A Decade Later
More than 10 years after designing and installing a custom robotic system that uses plastic media to strip paint from aircraft, SwRI engineers successfully revived and completed a major upgrade and retrofit program to help improve worker safety, reduce maintenance costs, and increase efficiency.
The F-15 aircraft robotic depaint system at the Warner Robins Air Logistics Center (ALC), Georgia, used the first large-scale custom robots developed by Southwest Research Institute (SwRI). The technically challenging program raised fundamental questions about the use of robotics for this application. Robotics can provide elegant solutions to complex problems, but what is the useful life of advanced robotics and automated systems? How will the assumptions and design decisions hold up to industrial use? How robust is the system design, and can enhancements be implemented? This article highlights only a few of the technical issues related to extensive retrofits and upgrades that led to the answers for the SwRI team.
The comprehensive retrofit of the robotic depaint system (RDS) and facility at Warner Robins recently concluded -- nearly a decade after the original concept and design were completed. Literally all elements of the facility, depaint process, and RDS were tested, evaluated, replaced, or retrofitted.
A Look at the Past
In April 1986, the SwRI Automation and Data Systems Division was awarded a U.S. Air Force contract to develop automated paint removal cells to depaint F-15 aircraft at Warner Robins and F-16 aircraft at Hill Air Force Base (AFB), Utah. Both systems were to use a plastic media blasting (PMB) process similar to sandblasting.
After the Warner Robins robots were completed in August 1990, but before the PMB processing equipment was put in place, the Air Force chose to discontinue the PMB process to instead investigate an experimental carbon dioxide (CO2) blasting process for the removal of paint. Development of the SwRI robotics-PMB system at Hill AFB continued, and it subsequently began operation in May 1991.
At the Warner Robins RDS and facility, a third-party contractor converted the PMB process to the CO2 process. The CO2 process was thought to have been ideal for the environment because after the frozen CO2 (dry ice) pellets stripped the paint, they were converted to harmless gases, eliminating the need for media reclamation. Three years after development began, however, the CO2 program was terminated because the process and its associated equipment proved unreliable and difficult to maintain. The mean time between failures for the CO2 system was approximately eight hours. In addition, it was found that the CO2 process could damage aluminum and composite aircraft skin.
In February 1997, SwRI was awarded a new Air Force contract to investigate a re-conversion and upgrade to the now-modified RDS and facility at Warner Robins. The challenge was to mitigate technical risks for the client while achieving performance goals. In the first phase, a team of SwRI engineers, called the "Phoenix team," conducted an engineering evaluation and overall assessment of the robots and supporting systems. The second phase included all elements of design, fabrication, component testing, and installation, and the third phase included facility and RDS system testing and shakedown.
Paint removal is a necessary part of aircraft maintenance to allow surface inspection, to perform repairs, and to keep an aircraft's weight at acceptable levels after several coats of paint have been applied. Paint removal can be achieved using a variety of methods. Two of the most predominant means are chemical stripping and PMB.
In chemical stripping, solvents soften the organic coatings, then scrubbing and rinsing remove the softened paint. In the PMB process, plastic particles are sprayed at the aircraft surface under high pressure to abrade the coating and remove it. Manual blasting can take a team of six to eight blasters 24 hours or longer to finish a job.
Though effective, chemical stripping generates several thousand gallons of toxic waste for each aircraft cleaned, and disposal becomes prohibitively expensive. In the pressurized PMB process, on the other hand, less than 10 percent of the media used during stripping is broken down to waste dust containing the hazardous material. This spent media is recovered and separated from usable media, then recycled. In addition, a robotic system requires two to three operators, provides a safer working environment, removes paint more consistently, and takes less time to depaint an aircraft.
RDS Evaluations and Findings
The Phoenix team first examined the facility and the modified robotics system to assess functionality and identify recoverable systems. The RDS had been in storage for almost three years. During that time, the electronics cabinets had not been pressurized or cooled, which led to concerns about the integrity of the power and electronics busses. Later, significant changes were found to have been made to the controller and control systems, including motor-drive amp replacement and wiring changes to nearly all the subsystems. This led to concerns that axis loads may have been too great. In addition, the robot controllers -- the "intelligence" -- had been replaced and the software code rewritten.
This examination helped define the state of the RDS relative to the original system. SwRI's software, hardware, and engineering support of the robotic depaint operations at Hill AFB, and familiarity with the manual depaint processes at Warner Robins, also proved to be valuable in reviving the system. Trade-off studies were conducted, including full versus partial PMB floor recovery, ventilation modifications, robot controller upgrades, and software modifications. The production goals helped identify the functional subsystems.
Mechanically, the robots were still sound. Electrically, the system was not balanced and the drive amplifiers were marginal, but they too were deemed acceptable. The robot controllers, however, raised concerns related to hardware supportability, obsolescence, and outdated software. In the original PMB configuration, SwRI engineers used Modicon® 5200 controllers and source code. During CO2 modifications the Modicon controllers were replaced with Adept® controllers and their source code, which introduced software trade-offs for the PMB process.
The robot's electronics, specifically the resolver-to-encoder cards as they interface with the controller, had been modified and were operated only intermittently. Other modifications were made to the ventilation and blast media reclamation and separation equipment. Overall, however, the software issues remained the greatest technical challenge.
Because of the complexity of the rework, the team established a means of evaluating the tasks. "Critical" was considered the highest level of significance. That is, without the completion of the critical recommendations, the RDS would not function. "Important" was the next level of significance. Without completing these recommendations, the RDS may not have met Air Force quality standards, reliability and maintainability (R&M) goals, or production goals. "Preferred" recommendations provided advantages that enhanced the overall R&M or production rates.
