When the Result of CAD is Failure (With Some Solutions)
by Ralph Grabowski
Provided we use their software, CAD vendors promise us output-perfection. So why then do CAD-based designs screw up, costing manufacturers sometimes billions of dollars?
The problem can be due to a flaw in the CAD software; but more likely mistakes stem from the ecosystem surrounding the software: the CAD operators, their managers, the design firm's culture, consultants who advise poorly, politicians in over their heads, and so on.
Here are four of my favorite design horror stories, along with guidelines to avoid costly errors.
The Wire that Created Airbus's Billion-dollar Delay
Perhaps the most embarrassing foul-up in computer-aided design's short history was the multi-billion dollar, multi-year delay in the debut of Airbus's giant A380 passenger aircraft. The problem itself was due to a change order involving electrical harnesses (those complete sets of wires with connectors at either end); the source of the problem might have been a consultant. (See figure 1.)
Figure 1 Example of wiring harness used in aircraft (Image credit BEA)
Airbus manufactures aircraft parts in several countries, with the A380's final assembly taking place in France. To design the planes, the company usesDassault Systeme's Catia software, but during this time it was making the difficult transition from Catia V4 to Catia V5. Airbus hired IBM to write a translator so that divisions using the new V5 could read drawings from the old and very incompatible V4. IBM Consulting, however, faced delays delivering the translator, and so French operators using V5 were forced to re-input by hand drawings created with V4. This was a contributing error.
The #1 concern these days in aircraft design is lowering fuel consumption, and one way is to reduce the weight of aircraft parts. Part way through the design process, a decision was made to switch all wiring to lighter aluminum. This would save 10% in weight over copper, or 500kg (1,100 pounds) per aircraft.
A shop in Germany designed and manufactured the harnesses using Catia V4, which meant the routing drawings could not be checked against the master plans in V5. When the harnesses arrived for installation in France, they were too short. The reason: aluminum wires require a greater bend radius than copper, and so have to be cut slightly longer overall. While the change had been made to the specification, the change was not reflected in the routing diagrams. This error cost Airbus billions in re-manufacturing costs, delayed deliveries, and lost sales.
Solutions. It is unwise to upgrade CAD systems during projects; always keep using the old system until the deliverables are complete. Behemoths like Airbus and Boeing cannot follow this advice, because drawings used to design an aircraft needs to be readable for maintenance up to fifty years later. For this reason, Boeing is keen on neutral file formats (like STEP) so that they are not dependent on the vagaries of CAD software suppliers and their proprietary formats. When a direct translator is not ready, I recommend using a neutral format, which, being open, can more easily be extended and inspected.
(While Autodesk's new subscription-only policy would force manufacturers to upgrade their CAD systems repeatedly during long projects, I have noticed that Autodesk has frozen the DWG format at R2013; perhaps the same will be true for its other file formats.)
The harnesses were made too short because at the manufacturing end someone forgot to recalculate the bend radii, while at the assembly end no one could check the routing information due to file format incompatibility. When a significant change is introduced to a design, I recommend doing a one-of prototype to check that the change will work.
Berlin's 21-year-old Non-operational Airport
Following reunification, the city of Berlin found itself with three airports. The historic Templehof airport was shut down, and plans were made to replace the other two aging airfields with a new one. It would be a showcase suitable for the capital of the largest, most-prosperous county in Europe. (See figure 2.)
Figure 2 Brandenburg airport under construction in Berlin (Image credit Peri Slabs)
Nearing completion in 2010, all of the airport's systems underwent testing: baggage, staffing, security, even people acting as passengers. The first system to fail significantly was smoke ventilation: passengers and staff were in danger of suffocating should a fire break out.
"For aesthetics, designers decided that the terminal would have smoke extraction ducts in its roof, but not on its rooftop. In case of fire, smoke would be pumped from the roof into exhaust ducts running down and below the structure, requiring the natural rising behavior of hot air in the duct to be reversed. Achieving this on the scale necessary for this airport is a unique undertaking and, thus far, this elaborate [18km-long] smoke extraction system has not worked as planned," reported German news magazine Der Spiegel.
After the ventilation supplier went bankrupt, the airport attempted to speed things along by proposing to hire 700 human fire spotters; this sadly hilarious solution was quickly rejected by the German agency responsible for building safety. Instead, entire walls are being torn down so that the ventilation can be rebuilt.
The results are costs and delays even greater than for Airbus. The design and construction had expected to take 12 years and cost e2.8 billion (US$3 billion). But politics and favoritism, lawsuits and bankruptcies, bribery charges and firings, the ventilation problem, and other construction flaws (like incorrect wiring, uninsulated hot water piping, and rain leakage) increased the project cost to perhaps $11 billion and life to 21 years. At this point, an opening date still has not been announced.
