Published by DELMIA - 05 June 2007

It is axiomatic that no large manufacturer undertakes any significant expansion without extensively simulating the proposed new facilities. How else to be certain the facilities will work as planned, that their cost was money well spent? At Lockheed Martin Corp.’s Aeronautics unit in Fort Worth, Texas, the F-35 fighter aircraft is rapidly moving into early production glitch-free, thanks to hundreds of simulations that cover every conceivable aspect of production and maintenance.

A major focus of simulations focus in Fort Worth has been the facilities, systems and methods that will apply coatings to the F-35s.

• Seven bays where surfaces will be prepared and coatings applied. Five bays are the size of barns—75 by 35 feet with 35-foot-high ceilings. Two bays for the fully assembled wing are even bigger.

• Robotic systems that apply the coatings use two customized material-handling robots and two customized positioners.

• The coating methods, off-line programming (OLP) and verification techniques, some of which are also still under development.


These simulations are being done with three DELMIA packages—ENVISION™ for facilities, IGRIP™ for robotic systems and UltraPAINT™ for the coatings. Based in Auburn Hills, Michigan, DELMIA is the Digital Manufacturing brand of Dassault Systèmes.

The bays and the equally massive spaces for getting aircraft in and out were developed primarily with ENVISION. Two even larger bays for coating fully assembled F-35s were designed to accommodate full-rate production.

IGRIP was used to develop the robot systems, which are material-handling robots with special coating tools on their end-effecters (or wrists) and not ordinary paint-sprayers. Paint sprayers were rejected as too imprecise and unpredictable.

UltraPAINT was used to develop the OLP systems that drive the robots, two per bay in two barn-sized bays. Also under OLP control are custom-built positioners let the robots reach most any point in the bays.

The huge F-35 requirement for simulations is due to the massive scope of the program: an estimated $290 billion (a Department of Defense record) for up to 4,000 aircraft to be built at the rate of one-a-day. Customers include the U.S. Air Force, Navy and Marine Corps and several foreign countries. With business and technical risks so huge, nothing has been left to chance.

The F-35 is being built in what was already one of the world’s largest factories, the U.S. Air Force Plant No. 4 in west Fort Worth. It was built during World War II for B-24 Liberator bombers. Its mile-long production floor has been in continuous operation ever since. Originally known as the Joint Strike Fighter, the F-35 was officially christened Lightning II.


CHALLENGE: It’s All About The Time Span

Arguably the F-35s could not be coated with existing best practices in anything close to the one-a-day rate without huge expenditures for additional bays. Such an investment could easily cost Lockheed Martin millions of dollars and production schedules could still be missed. “For us in coating, it’s all about the time span, the amount of time each fully assembled aircraft spends in here,” said Randy W. Scroggins, manager, F-35 final finishes.

The highly engineered coatings have a variety of critical functions:

• Radar evasion, or Stealth. The F-35 is the first supersonic multi-role radar-evading aircraft but specifics are classified.

• Abrasion resistance on the nose and the leading edges of the wings, elevators and twin tails. The supersonic impacts of millions of water droplets can leave leading-edge coatings looking sandblasted, Lockheed Martin said.

• High temperature resistance at the fuselage rear and F-35 turbine exhaust.

• Three dimensionally different versions of the F-35 whose coatings vary with intended missions.

Among the many factors that have to be optimized and balanced with simulations are constantly varying application rates—speeds, feeds and flows. Scroggins and Application Engineer Jason LeFever have to deal with:

• Differing coatings on specific areas of the aircraft.

• Applying multiple coatings, some requiring many passes.

• Variations in coating thicknesses over aircraft surfaces, and thicknesses that ramp up and down in transition areas.

• The tendency of liquids on concave surfaces to rise slightly around the edges (the meniscus); the slightly concave surface must be accommodated in the coatings programs.

• Calculating the “plume” of each coating as it is applied. During coating, air constantly moves through the bays.

Add to this the geometric complications. Almost no surface on the F-35 is flat. Many of the components to be coated are saddle-shaped—compound contours—and some are U-shaped. This means no easy OLP solutions such as constant-rate programming can be used.

