I'm writing this in the aftermath of Hurricane Irene. Up where I live, we had some wind and rain, and lost power for good 15 hours. We got off lightly – I know there were many who did not and my heart goes out to them.
Although aiming a couple of monster Maglites straight up at the ceiling makes for a nice ambient glow, it's not so great to read by, which left me a lot of time for thinking. Not surprisingly, one of the things I thought about was disaster preparedness – not just personal preparedness, but how industry copes with disaster. Business today is hard enough when everything goes well, but these days, we have more than our share of catastrophes. Quite aside from the human toll, such events take an economic toll, especially for companies caught unprepared. With our global economy, almost any natural disaster reverberates around the globe via the supply chain. In the aftermath of the Japanese earthquake, for example, supply-chain disruptions in the automotive industry left factories idling half a world away. Does your company have a plan in place to deal with this type of worst-case scenario? Perhaps just as important, do your vendors?
As engineers, we design systems. We develop detailed error budgets, we plan around failure modes, we research and test components. We concentrate on designing the system that will deliver the performance customers need, and then we go hunting for components, focusing on specifications, cost, quality assurance, and delivery schedules. But there’s another aspect to keep in mind – availability. You can’t build products and systems without the raw materials and components. Especially if you’re working on a contract with high penalties for late delivery, you need to know that your vendors have a strategy to come through no matter what. Part performance and on-time delivery statistics aren't enough. You need more. Do they have geographical redundancy in their production facilities? How about their supply chain? Do they have a plan for recovery? If so, how robust is it?
As the one-two punch of Japan’s monster earthquake and tsunami vividly demonstrated, some catastrophes defy all efforts at preparation. It’s one thing to plan for a disaster like a fire that destroys your factory. It’s another when the disaster destroys your entire town. In a normal storm, for example, power crews are out restoring power within days, if not hours. But in the type of widespread destruction experienced in Japan, or the southern United States after Hurricane Katrina, it’s not just the power lines that are down, it's the power company facility, the crews, the roads, and in some cases, even the power plant itself. Suddenly, recovery isn't a matter of days or weeks; it stretches out to months, even years.
Whether it's earthquakes or killer hurricanes, tsunamis and floods, volcanic explosions, tornadoes, or wildfires, every place on earth is vulnerable to some type of natural disaster. Amid all the fearful news, you can find inspiring stories, however, and tales of companies who've gotten it right. If you haunt the EE Times website, perhaps you saw the story about Renesas Electronics restarting its wafer lines. The microcontroller manufacturer had plenty of geographic redundancy – a total of 21 facilities spread throughout Japan – but the quake interrupted operations at eight of them. Worse, Renesas’ main fab in Naka, Japan, suffered severe damage, with debris everywhere, machinery destroyed, and structural damage that allowed outside air into the clean rooms. Management initially estimated that it would take six months to recover. Thanks to customers, competitors, and the heroic efforts of employees, the Naka facility resumed operations in half that time.
When you're choosing a supplier, it's essential to keep in mind the worst-case scenario – and to know that they’re thinking about it, too. One approach is to choose a multinational vendor with at least two geographically separated facilities that can manufacture your parts. Alternatively, you can establish a backup supplier, selecting and qualifying them and their components in advance. You’ll probably still experience scheduling delays, but the disruption will be minimized. Of course, either of these options is easier to accomplish if you're working with commercial-off-the-shelf (COTS) components, which present challenges of their own. If you're working with custom components, another alternative is to keep a backup supply of components on hand, but that forces you to tie up capital in inventory.
How does your company deal with the whole issue of disaster-preparedness and vendors? Is it part of your vendor qualification program? Do you add language to the contract? Do you have contingency plans for every project?
Did you find this article of interest? Then visit Military & Aerospace Designline, where we update daily with design, technology, product, and news articles tailored to fit your world. Too busy to go every day? Sign up for our newsletter to get the week's best items delivered to your inbox. Just click here and choose the "Manage Newsletters" tab.
2nd sourcing used to be standard practice. At one time a lot of purchasing departments could rule out the use of a component simply due to its being sole-sourced. Getting around those rules might have taken just short of an act of congress.
Today, with so many different specialized chips, it is, in many cases, simply impossible to have a second source. If you use Microchip MCUs, you can buy them from Microchip. That's it. Atmel from Atmel. An Allegro motor controller will only come from Allegro.
I do, however, think it's possible to prepare for limited disasters. With the Microchip example - say they have a major supply disruption. The supply of the particular MCU you use runs dry. Mnay of their parts maintain a lot of pin compatibility within the same pin-count packages. If your MCU is not available, they might have another close one that could do the job with minimal code changes.
Motor control chips, blue tooth and ZigBees might be more of a challenge, but the functionality is fairly standard. PCB real estate might not allow for having two different footprints, but maybe a small daughter card could fit the bill.
To be fair to Japan, the earthquake had little effect. It was really the tsunami that caused the most damage.
While having penalty clauses etc in contracts might give you a certain degree of comfort it does not help you if parts cannot get delivered. Or, put another way, it might give you someone to blame, but it does not give you a solution.
I recently had this discussion with some people and we had the idea of designing boards with overlapping footprints for key components like micros. Design for two equivalent parts from 2 vendors with supply chains in different countries. Choose one part as your primary part. If you lose that vendor then hopefully the other will survive ad you can respin your product quickly.
I live in Christchurch NZ where we got hammered by earthquakes. Luckily we didn't lose much manufacturing.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.