Mark your calendar, Mr. Vice President. In six short years, 2011, the first group of baby boomers will begin retiring from your company, each taking 40+ years of experience with them. Companies that make a living on problem solving and innovation will really feel the pinch. And it will only get worse. In 2012 the exodus will accelerate as the number of retirees grows and subsides over a span of about 19 years.
How will you deal with the coming brain drain? There are several ways to capture the expertise, knowledge, and know-how of the near retirees and make it accessible to the rest of the firm.
BUILDING A KNOWLEDGE-BASED SYSTEM
A few companies in the early 1990s, such as Icad and Wisdom Systems, tackled the problem with software that could be primed with design rules for a company product. Then with a few inputs, the software would spit out production drawings and generated costs for building complex products. The software was aimed mostly at generating quotes and drawings quickly. Companies using the programs reported definite benefits, but the software was difficult to set up, took the best people away from design tasks, and needed a lot of trouble shooting and coddling. These two companies do not exist today. But in their demise, they left the seeds for new software more focused on specific problems.
For instance, Enductive Solutions, Lebanon, N.H., specializes in capturing design rules for building products and then running the products through simulations to test and evaluate their performance. Mathsoft has developed a server-based repository for saving and sharing calculations generated by company engineers. This Web-based system lets others in the company pull up equations with commentary from their author and modify them for more current research. Caterpillar, a manufacturing company, has developed a system-that lets engineers find others in and outside the company with the knowledge to solve specific types of problems.
Most companies that successfully apply knowledge-based engineering (KBE) build similar products in many variations, such as heat sinks. "Even though making heat sinks is a commodity business," says Bart Patel, President of Aavid Thermalloy, Concord, N.H., "there are good reasons for applying knowledge-based design. For instance, our design expertise is mostly in the heads of a few engineers, each with different experience levels. If you give them the same inputs, each would design a slightly different heat sink. So consistency becomes the first reason. In this fast-moving business, we want a design that can be built in any facility around the world."
"We could have made design work more uniform by insisting everyone work from the same manual. But these are expensive to develop and update. Plus, someone could walk off with one. Then we would have a competitor with all of our know-how. So security is an issue and protecting intellectual property is a major concern, especially in places like China," he says. Aavid's KBE system solves these by staying on a server in New England with access limited to a few dozen people.
Aavid's KBE system was developed by Enductive Solutions. Their Director of Application Engineering, Scott Del Porte, tells how it was developed.
"For the first part of the threemonth project, our engineer collected information in the Aavid offices. He flowcharted their design process to identify where data came from and where it goes. Then we wrote a description outlining what the system should do and presented it to Aavid engineers. After an iteration or two, the programming began. The software is now installed and Enductive maintains it," says Del Porte.
Patel says the system has performed as expected and generated the anticipated benefits, such as faster turnarounds, higher-quality heat sinks, and fewer errors. But there were a few benefits he didn't anticipate. "For instance, it's Web based, so sales engineers collect all the needed information the first time. Also, all designs are now archived in the same database, not in private file cabinets and desks, so new designs start at a more advanced point, not square one. The software also makes it easy to run sophisticated CFD and FEA simulations at the push of a button. There is an audit trail to show what was done and why. And lastly, a cost estimate tells us when we are headed in the right direction with a design."
Engineers in the precomputing era would often develop calc sheets for projects. These would step through a series of equations to show unit conversions and validation columns. But the idea got lost when people started writing programs and spreadsheets unless they did a good job of documenting what they did.
Mathematics has been the language of engineering for hundreds of years, yet calculations are now stored mostly in formats unsearchable by computers. Somehow the evolution of electronic documents left out electronic math.
Mathcad V12 and two other ideas may reverse the trend with server-based technology that stores worksheets in an open XML format. "XML codifies everything in the worksheet in a nonproprietary way," says Allan Razdow, CTO and vice president of products and technology at Mathsoft. "It's a description of the applied math engineers used to find graphs or other data. The math software revives calc sheets in the form of worksheets with new possibilities. For instance, the sheets show a calculation's inputs and how outputs were derived."
Razdow's additional programs, Designate and Provenance, lets engineers copy and reuse existing worksheets for new tasks, and record from where worksheets and equations came, especially if they came from another worksheet. "This lets users view and search worksheets in a browser," he says. "Users can ask questions that were previously difficult to answer, such as: What worksheets were used to find the strength of a particular structure? And if you want to do something similar to the calculations on my existing worksheet, I'll e-mail it to you, you open it in Mathcad, and rerun the equations with your inputs and modifications. Later, someone can ask who reused my original worksheet and the system comes up with them."
"No matter where an equations goes, as long as its been copied by Mathcad or a compatible application, it's traceable. The three programs let companies trace numbers, equations, and anything on a worksheet. It captures user knowledge and lets others search it, and all on standard computers," he says.
Razdow adds that his company is working with the National Institute of Standards and Technology on UnitsML, a standard that will, among other things, keep people from mixing units in electronic documents such as nanometer and nautical miles, which are both abbreviated nm.
HOW CONCEPTS TAKE SHAPE
Narrowing design concepts into a final product calls for discarding what some consider useful intellectual capital. It's useful because it explains why something did not work. But what system or procedure can capture a designer's thoughts? Early schemes for nailing down engineering thought required a lot of extra writing or typing. No wonder they flopped.