Meanwhile, it costs $12 million a month to keep the non-functioning airport humming. Nevertheless, the citizens of Berlin don't mind the delay, because the new airport is 18km further away from town and the old ones continue to operate just fine.
Solutions. The person in charge of the new airport project had no construction experience, his other job being mayor of Berlin. A project this prestigious inevitably suffer from cost overruns and delays; it is a law of nature. I recommend that government-initiated projects become public-private partnerships, where a private project management and construction firm takes on the full risk, but then profits from the operation of the completed facility. This minimizes the problems created by political motivation, and protects tax payers from cost overruns.
One of the reasons I left engineering was because it discourages innovation, which I wanted to implement. But it was best I did not. There are too many lives at risk should engineers be allowed to experiment with radical new designs -- such as ventilation systems that try to force smoke against its natural will, downwards. (I am guessing that someone wanted clean roof lines for the viewing pleasure of passengers on incoming aircraft.)
I recommend that engineers refuse to create designs that buck the laws of physics -- or common sense.
The Safest Saw No One Can Afford
On to some smaller design missteps. A Solidworks World event several years ago featured the inventors of a table saw that stops instantly should a finger touches the blade. When it is demo'ed, a wiener is used instead of the demo jock's finger. (See figure 3.) Nevertheless, it's a pretty amazing invention and one you'd think would be a standard feature by now. But it's not.
Figure 3 Checking the saw's brake with a finger-like weiner (Image credit Woodworking Adventures)
I asked a friend, who runs a chain of tool stores, about it. It's just too expensive, he told me. The brake works by sensing the finger's natural electrical current, then slamming a chunk of metal into the saw, ruining it. So no other vendor has adopted the technology.
Solutions. We see here a radical new idea (instant stop) costing too much, but then sparking the development of a cheaper solution after decades of design inertia. Some saws now employ an electric brake that briefly reverses the direction of the motor for a stop that takes a second or two. Not instant and not when a finger touches the blade, but it saves the saw and works when fingers are most likely to touch the blade: after the cut is finished, rather than during the cut.
I recommend rethinking expensive solutions, because good-enough solutions become far more widely adopted, being more affordable than perfect solutions.
The Imperfectly Designed Perfect Toaster
At another Solidworks World keynote, a pair of industrial designers from England showed us how they did their work. Instead of assuming how a consumer product should work, they spend a week watching how people use existing models -- to learn the good aspects and bad. They watch instead of ask, because people often give answers they think researchers want to hear.
Only after spending a week observing do they begin developing the new design. One of their solutions was a new toaster: it has a slanted top. From watching people, they found that consumers wanted to be able to look inside to see how the toast was doing. Cool, I thought. The next time I needed a new toaster I bought their design; after all, I actually had met its designers! (See figure 4.)
I quickly found that the slanted top is a pain. When the toaster is done, it shoots the toast out with force, often landing on the kitchen floor. As engineers, we know that 45 degrees is the ideal angle for maximizing projectile distance. Nor can I toast a third slice at the same time, because it slides off the sloping top. Plus, it's not particularly good at toasting: one side is always less done than the other, like every other toaster. In any case, I found I never needed to look inside to see how the toast is doing -- maybe that's just a British thing.
Solutions. I like innovative improvements to otherwise mature products, like table saws and toasters. (My current favorite is the addition of a laser beam to mitre saws.) CAD, unfortunately, allows us to be innovative for the sake of marketing. That toaster with the slanted top certainly stands out on the retailers' shelves, but otherwise is a fail.
I recommend designers employ a feedback loop, where radical new designs are tested in situ: does it solve the problems originally uncovered? Had the design duo done this, they would have found that it didn't work for consumers, and perhaps they could have turned their attention to the real problem: evenly toasting bread.
CAD vendors are fond of trumpeting uses of the software that cleanly solve design problems. They are loathe to admit otherwise. So it becomes our job, as customers, to be ready to overcome their hubris.
Just because we use CAD for design doesn't mean the design will work out for the customer. Nor can CAD software overcome post-design delays and human-induced problems. The solution is to be hyper aware of these limitations, to factor in incremental checking of radical changes, to even be ready to kill your pet project when it appears that the outcome won't match the initial expectation. No project is too big too fail.
[This article first appeared in Design Engineering magazine, and is reprinted with permission.]
Sponsor: Okino Graphics
== Professional 3D File Conversion/Viewing/Rendering Software ==
For over 2 decades Okino (Toronto) has provided mission-critical 3D conversion software used extensively by tens of thousands of professionals. We develop, support and convert between all major CAD, DCC & VisSim formats. Robert Lansdale (CTO, firstname.lastname@example.org) tailors each package to the specific conversion requirements of each customer.
