Reverse engineering tricks with lasers allow designers to resurrect old aircraft designs quickly and efficiently
Your father’s Oldsmobile may be long gone but his B-52 is still pulling missions, and they haven’t built the “BUFF” (Big Ugly Fat Fellow) since 1962. The last KC-135 tanker was built in 1965. Besides aging warbirds (the average plane in the US Air Force is over 28 years old) there are hundreds of ancient civilian airliners carrying friendlier payloads everyday. The key to doing this safely is of course excellent maintenance and periodic upgrades. Laser scanning plays an essential role.
Why Reverse Engineer?
Replacing a worn or damaged aircraft component quickly and efficiently often requires reverse engineering the part, especially on legacy aircraft. The original vendor may have gone out of business, or be unwilling or unable to supply a replacement at a reasonable price and lead time. In many cases, the documentation needed to make another part doesn’t exist.
Faced with this challenge, the aircraft operator must reverse engineer the part and assemble enough data in the right format to enable remanufacturing. And if his multimillion dollar plane is grounded, he has to do this quickly. Luckily today’s laser scanners and associated software offer exactly this capability, albeit with limitations.
How Laser Scanners Work
Laser scanners use one of two different approaches to define a part’s shape. One approach uses triangulation. The scanner sends a laser line or single point across an object and a sensor picks up the reflected laser light. Because the system knows the distance and angle between the laser source and the sensor very precisely, it can use trigonometric triangulation to calculate the distance from the scanner to the part. By building up millions of these measurements it creates a point cloud that defines the shape of the part. This approach is suitable for small-to-medium sized parts in situations in which the scanner can be positioned within about 1 m of the part.
For mid- to long-range scanning (over 2 m from the part), “time-of-flight” systems are the better solution. This technique measures the time it takes pulses of laser light to reflect back to the sensor. Since the speed of light is a known constant and since these systems can measure the time interval to within picoseconds, they can accurately calculate distances based on the length of the intervals. These systems often rotate the laser and sensor to capture up to a full 360° view of the area, further enhancing their ability to scan entire airframes or other large components.
Beyond the two basic approaches, there are variations in mounting and moving the laser scanner, systems that combine laser scanning with other techniques, different workflow and output options, and of course differing degrees of accuracy depending on the quality of the equipment. Either system can be highly accurate and both have inherent limitations.
Scan, Probe, or Both?
As Stephen Strand, applications engineer at API Services (Newport News, VA) said, “Choosing the right laser scanning solution depends on the work scope, both in terms of the size of the application and the accuracy requirements, plus factors like the surface type. For example, certain lasers won’t get any return on a chrome or mirror-like finish.”
In fact, most light-based systems have problems with highly reflective parts, like turbine blades. Strand said the usual work-around is to “spray a developer on the part to create a matte surface. But with aerospace components you’re often working with very tight tolerances, so the thickness of any coating may be a significant consideration in any measurement.”
In many such situations the best approach is to combine laser scanning with another technique. Steve Kersen, VP of sales and marketing for NVision (Southlake, TX) said, “Laser is often used to scan an entire part because it’s quick and gives engineers a fast way to make a 3D model. One of our bigger units scans with a 9″ [228.6-mm] stripe. You can capture many parts very, very quickly. We may then go back and touch-probe certain portions that need to be accurate within tenths, and adjust the model accordingly.”
Strand echoes this, explaining that, “A surface profile typically doesn’t have the same tolerance as a datum or a feature, so we’d typically use a laser to create a point cloud of the entire object and then use another method to measure specific areas like a hole, a bore, a planer face, or a turbine blade’s root form.”
Accurate to Within a Tenth…or Maybe Not…
That’s not to say that laser scanning isn’t accurate. Kersen said, “A good laser in a bridge-type CMM will be accurate into the tenths. A portable system in which the laser is mounted to an articulating arm in a movable CMM can be accurate to plus or minus a thou. So the quality of the laser and the method for orienting the laser are both critical factors.”
Kersen cautioned that, “There are some cheap $10,000 lasers coming out of Russia now that aren’t very accurate. Price tends to determine the accuracy. These systems are not very accurate while some of the better laser systems we and others sell cost over $100,000.”
There are also systems that use laser-tracking technology to accurately locate a hand-held probe in space, giving the user tremendous freedom in measuring features best suited to tactile measurement. For example, API Services’ parent company in Rockville, MD, Automated Precision Inc., makes a wireless hand-held probe called the vProbe that interfaces with their Omnitrac2 laser trackers. The volumetric accuracy of the Omnitrac2 system is ±0.0016″ (40 µm) within a 5-m scanning range, while the probe can deliver 3D point accuracy as low as ±0.002″ (55 µm).
Strand explained that, “The tracker shoots the laser to the embedded prism on the probe. There are no attached, articulating pieces. In order to orient itself in space, the probe has sensors that reference a gravity frame measured by the tracker. The prism in the hand-held unit can we swiveled so that you can probe an area that might be hidden from your line of sight as long as the tracker can still see the embedded prism.”
Quick Work in the Real World
Kersen also said that with the possible exception of jobs on top of an airplane, a portable arm-based system can handle most of the scans required in field work. “We just had a rush job for Southwest Airlines. What the industry refers to as ‘airplane on the ground,’ meaning the plane is grounded until the problem is fixed. A bird had hit the wing. They called at three o’clock. We had engineers on the scene at five. By then the airline had removed the sheetmetal covering the wing so they could get to the strut and cut out the damaged part. Our engineer rolled up to the wing with a portable CMM laser scanner, mounted on an extendable tripod. He scanned the area and created a model of what’s called the ‘bathtub,’ an insert that would provide the required strength in the damaged section of the wing. We converted that to an STL file and then an Inventor file by eleven that night. The airline then cut a new part and had the plane flying the next day.”
