GreyPilgrim HTI Test Results
Ron Graham
A waste retrieval system consisting of the EMMA
robotic manipulator (by GreyPilgrim LLC) and the
lightweight Scarifier (by Waterjet Technology, Inc.)
was used in 1997 to demonstrate the retrieval of
simulated radioactive wastes on behalf of the Hanford
Tanks Initiative (HTI).
The EMMA was 33 feet in length, and consisted of four
hydraulically-controlled stages with weight and length
given as follows:
|
| Stage |
# of couplings |
Length, ft |
Diameter, ft |
Weight, lbf |
| One |
4 |
12 |
2 |
380 |
| Two |
2 |
7 |
2 |
205 |
| Three |
3 |
8 |
2 |
235 |
| Four |
2 |
6 |
1 |
155 |
| Total |
11 |
33 |
- |
975 |
The end-effector was mounted on a two-way movable
connector, allowing the Scarifier to be kept normal
to the waste simulant during testing and operation.
The deployment frame to which the manipulator was
mounted was controlled for vertical motion with a
range of five feet, thereby simulating the movement
of a deployment mast in a tank environment. The
Scarifier and conveyance hoses added approximately
100 pounds to the manipulator's total weight. This
system provided the following advantages:
- All electronics and other sensitive components
are kept out of the tank, thus ensuring as long a
life as possible for the in-tank system.
- Waste retrieval based on air conveyance leads
to an arm lighter in weight, and having less weight
offset, than if based on water conveyance. A light
arm is more easily and safely deployed through
existing Hanford tank risers, thus contributing to
system reliability and economy.
- The heritage of the end-effector and conveyance
system enables the development of a full waste
retrieval system based on known relationships between
subsystems. In this way, most of the waste retrieval
system can be designed concurrently with the
manipulator.
Arm with end-effector and conveyance hose.
The following tests were performed:
Advanced Stage Testing
- Demonstrate that EMMA will provide sufficient
stiffness and load-bearing capability to ensure
acceptable waste retrieval performance.
- Provide a component-level understanding of the
waste retrieval system.
Control System Testing
- Demonstrate a simple, intuitive operator
interface to the waste retrieval system.
- Demonstrate the ability to acquire and hold
end-effector position and pointing direction, and to
achieve motion trajectories.
- Define and estimate the cost of a control system
suitable for Hanford operations.
Integrated Retrieval Testing
- Demonstrate through IGRIP simulation the
deployment of the manipulator into a tank.
- Determine system operating characteristics.
- Demonstrate system flexibility over a range of
retrieval scenarios.
- Determine retrieval rate capability.
- Ensure safety and reliability in retrieval
process through all aspects of operation.
The following failures occurred during the testing
described here:
- Several storage barrels used to receive conveyed
waste and water were imploded during retrieval
testing. This was a result of an instantaneous seal
being made between the end-effector and cleared
surfaces of a waste tray. Those failures are well
understood and procedures to overcome them in the
future have been documented.
- A coupling failed during testing of Stage Four
prior to waste conveyance and was replaced by a long
rigid segment. This failure was due to operator
error: load cells were not activated prior to control
of horizontal motion in Stage Four, and thus tension
was not fed back and could not be limited by the
control system. This operator error has been
corrected in the control system startup
procedure.
General Procedures
- All components are labeled with tags. These tags
were reviewed prior to each test run to ensure that
each component is subjected to an appropriate
environment:
- stage designation
- weight
- dimensions
- weak points (where applicable)
- operating limits
- Mounted components were checked for proper
fastening, symmetry, pre-test damage or
deformation.
- End-effector and conveyance hose were inspected
for proper functionality.
- Background noise was minimized; lights turned
up; non-essential personnel sent away from immediate
test area; final safety checks and test announcement
made.
- Each test run was videotaped and labeled
on-camera.
- Data was recorded in Excel spreadsheets within
the immediate test area.
- Qualitative test results (e.g. observations as
compared to expectations) were recorded both on
videotape and in spreadsheet.
- Digital snapshots were taken as needed for
documentation.
- After the test, cables were relaxed; control
system user interface and actuator pump turned off;
end-effector turned off (where applicable); tags
reviewed as needed; and spreadsheet copied to
diskette for redundancy.
Results
Static Deflection. Deflections induced by
static loads on the arm were shown to be correctable
with operator intervention, and corrective actions
were well within the capability of actuators and
cables. Programmed tension limits were shown to be
sufficient to bend the Stage Four beyond its
planned maximum bend, if necessary.
Arm Curvature. Each stage was shown to share
its total bend approximately equally among that
stage's couplings.
Free Vibration. Vibration response to slewing
of an individual stage is approximately the same
whether Stage Two, Three or Four is slewed. The
natural frequency of these vibrations was about 0.5
Hz.
Steady-State Response. The Scarifier was
limited by the accuracy of its motor to frequencies
no lower than 1 Hz. With city water flowing through
it at 4.5 ft3/min,
the spin rate of the Scarifier was slowly adjusted
between 1 and 5 Hz, with the arm almost not being
excited at all.