All these geometric considerations add complications because the coating process requires a very short “stand-off” from the working surface. That increases the risk of dents and dings in the aircraft surfaces, which are unforgivable. Any contact would require the part surface be scrutinized to ensure the aerodynamics is unaffected. The underlying airframe structure also would have to be painstakingly inspected to ensure against any possible hidden damage.

The latter might require partial disassembly. Inspections could take from half a day to a full week and could cause a traffic jam in finishing. “We must not in any way affect the aircraft’s OML” or outer mold line, which determines aeronautical capabilities,” said Scroggins. “Our need for precision is critical. With the OLP software, we get very tight control of very precise hardware.”

Clearly what Scroggins, LeFever and Lockheed Martin are doing with the F-35 has almost nothing in common with paint spraying. They almost never utter the words “paint” and “spray.”

SOLUTION: Hundreds of DELMIA Simulations

To make sure they had the robotics aspects of F-35 coating properly worked out, Scroggins and LeFever created more than a hundred simulations and analyses. And they were able to do so in about one-fifth the time and effort that would have been needed just a few years ago with earlier releases of the DELMIA V5 suite of modeling and simulation tools. The F-35 was designed in CATIA V5, also from Dassault Systèmes.

For the systems aspects of their work, Scroggins’ and LeFever’s approached the coating process as if working with machine tools. They are using Fujitsu-Fanuc Model R-2000s, six-axis material-handling robots with extended eight-foot reaches.

Built to the mechanical stiffness of a machine tool, the R-2000s were partly customized with IGRIP simulations. Stiffness equals repeatability in a robot, which in turn ensures that coating programs will always achieve the identical, precise results. “You can shake and wiggle a spray-painter robot,” Scroggins observed, “but you can’t make an R-2000 move. You can tug and push on it all you want.”

Nevertheless, reach presented some mechanical issues. First, to maintain the necessary precision application paths of the coating tools, Scroggins and LeFever stay well inside the maximum reaches of the R-2000s. Despite their stiffness their arms are still subject to the pull of gravity, enough to affect OLP calculations for standoff distance and coating consistency.

Because of the huge size of the bays, the robots themselves had to be mobile. Lockheed Martin purchased a pair of positioners that have 65-foot horizontal travels along the bays’ side walls and 15-foot lifts. An R-2000 robot weighing several hundred pounds is mounted on each. The positioners were built by systems integrator CTA Inc.,* which specializes in large robotic coating systems.

Simulations with IGRIP ensured that the robots and the end-of-arm tooling would not hit the bays’ walls or ceilings (studded with fragile sprinkler heads) or collide with each other. IGRIP’s Realistic Robot Simulation (RRS) module links to Fanuc’s Robot Control Software (RCS) that deals with speeds and feeds.

Another series of simulations developed the “dress” for the bulky flexible tubes that deliver coating materials to the ends of the robot arms. Other simulations ensured sufficient access for maintenance people and even forklifts beside and behind the CTA systems.

Some of the most demanding IGRIP simulations, LeFever said, were studies of robots’ reach with the coating tools and optimized part orientations and placements within the cells. These simulations accommodate arc-like travel in any two planes (or all three) plus roll, pitch and yaw of the R-2000 wrists.

LeFever’s work with RRS and RCS also prevents robot tilt heads from “flipping” in midair. Flipping occurs when a robot axis needing to move one degree, clockwise for example, instead goes the other way, 359 degrees counterclockwise. Given the short standoffs Lockheed Martin is using and the size of coating tools—as big as a breadbox—a flip could do significant damage.

“By using IGRIP, we can now verify offline that all of the robot motions are correct before we implement our processes physically, on aircraft parts,” Scroggins said.


RESULTS: Physical Facilities Speak For Themselves

Asked about results of using DELMIA software, Scroggins gestured at the cavernous facility replied: “Everything you see here in the finishing facility.” The bays and the huge vestibule and corridors were sized to move fully assembled F-35s in and out quickly—and without bumping each other. ENVISION was used to reverse-engineer the F-35’s largest components.