A recent requirements-capture tool, say its developers, cuts time from design and development cycles, and captures best practices without additional administrative overloads and with immediate benefit to the end user and enterprise. "Valuable intellectual capital gets thrown away in any design process as concepts are refined into a final product," says CEO Igal Kaptsan at KollabNet Inc. A technique called Engineering Information Modeling (EIM) makes design engineers more productive by capturing and linking to data created in the everyday use of programs such as office productivity, PDM, FEA, and CAD software.
Kaptsan says EIM, intended for manufacturing firms, captures and tracks knowledge. It lets users do things that are otherwise impossible, leading to the preservation of requirements throughout the engineering process, capture of knowledge and alternatives, and design-process auditing. The technique also encourages reuse of engineering knowledge. It can also perform an instant impact analysis for any what-if scenario, and increase profitability, and reduce error rates, an effect that improves over time.
The software works like this: A designer in a traditional workflow gets a data package along with an assignment. The package is a stack of documents from which the designer is expected to extract design constraints. Such documents today are often electronic. A team leader would then take out a pad and pen and make notes while poring over drawings, and reading e-mails and Word files. KollabNet software replaces the pad and pen.
So when users find a statement that describes a design constraint, such as "Operating temperature range: 40 to 50°C", they simply select or highlight the detail and switch over to KollabNet. There the designer selects a constraint DesignBlock from the software's palette and places it on the software's work area or DesignMap. A constraint block holds any string with specific design targets such as weights, sizes, and operating temperatures.
Then the designer clicks on a button labeled "Word" to indicate the requirement came from a Word document. The action captures the selected string of text and where it came from. "The action eventually builds a symbolic model of the design by linking requirements and constraints. The software, meanwhile, is doing the accounting. It's keeping track of what changes, how, and why, so users have an audit trail of events," says Kaptsan.
"The power of the software is in capturing the rationale behind a decision. Every DesignMap stores reasons, constraints, and design choices for every change, starting with requirements in the concept phase, through design, and release to manufacturing," he says. "This way, a complete history is available. The software doesn't create knowledge nor tell how to design. It just helps designers track what they do, why they did it, and what is affected if they change it," says Kaptsan. "What's more, a DesignMap developed by one engineer can appear as a component in another DesignMap to create a top-down environment. This embedding builds knowledge as users share and link DesignMap throughout an organization and with those from subcontractors," he adds.
DRIVING CAD WITH RULES
One of the earliest and still widely used ways to drive CAD systems is by linking them to Excel spreadsheets. The idea works for simple parts, but gets unwieldy trying to make and manage parts of even moderate complexity. At least three companies, RuleStream, Vision KBE, and DriveWorks, have tackled the limitations of spreadsheet drivers by designing equipment based on ifthen statements, look-up sheets, rules of thumb, and of course, equations. The three programs also put more order into the driving rules and make it easier to pass such engineering know-how to others in the company.
All three programs drive the SolidWorks CAD system and provide links to MRP and ERP systems to facilitate production. A DriveWorks demo also shows how companies might let clients design complex products online through a Web interface. PartStream.net would then allow a visual examination of a 3D model of the final device, which could also be purchased online. All programs also link to several PLM systems to store designs for quoting and reuse.
The Babcock & Wilcox Co., for example, uses RuleStream to design and build power-generation equipment, most of which is engineered-to-order as part of a request-for-proposal bidding. The company had been using an aging knowledge-based engineering (KBE) system. Management wanted to upgrade to something more modern, so it eventually selected RulesStream because it could drive the company's CAD system in Windows. It also has an interactive development environment, uses Visual Basic for scripting, and has a rule and project database.
"Customers typically have unique requirements for fuels, emission standards, and boiler arrangements," says Manager of Engineering Systems Kip Alexander. "A design is also constrained by our own rules developed over years of experience."
Since transitioning to the software, B&W develops burner designs in days instead of weeks, while turning out proposal concepts that are close to final designs. Alexander also says this is the first time a nonprogrammer has been involved in development. The company plans to use the software for other devices not previously automated, and expects to reduce overall time to final design by as much as 90%.
The human side of knowledge networks
The Knowledge Network (KN) developed by Caterpillar Inc. connects company engineers through thousands of "communities," groups with common interests or fields of expertise. So far, there are 3,900 communities covering everything from project management to bolted joints. But the network extends outside the company to bring in suppliers and some of the most experienced people that don't work there anymore: retirees. About 1,200 of them participate. "Some retirees like to stay engaged," says Reed Stuedemann, IT Manager and KN administrator. "It's what they've done all their career. It's their identity. In the middle of winter they can be lounging in Palm Springs and still help solve problems in Peoria," he says.
The network is about people, not the data, he adds. "When someone has a question on bolts, for example, they go to the right community, review and search reference material and specs, and post questions. If that doesn't solve the problem, they can e-mail or call an expert on the topic. Many companies get hung up on tools, yet 80% of the knowledge is in the heads of their people."
Stuedemann says the system is intuitive and needs little training. "Our studies indicate the biggest barriers to using it are lack of computer skills and the fear of looking foolish."
The company developed the network at its research facility and rolled it out in 1999. It now plans to license the network.
Enductive Solutions Inc.