M. D. Fowley and J. R. Brault
Westinghouse Savannah River Company
Aiken, SC 29808

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy.

Available for sale to the public, in paper, from:  U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161,  phone: (800) 553-6847,  fax: (703) 605-6900,  email:  orders@ntis.fedworld.gov   online ordering:  http://www.ntis.gov/support/ordering.htm

Available electronically at  http://www.osti.gov/bridge/

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062,  phone: (865 ) 576-8401,  fax: (865) 576-5728,  email:  reports@adonis.osti.gov

It has been decided that excess, weapons-grade plutonium shall be immobilized to prevent nuclear proliferation. The method of immobilization is to encapsulate the plutonium in a ceramic puck, roughly the size of a hockey puck, using a sintering process. This method has been officially identified as the Plutonium Immobilization Process (PIP). A Can-in-Canister storage method will be used to further immobilize the plutonium. The Can-in-Canister method uses the existing design of a Defense Waste Processing Facility (DWPF) canister to house the plutonium pucks. The process begins with several pucks being stacked in a stainless steel can. Several of the stainless steel cans are stacked in a cage-like magazine. Several of the magazines are then placed in a DWPF canister. The DWPF canister is then filled with molten glass containing high-level, radioactive waste from the DWPF vitrification process. The Can-in-Canister method makes reclamation of plutonium from the pucks technically difficult and highly undesirable.

The mechanical requirements of the Can-in-Canister process, in conjunction with the amount of time required to immobilize the vast quantities of weapons-grade plutonium, will expose personnel to unnecessarily high levels of radiation if the processes were completed manually, in glove boxes. Therefore, automated equipment is designed into the process to reduce or eliminate personnel exposure. Robots are used whenever the automated handling operations become complicated. There are two such operations in the initial stages of the Can-in-Canister process, which required a six-axis robot. The first operation is a press unloading process. The second operation is a tray transfer process.

The first operation is prior to the sintering ovens. Plutonium, in the form of plutonium oxide, and ceramic precursors are blended and fed into a press that produces the puck. The unsintered puck is white in color (due to the precursors) but is called a "green" puck (see figure 1). The six-axis robot is required to remove a puck from the press, move it to the inspection area, and place it on a furnace tray for the sintering process. The furnace tray will hold an array pucks, therefore it is necessary for the robot to palletize the pucks on the tray. The size and shape of the array will be determined through furnace testing with the tray and green pucks. The inspection area contains three separate devices: a platform scale for weight analysis, an array of lasers for dimensional analysis, and a digital camera for topological analysis. The puck will be palletized if it passes all of the inspection stages. If not, it will be placed in a bin for recycling.

The second operation, the tray transfer process, is similar to the first operation and is located down the production line from the sintering ovens. Another six-axis robot is required to remove the pucks from the furnace tray, move them to an inspection area, and palletize them onto a stainless steel transfer tray. The inspection area is very similar to that of the first operation. A puck will be palletized if it passes all of the inspection stages. If not, it will be placed in a bin for recycling. The transfer tray will be used to transport pucks to the next workstation in the facility. A different tray is used to transfer the puck to increase the operational life of the furnace tray. Frequent handling by, or interfacing with, automated equipment can diminish the operational life of the ceramic furnace tray.

To successfully accomplish the operational tasks described in the two operations above, the end-effector of the robot must be versatile, lightweight, and rugged. Versatile to interface with the numerous equipment. Lightweight to retain maximum capacity of the robot. Rugged to endure the dusty and radioactive environment. As a result of these demands, an extensive development process was undertaken to design the optimum end-effector for these puck-handling operations.

As an overall requirement, it was desired to keep the design of the robot end-effector as simple as possible. There were pros and cons for either type of actuation method (pneumatic or electric). But, pneumatic actuation was chosen for its simplicity and durability in a radioactive environment. For simplicity, the number of end-effector sensors would be kept to a minimum. However, sensor feedback was necessary for the delicate movements around the furnace tray and for detection of process interruptions.

It was determined early in the design process that at least two different types of end-effectors would be required for each of the operations. Therefore, a tool changer was incorporated into the end-effector design. The tool changer would also provide for simple end-effector maintenance when used in the PIP process.