• Bay width was based on the aft fuselage, the F-35’s bulkiest component.

• Bay length and height were determined by the 35-foot wingspan of the aircraft-carrier version, the largest component by surface area as well as the aircraft’s longest dimension. (Wings are to be coated in the vertical position.)

Similar dimensional considerations verified there is sufficient space around the scaffolding and ladders in the surface preparation bay was sufficiently roomy. F-35 components shipped to Fort Worth from prime subcontractors and business partners require sanding and cleaning of protective finishes. “Ninety-five percent of a good coating job is good surface prep,” Scroggins observed.

There are three smaller bays for prepping and coating small and medium-sized components such as landing-gear and weapons-bay doors, access panels and covers, rear elevator wings and the F-35’s twin tails (rudders).

Nothing is left to chance. “We want to know all the feedback so we understand what the system is doing and how to gain even better control over the process,” Scroggins said. “Without this, OLP will not work very well. We are finding tons of little things you never would have suspected,” and they are going into a programmer guidebook.

BENEFITS: Driving Down The Cost Of Quality

Good coatings on supersonic Stealth aircraft are all about thickness, and the capability of predicting it in OLP simulations. While using DELMIA UltraPAINT for coating OLPs, Scroggins and LeFever found a huge opportunity for productivity gains through statistical process control (SPC). The simulations’ SPC data helped reduce or eliminate the need for inspecting every one of the F-35’s coated surfaces—hundreds of them, perhaps several times.

“With SPC built on UltraPAINT, we are trying very hard to eliminate the need to inspect coating thickness at every one-foot interval,” Scroggins explained. “Doing those inspections manually would take two people two days per aircraft, plus a lot of costly special equipment. We have neither the time nor the space for anything like that,” he added.

So to Scroggins, LeFever and Lockheed Martin, coating F-35s is far more than a matter of time. It is also a cost-of-quality issue with process consistency the key to success. “We intend to predict coating thickness and that will allow us to operate flawlessly,” Scroggins said. “With some help from DELMIA R&D, we are on the verge of getting that working really well. DELMIA has also written us macros for special functionality.”

Not only does every process have to be simulated, Scroggins pointed out, “but every process has to be verified, too. UltraPAINT gives us the intricate, very detailed, and automatically generated NC programs for the robots that we need,” he added.

DELMIA Senior Robotics Consultant Jay Johnson pointed out that “the programmers will be able to do an entire simulation, including multiple robots, fixtures, and material handing devices by using a simple graphical programming interface.” His work at Lockheed Martin included refinements to line tracking, where a stationary robot coats parts as they move continuously through the bays, and rail tracking where the robots on the CTA positioners follow the parts through the bay.

“This gives Lockheed Martin real-time checking of reachability as a part moves in front of the robot,” Johnson said. “This is a key advantage of simulation. The simulations help make decisions on how soon after a part enters a bay can coating start and on the order in which surfaces are coated.” For the latter, he explained, “the issue is which areas of a moving part should be painted first in order to leave until last the surfaces most reachable when the part has past the robot.”

In addition, UltraPAINT simulations help:

• Calculate how far each application pass must extend beyond the edge of a workpiece to ensure the surface is completely covered while minimizing waste. Some coatings are very expensive.

• Predict the thickness of the coating buildup on overlapped surfaces between passes. Buildups add to weight, which affects aerodynamics.

• Determine coating viscosity and drying time, which require tracking (and simulating) temperature, humidity and rates of change plus in-bay air flows as high as 200 cubic feet a minute.

“Ultimately,” Scroggins said, “we want the programmer to be able to look at the automatically generated OLP toolpath from UltraPAINT and apply their coatings expertise and robotics skills to improve the programs. We want to remove the repetitive tasks that programmers usually have endure and that take so much time, even though with UltraPAINT we can program in a couple of hours what took two weeks in our legacy programs.”

He added that “we also must be able to accommodate change,” he added, “especially the things we cannot yet foresee. We know we will get new bumps in the plane’s exterior, as for new radars. Some of the foreign customers may need changes, too. All those will impact the way we program. And they will have to be simulated all over again.”