Manufacturing has adopted embedded chips on a massive scale since 1975. These chip-based production systems are date-sensitive. The faulty chips must be discovered and replaced, one at a time.
Most U.S. businesses are not through the assessment phase. I know of no business that has announced, "We're finished!" (Actually, they're all just about all finished. They just don't know it yet.)
This article from Digital's site is forthright in its description of the problems. The author offers no industry-wide solutions. That's because there aren't any.
* * * * * * * *
Recent research indicates that the cost of fixing the manufacturing problems at the plant-level may be at least half of what a company spends to fix overall data center issues.
Machines on the factory floor are very sensitive to incorrect dates - more so than was expected. For example, a modern pharmaceutical plant maintains 83 computer systems with three million lines of code. Within that code are 120,000 date references with potential Y2K problems. In addition, the plant runs 138 automated production systems with 400 date references plus 200 machines with embedded software. . . .
The problem on the factory floor began 20 years ago when manufacturing found that computers could streamline their operations, making a company more efficient and thus more profitable. In those early days, off-the-shelf software was practically non-existent so each plant developed programs that suited individual manufacturing specifications. The result was that custom software ruled the factory floor.
During this time period, about half of the software written for manufacturing was written in Cobol. The remaining software was written in a variety of computing languages that might as well be gibberish. What this means is that although there are tools currently to hunt for zero-zero (00) date errors in Cobol and a few other languages, few exist for the vast number of so-called embedded systems.
Embedded systems are chips and programs (not readily accessible or even visible) which are integral parts of control and production equipment. Many must be decoded and fixed individually. Repairing devices and software programs is tricky since it is a 'given' in the industry that new program errors will be introduced in seven percent of routine repairs.
Compounding the problem is that many of these programs can't be fixed because they are inscribed on silicon chips. In those cases, manufacturers are forced to scrap any date driven plant equipment. The only good news is that these moves force manufacturers to purchase leading-edge products that will improve their efficiency and overall competitiveness.
To date, the majority of U.S. manufacturers haven't even completed a plantwide assessment to learn the depth of the Y2K problem. With the economy booming, manufacturing plants are running three shifts, seven days a week. Companies find it difficult to replicate Year 2000 conditions before they happen. Because the factories can't afford to close down, the solution involves testing during off-peak hours, over planned shutdown periods or buying expensive back-up equipment.
General Motors serves as a good example of how traumatic the situation is. With over two billion lines of code, GM is the world leader in the number of computerized systems. As part of its Y2K program, the company is retiring 1,700 obsolete computer systems. Estimates to eradicate the millennium bug at GM run between $400 and $550 million.
The severity of the problem at the giant auto manufacturer was recently brought to light when the company ran a test with some of its robotic devices. Y2K problems caused the robots to freeze - an act that could shut down the entire assembly line.