A two-jaw, angular gripper was chosen early in the developmental process to be the primary gripper for each operation (see figure 2). The two-jaw configuration was chosen for two reasons. One, the puck could only be approached from the side while it was in the press. Two, the top and bottom surfaces of the puck were required to be free of obstruction when the puck was presented to the digital camera by the robot. The two-jaw configuration met both of these requirements.

Angular jaw movement of the gripper was chosen for durability and a greater range of travel. The environment around the press was expected to be dusty due to the ceramic precursors. Ceramic powder collecting on the exposed jaw surfaces could work itself into the jaw mechanism and cause damage. In most case, an angular jaw had less of an exposed jaw surface than a parallel jaw of similar travel. Therefore, an angular jaw configuration was thought to have greater durability for the puck-handling operations.

To firmly grasp a puck, the fingers of the gripper will need to completely envelop the puck perimeter. An angular gripper, with a jaw travel of 15° or greater, could be fitted with fingers that completely envelop the puck and permit a sideward approach. Parallel jaw grippers could not be fitted with fingers that completely enveloped the puck and permitted a sideward approach. It was later determined, when more of the interface components were specified, that a large amount of travel was undesirable. The angular jaw movement was still retained due to the durability issues. However, the redesigned fingers would not completely envelop the puck perimeter.

A two-jaw gripper will be used exclusively for the press unloading operation. However, in addition to a two-jaw gripper, a three-jaw gripper will be required for the tray transfer operation (see figure 2). The two-jaw gripper is necessary for the inspection area, requiring a tool change in the middle of the process. The three-jaw gripper is necessary to remove the sintered pucks from the furnace tray. During the sintering process the pucks will shrink (diameter and thickness). There is a possibly that the pucks will shift from their original position as they shrink. The robot will have to know the exact location (in X, Y, and Z coordinates, relative to the robot origin) of each puck to successfully acquire and transfer them. One method to determine the new location of the pucks is with a digital camera other visual means. This method is used extensively in the automated industry. However, to keep with the dictum of simplicity, a three-jaw gripper with adequate travel is used instead of a visual system. When a puck shrinks, it will shrink within the original footprint of the unsintered puck. A three-jaw gripper, with adequate travel, could envelop the footprint of the unsintered puck and successfully re-acquire the shifted, sintered puck when the jaws are closed.

Prior to acquisition of a six-axis robot, a very simple pick and place machine was constructed to test the dynamic handling requirements of the end-effector (see figure 3). The Pick and Place machine was built from spare linear and rotary actuators and pneumatic valves. It could be fitted with a two-jaw gripper or a three-jaw gripper. A PC, using National Instruments control software to sequence the pneumatic valves, controlled the pick and place machine. The pick and place machine provided valuable information relating to grip force, finger design, flow control valves, and much more.

Two criteria were initially established for the design of the gripper fingers. The first was that the puck had to be firmly held. The relative position of the puck top and bottom surfaces needed to be well known throughout the press unloading and tray transfer operations (for gripping, inspection and palletization). The second was that the gripping and puck handling process needed to be delicate and graceful, to prevent chipping, cracking, and other unnecessary damage to the puck.

Grip force is controlled by the amount of air pressure supplied to the gripper. The greater the air pressure, the greater the grip force and the firmer the grip on the puck. Unfortunately, greater air pressure creates a harsher impact on the puck when the gripper fingers snap closed. At this stage in the development process the gripper fingers were made from aluminum for manufacturing ease and light weight.

Flow control valves were first used to offset the impact and retain the firmness of grip. The flow control valves restricted the flow of air so that the pressure built up gradually. With flow control valves, the fingers would softly impact the puck, then firmly grip the puck as the pressure built. The flow control valves lessened the impact on the pucks but required extra time for the pressure to build. The extra time only amounted to seconds but, as the overall immobilization process developed, a shorter cycle time was required to meet production schedule. The shorter cycle time necessitated the elimination of the flow control valves in favor of a quicker method of lessening the gripper finger impact on the pucks.