The Corpus Christi Army Depot (CCAD) credits its NVision arm-based laser scanner with cutting the time it takes them to reverse-engineer helicopter components from two weeks to two hours. That should make the taxpayer happy, because CCAD is the largest facility in the world providing overhaul, repair, modification, recapitalization, retrofit, testing and modernization of engines and components for rotary wing aircraft. They’re keeping 30–40 year-old helicopters in service, many of which were designed without CAD and lack even blueprint documentation.
Like the system NVision used at Southwest Airline, CCAD’s laser is easily moved by hand, but mounted on an articulating arm that enhances the accuracy. Vernon Hull, machine shop supervisor said it, “…moves about the object, freeing the user to capture data rapidly and with a high degree of resolution.” Whereas it took days to tediously collect tens of thousands of points to define complex surfaces with their CMM, the NVision scanner “collects millions of points on the aircraft in only 30 minutes…We can now capture thousands of points every second, making it possible to more accurately define the part surface in a matter of hours. The net result is that we can get aircraft back into service faster.”
To CAD or not to CAD
These case studies are a good example of both the speed with which a laser scanner can get you the 3D model you need, but also the necessity of having the right software and the skills to use it. It also points to the question: What’s the ultimate goal? Strand says one of the most common customer misconceptions is the belief that they need a CAD model.
“Too many assume that CAD is what makes everything work. But there are many possible deliverables that can be manipulated to meet various needs. An STL file is the most important example, and they’re much easier to create than a CAD. For example, an STL file is all you need for prototyping and 3D printing. You can also do inspection from an STL if you have the right software. And an STL is a more direct representation of the part, because when you go from a point cloud to a CAD you have to do some interpretation. When you’re just triangulating points, you’re just one step away from raw data. The only disadvantage of an STL versus a CAD is the file size, since a CAD file is generally much smaller than the STL for the same object, depending on the level of detail.”
Naturally the relative ease of creating an STL results in much lower costs. Strand said, “Most scanners will scan straight to an STL. So if you go that route you’re already at your deliverable after a little clean up. Depending on your scan time, clean up is a tenth to a fifth of the time, depending on complexity. But if you need to create a CAD file, you would need anywhere from scan time to four times scan time. All those hours cost money. You also need a trained operator to do the work of going from STL to CAD. The software doesn’t do it automatically.”
Strand said when people think they need to go from point cloud to CAD, or point cloud to CAD with the addition of 2D drawings, they sometimes consider hiring “Joe Blow with a laser scanner in his trunk” in order to cut their costs. “But if you just explain the functionalities of the different deliverables, you can get them what they want for much less money.”
Three Votes for Going All the Way to CAD
On the other hand, there are knowledgeable users who do need to create a CAD file, and that may mean changing the measurement mix. The Naval Air Systems Command (NAVAIR), in Patuxent River, MD, runs a “reverse engineering center of excellence” for the express purpose of keeping legacy aircraft flying. NAVAIR’s Danny Campbell said it usesseveral arm-based laser scanners (a Faro Edge 9′ [2.7-m] arm with a Faro ES Laser Line Probe and a Faro Titanium 8′ [2.4-m] arm). But he prefers using the touch probe when possible because the measured data is almost immediately usable in their CAD software. “The laser is really only used for complex or convoluted surfaces that don’t lend themselves well to touch-probing.”
They use PolyWorks or Geomagic software to process their scanned point clouds into usable CAD data. Campbell takes advantage of any available drawings as they sometimes provide design intent and critical dimensions, though he also says it’s common for airfoil drawings to have insufficient detail. Recent projects have included critical safety items on the AV-8B Harrier jump jet and engine ductwork on the V-22 Osprey and Bell AH-1Z Viper attack helicopter. The latter two included computational fluid dynamics, requiring both laser scanning and translations to CAD files.
The Corpus Christi Army Depot also typically creates CAD files, taking the scanned point cloud into their XOR software and then converting it to either an IGES/STEP Parasolid or a fully parametric model with history tree. They then check and tweak the resulting 3D CAD model in SolidWorks and finally go to a CAM package for the machining program.
Strand said API is getting lots of requests to use laser scanning to reverse engineer electronic components, in part because they’re harder to measure using other methods. His first such project was for an aircraft circuit board and it’s another good example of when it makes sense to employ CAD.
“The critical factor in creating a replacement part was to be able to define the connection points and integrate the replacement with newer parts. The customer needed to change the layout a bit and reverse engineering helped make sure everything would fit with existing components. We captured the data using Polyworks and exported into DesignX to clean up the mesh and create the CAD. We modeled all the components separately. That way if we have any future requests with the same components we can use the CAD from our catalog, instead of wasting time rescanning and cleaning up the data and modeling over again.”
In-House or Outsource?
If you’re still a bit bewildered by the options and trade-offs in this area, take comfort in knowing that you can hire a company like NVision or API to come in to scan you parts and deliver an STL, CAD file, or another format on demand. If you’re ready to take the leap into doing it yourself, you can hire them to study your needs and deliver training and a complete turnkey solution. Or you can go straight to vendors like API, Wenzel America (Wixom, MI), and Basis Software Inc. (Redmond, WA) and assemble your own system. Just don’t stand too close to the B-52.
The original article can be found here.