Popular formats include 3ds Max, Maya, C4D, LW, ProE, SolidWorks, Inventor, SketchUp, DWF/DWG, DGN, CATIA, IGES/STEP/Parasolid, 3D PDF/U3D, JT, FBX, Collada & more. We know data translation, and provide immaculate developer-to-customer relations. http://www.okino.com
And One More Thing...
I like it when a CAD vendor sends a cleverly-worded press release and so I am happy to share this one from CCE. Here is the text, edited by me for brevity:
At a time when a lot is being written in the press about new cloud solutions for engineering data management, CCE found that a lot of SMBs [small and medium businesses] are too small to justify investing in an expensive PDM solution, however, they have enough product data in their organization to make version control and back-up a nightmare.
These customers are CAD-enabling Dropbox to manage product data across the organization. The part that is missing from Dropbox is its ability to handle engineering/CAD data. CCE's customers use the free CAD viewers available on CCE Labs to address that problem: EnSuite View and LAVA-Cat5. (See figure 5.)
Figure 5: Home page for CCE Labs (Image credit CCE)
"No knock on other new, cloud-based engineering data management products with multi-CAD support that supposedly do the exact same thing, since they likely use Dropbox infrastructure under the covers, but we have been able to successfully address this a long time ago without having to reinvent the wheel. So, Dropbox plus the free CAD viewers from CCE Labs is your engineering data management solution," said CCE vp of sales and marketing, Vinay Wagle. http://www.cadcam-e.com/CCELabs.aspx
Note that these free products require an Internet connection so that CCE can send ads to the software.
Even More News
Read the blog at WorldCAD Access as I write more about the CAD industry, and give tips on using hardware and software. You can also keep up with the blog through RSS feeds and email alerts. Some of the posts from the last week:
Autodesk settles with ZwSoft: the details
foto of the sunday: frozen leaves
I'm also on Twitter at @upfrontezine throughout the day with late-breaking CAD news and wry commentary, such as....
upFront.eZine (@upFronteZine) Nov 9: Remember when CAD execs excitedly told us our phones are more powerful than early NASA computers? Not seeing phones planning moon missions.
Thank You, Subscribers!
Thank you to you subscribers who donate towards the operation of upFront.eZine.
Ian Mackenzie: "Thanks for your thoughtful and balanced CAD commentary -- keep it up!"
Should you wish to support upFront.eZine through PayPal, then the suggested amounts are like these:
Or you can mail a cheque (US$ or CDN$ only, please) to upFront.eZine Publishing, Ltd., 34486 Donlyn Avenue, Abbotsford BC, V2S 4W7, Canada
Letters to the Editor
Re: State of European CAD
I read in some of your recent articles that Graebert has a “secret” agreement with DS restricting our R&D in 3D. I apologize if I ever gave you this impression. SiteMaster BIM is already a counter-example but the main reason why we don’t concentrate our R&D on 3D is different: Graebert is simply customer-driven and we see currently that we can bring more value by investing in Mobile and Cloud.
A student once asked me how Graebert could compete and survive against such large companies after so many years. My method is always the same:
Study the market
Invest soon enough in technologies before they become a must-have
In doing so we have something unique to offer, an opportunity to lead the market and have collaboration opportunities with industry leaders. It’s a head start.
I watch some of our competitors instead trying to catch up with mature technologies, such as 3D MCAD. The way I see it is that these segments of the market are already dominated by giants with infinite resources and years of R&D in advance. - Wilfried Graebert, ceo Graebert Gmbh, Germany
Re: Where SolidEdge is Heading
"...are 61% happier... " I've heard it claimed that 83.7% of people are fascinated by statistics. and that 92.4% of all statistics are made up on the spot. - Bill Fane
Following is the source/background information on the Net Promoter Score referenced in John Miller's presentation at SEU15: Net Promoter Score (NPS) is one of the most widely accepted ways to measure customer satisfaction. It asks the question "On a scale of 0-10, how likely are you recommend product/service to a friend or colleague. The overall Net Promoter Score for Solid Edge is very good. And the score among users who frequently use Synchronous Technology is even better -- in fact, it is 61% higher than the NPS of those who do not. - Kate Eby, public relations PLM Siemens
Another well-written issue! Your review of SEU 15 was very thorough and informative. Without your articles I don't think I'd even want to try to keep up with what's going on in the engineering and design technology space. - David Stein
I wanted to thank you for the mention in your latest issue and ask if you'd be willing to add a link to our site -- www.fra.me. We've had a lot of folks mention they can't easily find our site given the unique URL we're using. We'll be addressing that shortly, but it would be great if you can provide a direct link for now. - Justin Boitano, vp of marketing Fra.me