Next, finger pads were used to reduce the impact on the puck. Various hard and soft elastomers were tried as finger pad material. The softer material proved best for lessening the impact. As a bonus, the elastomer pad provided a firmer grip due to the increase in friction between the pad and the puck. The increase in firmness due to the increase in friction allowed a reduction in air pressure, which further reduced the impact of the fingers on the puck. Finger pads proved to be the best solution for the harsh impact problem.

Fingers, such as the ones shown in figure 2, were initially designed and tested with surrogate green and sintered pucks. These circular fingers conformed to the puck perimeter and distributed the grip force over a large area, which in theory would provide a very firm grip. However, the first surrogate pucks (made from plastic pipe) were not made to exacting standards. The diameter varied slightly from puck to puck. The pucks with diameters different than that of the gripper fingers did not make uniform contact. These pucks made multiple-point contact, which would concentrate the gripping force and increased the likelihood of chipping or cracking. In some cases, there was only two-point contact. The two-point contact caused an uneven load between the fingers and created instability when the puck was moved. These initial results, although caused by poor surrogate pucks, were unfavorable and provoked a change in finger design.

Parallel studies at another DOE laboratory used a two-jaw, angular, gripper finger design similar as the fingers shown in figure 4. One of the benefits of the angular design was discovered to be the ability to re-center a puck that was not in the center of the gripper target area. The flat sides that first made contact would push the puck to center. This "re-centering" would occur with the circular finger, but to a much lesser extent.

By design, the angular fingers create the undesirable effect of multiple point contact (four-point contact with the two-jaw and eight-point contact with the three-jaw). However, with the use of soft finger pads, the harsh impact would be softened and the load would be better distributed. The angular finger design proved favorable during many of the initial gripper tests.

The re-centering capabilities of the angular finger design give the puck handling process a certain degree of versatility. That is, the position of the puck, relative to the robot, can vary (slightly) without interrupting the process flow. This versatility is paramount for the gripping of the sintered puck from the furnace tray. As explained previously, a sintered puck can shift as it shrinks during the sintering process. The re-centering capabilities of the angular finger design will successfully capture a shifted puck.

A study was performed with a sintered puck to determine the limits of the re-centering capabilities of the angular finger design. In the study, a sintered puck was randomly placed off-center in the footprint of an unsintered puck. A three-jaw gripper was used for the study. Initially, the puck was placed on a wooden workbench, the fingers did not have pads, and a stainless steel mockup of a sintered puck was used. The stainless steel puck was easily re-centered at all extremes. Next, a surrogate sintered puck was used. The sintered puck also re-centered, but with slightly more difficulty due to the increased friction between the puck and the table. More difficulty was encountered when gripper pads were placed on the fingers. Once again the increased friction between the fingers and the puck hindered the re-centering capabilities. In every test, increasing the air pressure to the gripper helped the re-centering capabilities.

From these tests it was apparent that the finger pads had opposing effects. They increased the firmness of the grip, but hampered the re-centering capabilities. To overcome these opposing effects a composite finger pad was devised and tested (see figure 5). The composite finger pad was a foam rubber pad covered with a strip of stainless steel shim stock. The foam rubber provided the cushion effect and the shim reduced the friction to enhance the re-centering capabilities. Any firmness of grip due to the shim stock could be accounted for with an increase in air pressure to the gripper. The composite finger pad worked well in every finger configuration tested.

A six-axis Mitsubishi robot was purchased and is being tested for the press unloading and puck transfer operations (see figure 6). The robot is a 3 Kg model with an extended upper arm that provides an expanded work envelop. The test robot is the standard version but is available in a clean room version, which will be used for radioactive operation.

The end-effector design that was developed with the pick-and-place machine was incorporated onto the six-axis robot. From this point on development of the robot and end-effector was mainly focused on the press unloading since the development paths of the two operations were parallel.

A mockup of the press unloading glove box was built around the robot. The mockup contained representations of the press, the inspection stations, and a furnace tray stacker/unstacker (SUS). The robot mockup provided a platform to develop the robot, coordinate automated equipment interfaces, complete cycle-time studies, and much more. The dimensional and spatial relationship between the robot, press, inspection stations, and a SUS was based on existing prints or actual hardware. The sequential operation of the automated equipment in the mockup is controlled with a PC. Communication between each devise and the PC is via the RS-232 serial port.