The control software written for the CO2 process was marginally functional for the PMB process. Given the schedule and cost constraints, and to minimize technical risk, a decision was made to build on this software. This build would include creating all the path files -- that is, the means of instructing the robot to depaint the aircraft in a particular pattern. These path files must be recorded accurately with respect to a master reference point file. However, during the second phase, the software was found to be unusable. For example, the robot base frame and the robot tool frame were incorrect. These frames are coordinate systems that provide a reference for robot motion and knowledge of the relative placement of the aircraft with respect to the robots. As originally designed and installed, the serial manipulator of each robot had seven joints or degrees of freedom (DOF) that moved in coordinated motion. The coordinated motion was lost in the PMB-to-CO2 conversion and had to be recovered.
Software problems alone might have terminated the program, but the team formulated a plan. The increase in costs associated with a complete software system build would have to be met with equal cost savings elsewhere. Potential savings were possible by recovering the original SwRI path files from 1991. This required additional functionality, which resulted in additional software tasks -- and technical risk.
The mechanical design proved to be robust, allowing a mechanical rework that included fitting the robots with collision switches for operator and equipment protection, tuning the motors and amplifiers for the PMB process, repairing door seals, and installing actuator limit switches. The robot end-effector design was resurrected, fabricated, and installed.
What was not evident in the initial evaluation was the degree the system was used or the payloads, end-effector velocities, and operational environment. The system shakedown revealed some loading extremes. Three cycloidal drives, which enable the robot arms to move, failed unexpectedly. In the first phase, the team determined that the axis drive, which allowed wrist-like movement of the media nozzles, had been resized from 500 inch pounds (in-lbs) to 1,000 in-lbs to accommodate heavier CO2 equipment loads. This change may have contributed to the failures.
Subsequent investigation revealed that, during initial cryogenic testing, the wrist joints froze because of conductive cooling as ice condensed on the robot wrist and end-effector. Complicating that problem, the facility's air conditioning system dehumidified the air. The ambient air dew point dropped, resulting in electric arcing between components. The arcing may have contributed to the reduced life of the motor drive amps, but team members believe the motor sizing mismatch had a greater effect. During the system shakedown, several drive amps failed. Investigation showed other factors led to the failures, such as unbalanced transformer loads and electrical noise. New digital drive amps, correctly matched to the motors and loads, replaced the drive amps.
Resurrecting Bay 4
The depaint facility at Warner Robins has five bays at which an aircraft is washed and prepared, depainted, primed, and repainted for service. Bay 4, used for the actual depainting, had been significantly modified and required ventilation, blast media reclamation and separation equipment, and additional compressed air.
Bay 4 was initially to be a stand-alone depaint facility. However, after the conversion, the facility did not meet the National Electric Codes Division 2/Class 2 standards necessary to prevent explosions caused by electric sources in dusty environments, nor did it meet Occupational Safety and Health Administration ventilation or visibility requirements for the depainters. In addition, a media recovery system did not exist. These issues were critical to the depaint operation.
The CO2 contractor had also filled the existing floor recovery foundation -- previously a 32 x 36 x 1.5-foot-deep pit -- with concrete. Several factors complicated the foundation's repair, the most tedious being that the work had to be conducted within inches of three 26,000-pound robots.
The team believed the floor recovery system to be pivotal to the PMB process. About 6 pounds per minute of media flow from each of the robot's three blast nozzles, requiring about 38,880 pounds of media to depaint an average F-15 aircraft. Without an automated recovery system, the Bay 4 facility could not be used. Fortunately, the removal of about 62 tons of concrete did minimal damage to the floor of the recovery system, and major rework was avoided.
A Success Story
After two years of retrofitting, the Warner Robins ALC robotic depaint system and facility are fully operational and have depainted more than two dozen F-15 aircraft. The electromechanical system, controller, and software have been upgraded to the latest industrial robot technology available, and the electrical system and electronics have been updated and retrofitted.
Overall, the robotic depaint system provides the Air Force with enhanced depaint capabilities with no additional staff requirements. The system requires two or three robot operators to depaint an aircraft, compared to six to eight operators for manual blasting. The rebuilt RDS has depainted aircraft in as little as 12 hours, with an average of about 16 hours, compared to manual depaint in about 24 hours. In addition, RDS operators are removed from the hazardous environment, and depaint quality remains consistent.
The team found that robotics provide practical solutions to complex problems and that the useful life of advanced robotics can be 10 years or more. The questions surrounding design assumptions and decisions can be answered by looking at the systems on a case-by-case basis.
The robotic systems at Hill AFB have been in operation for nearly 10 years. At Warner Robins, SwRI had an opportunity to make significant design changes where warranted. However, in the case of the mechanical systems, when operated within intended limits, these components held up to industrial use. Electrically, the changes to the original design affected the system operability, and since returning to the original design, the electrical subsystem is sound. Electronically, some subsystems that were analog in nature have migrated to their digital equivalents, while providing the same functionality. In the area of software, robot controllers provide enhanced capabilities that were not available 10 years ago. In summary, the robots have demonstrated their robustness.
More than 40 major tasks were completed to meet the RDS and process goals, including all of the "critical" and "important" items and most of the "preferred" items -- expanding capabilities beyond the initial goals. The SwRI Phoenix team anticipates that the Warner Robins ALC robotic depaint system will continue to support the reliability, safety, and maintainability of the Air Force depaint operation for many more decades to come.
Published in the Spring 2000 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Maria Stothoff.