murray says, "at this point, we are glad to know that we can get quality data that goes deeper than just inspection level. We also like knowing that we can get comprehensive data from throughout the supply chain."

That seems to a growing sentiment among manufacturers who are adopting the new breed of SPC systems.