The press representation was constructed using the design of the actual puck press used at another DOE facility. Representations of the press platen and associated hardware were included in the mockup to simulate spatial restrictions for the end-effector (see figure 6 and 7). The press is located at the side of the glove box to keep the axis of the glove box free for linear transportation of material below the robot.

The SUS will present an empty furnace tray to the robot. An empty tray must be presented in the same position each time for the robot to successfully palletize the pucks. At first, the SUS was represented as a platform, which was the approximate size of the furnace tray. As the design of the furnace tray progressed, the platform was replaced by a plastic mockup of the tray (see figure 8). The actual SUS was placed into the mockup after it was designed, assembled, and tested (see figure 9).

The robot will remove the puck from the lower press ram. In similar press designs used in other industries, the press product is pushed out onto the platen, away from the ram, as the press shoe delivers more powder. This action, if used for the press unloading operation, would present an easier target for the robot. However, this option was avoided to reduce waste material. Early tests with the ceramic precursors indicated that minor bonding occurred between the puck and the lower ram. It was deemed that removing the puck with a twisting motion would produce the least amount of residue on the ram, when compared to a pushing motion by the shoe. The robot will provide the twisting motion for removing the puck from the press.

The inspection area consists of a platform scale for weight analysis, an array of lasers for dimensional analysis, and a digital camera for topological analysis. During the inspection sequence, the robot will first present the bottom and top surfaces to the digital camera. The end-effector must not obscure either surface during the visual inspection. The visual inspection process, whether manual or automated, will assess the surface quality of the puck. The platform scale and array of lasers are combined in the next step in the inspection sequence. As the puck rests on the scale the lasers measure diameter and thickness.

The puck is palletized on the furnace tray if it passes all of the inspection steps. The furnace tray has a four-wall design to support other tray stacked on top. A neutral material will be sprayed or laid in the tray bottom to prevent undesirable chemical reactions between the puck and the tray. The walls of the tray present an obstacle for puck palletization (see figures 1 and 8).

To reduce the size of the tray the spacing between each puck and the spacing between the pucks and the walls is constrained to ¾". This spacing limited the travel of the gripper jaws on the end-effector. Given the physical dimensions of the puck, and the design of the end-effector up to this point, the total jaw travel from gripping the puck to full open was around 7° (per jaw). Unfortunately, the gripper of choice had 15° of travel per jaw. Placing the point where the puck is gripped at mid-travel retained the gripper of choice. This method proved helpful in other aspects, as will be discussed later.

Sensors were incorporated on the end-effector to provide robot operational information. The PC will monitor the output of the sensors directly or indirectly depending on the sensor.

Hall effect switches are used on the gripper to detect the full open and full closed positions of the jaws. The robot will monitor the output of these sensors directly and passed the information to the control PC. The puck is firmly gripped at mid-travel of the jaws. When the robot is ready to grab a puck the jaws must be fully open. The robot will check the state of the gripper prior to gripping the puck. If jaws are not fully open the robot is not ready, and a message to the control PC will stop the process. Conversely, the robot will check the state of the sensor after it has grabbed a puck. A puck is not in the gripper if the state of the sensor is fully closed, and a message to the control PC will stop the process.

Two different types of sensors were tested to prevent unintended contact with the furnace tray. The first was a collision sensor, having rotational, angular, and axial compliance as well as a trip sensor to indicate excessive compliance. The second was a multi-axis force/torque (F/T) sensor, which provided a linear output corresponding to a force and torge applied to the sensor. The control PC monitored the output of these sensors directly.

The collision sensor provided substantial compliance. The control PC monitored the tripping switch. When tripped the control PC would stop the process. The force or torque required to cause the sensor to "give" was adjustable with air pressure. A higher pressure was used when the puck was removed from the press and inspected, to add stiffness to the end-effector. A lower pressure was used when the puck was around the furnace tray during palletization, to detect minor impact. This method worked reasonably well. However, the cantilevered weight of the gripper and puck caused a bias force in the Z-direction. The tripping force could not be set below this bias force.

The F/T sensor provided force and torque feedback to the control PC at a rate of about 0.2 seconds. The control PC would monitor the feedback and "react" when a force or torque limit was exceeded, indicating an impact. Different force and torque limits are used during different steps of the press unloading process. A higher limit is used when the puck is removed from the press and inspected. A lower limit is used when the puck is palletized. The speed of the robot is also reduced while operating around the furnace tray during palletization. This allows sufficient time for the control PC to react when a limit is exceeded.

The ceramic furnace trays had relatively large manufacturing tolerances due to the manufacturing process. The relatively large tolerances would add up to a significant tolerance in a stack of trays. Lasers focused on the furnace tray will determine offset of the tray in the X and Y direction. The control PC translated the offset into robot palletization coordinates.

The F/T sensor will determine offset in the Z-direction. As the puck is lowered into the tray the F/T sensor will "feel" for the bottom of the tray. The control PC will signal downward motion of the puck to stop when the limit is exceeded.

As mentioned previously, the three-jaw gripper will be used to re-center and pick up a sintered puck from the furnace tray. One of the conclusions that was alluded to in the re-centering study was that re-centering became more difficult (more force was required) as the friction between the puck and the tray surface increased. The tray surface will be a neutral sprayed-on of laid-on material. A few of the laid-on materials being considered are very hard and porous ceramics. The surface quality of this material will create high friction with the sintered puck. The force required to re-center a puck on this material may approach the robot motor overload limits. Furthermore, the act of re-centering the puck may chip particles off of the bottom of the puck, creating unwanted contamination.

There are several ways for the robot to locate the sintered puck on the tray. Two ways will be explored using the glove box mockup if the three-jaw method develops insurmountable problems. One is with a digital camera. This method has been employed in other glove box mockups and can be used in the tray transfer glove box. However, a concern exists that the furnace trays could discolor from the sintering process so that it would be difficult to distinguish the pucks from the tray (the sintered pucks are typically dark gray and the trays are typically white). It would become difficult for a digital camera to find a puck in a discolored tray.

The second method will use the F/T sensor to "feel" the offset of the puck. The X and Y forces will be zero if the puck is exactly centered in the closed jaws of a three-jaw gripper. There will be a non-zero value for either the X or Y (or both) forces when the puck is off center as the jaws try to force the puck to center. The control PC can recognize the non-zero forces and, by vector analysis, signal the robot to move towards center.

The radiation levels in the actual glove boxes slowly damage certain materials and electrical components. The materials and electrical components of the end-effector are extremely prone to damage due to their proximity to the puck. If possible, radiation resistant parts will be used. If not, the ALARA concept will be used to increase the life expectancy and decrease the maintenance requirements of susceptible parts. ALARA is a method of reducing exposure to radiation workers by keeping exposure As Low As Reasonably Achievable. This is accomplished by the three principles of 1) reducing the exposure time, 2) working at a safe distance, and 3) using appropriate shielding. Keeping the radioactive materials away out of the glove boxes until they are ready for automated handling can reduce the exposure time. The movements of the automated equipment can be programmed to keep susceptible parts away from the radiation source. And shielding can be used as needed or when the first two principles cannot be employed.

Dusting has occurred with several ceramic powder recipes that have been tested for the sintering process. In the actual press unloading glove box, the dust will contain radioactive material, which will contaminate other equipment in the vicinity. Contamination is defined as the presence of radioactive material in a location where it is not wanted. The radioactive material is accountable and must be contained and reused.

The contaminated dust can be managed several ways. Vacuum lines can be placed around the press where dusting is more prevalent. The vacuum line can remove the dust as it is produced. In addition, the glove box can be designed so that the dust collects in areas that are easily policed. The robot, or other programmable, automated equipment, can clean these areas at predetermined intervals.

Disposal of replaceable equipment in the glove box will be difficult and expensive when the equipment becomes contaminated. In most cases, the equipment will be located as far away as possible from the dusty areas. Removable covers that are easily decontaminated will be employed when relocation is not an option. The robot will be a clean-room version. The clean room version has tighter seals and a method to pressurize the housing. Pressurizing the housing will prevent contaminated particles from entering.