Target Acquisition. The arm was slewed with
the intent of acquiring a small target after the
motion was completed. An operator with little
training was able to repeat acquisition to within
1.5 inches, for repeated slews with sudden stops.
Inverse Kinematics. The inverse kinematics
algorithm performed the calculation of joint angles
for slewing of two stages in a plane in less than one
second of computing time to achieve an accuracy of
less than one-tenth of an inch. As more stages
become involved in a move of the end-effector, the
algorithm of course becomes slower, though there is a
tradeoff between computational speed and final
position accuracy.
Obstacle avoidance.
Obstacle Avoidance. A cylindrical object was
placed such that a sweep trajectory like those used
for retrieval would bump the object. The operator
slowed the arm down and through the use only of
Stages Three and Four, and the end-effector pivot,
avoided the obstacle, which overlapped the basic
sweep path first by one inch, then by three, then by
four inches.
Approaching the waste.
Physical Deployment. The deployment system
was used successfully to move the arm up and down,
and was regularly involved in repositioning the arm
for passes along the waste surface during simulated
retrieval. At no time were loads passed through the
arm to the truss structure sufficient to impede
vertical motion capability, nor were vibrations
passed along the arm from the end-effector observable
in the truss structure.
The end-effector pivot was also useful in extending
the capability of the manipulator: as the end-effector
position is raised and lowered, the pivot enabled the
Scarifier to be held normal to the waste surface.
This capability was not used often during simulated
retrieval, because of the limited range of motion
required to access waste simulants, but when used
was always effective.
Successful conveyance. The waste conveyance
system prepared for this series of demonstrations
performed successfully in that
- About 12 cubic feet of salt cake and four cubic
feet of dried sludge were accessed by the
end-effector.
- The manipulator delivered the end-effector to the
waste surface.
- The end-effector cut through and fluidized the
waste simulants for conveyance.
- The conveyance system retrieved the fluidized
simulants and delivered the waste stream safely to
storage barrels.
- Two kinds of waste simulant (dried sludge and
salt cake) were used in testing.
- The arm remained fully controllable in the
presence of the waste stream.
- Expensive components in the conveyance system
(e.g. Scarifier, jet pump, cyclone, and blower) were
by all indications undamaged by the retrieval
process.
The testing was, however, ended somewhat prematurely
as a result of the failure of the conveyance hose.
The failure did not present any new issues; the
conveyance hose was merely the weakest part of the
system.
Description of conveyance. The conveyance
approach used here consisted of repeated passes of
the Scarifier over the waste surface in a "windshield
wiper" trajectory, which involved the use primarily
of Stages Three and Four of the manipulator. After
a single pass yielded as much conveyed waste as
possible, the elevator, Stage Two, and the pivot were
used to reposition the Scarifier for another pass
over a fresh section of the waste surface. (This
plan was repeated either until drum failure or until
sufficient waste was removed to give the operators a
good idea of system performance. Of the four trays
of waste used, only one tray of salt cake did not
have at least half its waste accessed by the
end-effector.) Between passes, the Scarifier and
high-pressure water were deactivated, and the control
system operator was given an opportunity to practice
the next pass and get used to the system "feel."
When the control system operator was ready, the water
jets were gradually brought from city water pressure
up to 30 kpsi, then the blower was activated. The
high-pressure water caused the Scarifier to lift up
from the waste surface, and the sucking of the blower
would tend to pull the Scarifier down. The control
system operator, left with no automatic means of
adjusting cable tension to compensate for these two
opposing phenomena, had a difficult time maintaining
standoff distance. For this reason, the end-effector
would spend nearly half the run time in a "dwell"
waiting for the operator to be able to change
direction. For this test, the dwell was not an
issue, as the water jets were unable to damage the
waste tray, but this observation should be checked
in future generations of the conveyance system.
The Scarifier would at times get stuck on the waste
surface, and the operator would need to pull both up
and sideways to remove it (a result of the overly
compliant shroud), and on these occasions, the
manipulator would move as though plucked like a
bow. This unintended impulse response revealed a
manipulator natural frequency in the neighborhood
of 0.5 Hz. The vibration was quickly damped out by
the manipulator's couplings and by the rough motion
of the Scarifier shroud. No higher-frequency
vibrations were observed.
Scarifier in action.
After the second pass through the third tray of salt
cake, the first drum failure occurred. A trench wide
enough to permit a seal had been dug in the waste
tray. Failures subsequently occurred more
frequently as the last salt cake, and then dried
sludge, were conveyed. It is possible that some
aspect of the conveyance system was compromised,
though not visibly so. In the first drum failure,
making the other failures occur more easily.
Safety measures. The following safety
measures were taken prior to waste conveyance, and
were proven effective:
At the point of operation
- Polycarbonate shielding, secured during testing
but easily movable, prevented operational
interference with observations.
- Operation of the deployment and end-effector
pivot control systems was on-off and simple.
- Power machinery was on-off and redundant e-stops
were available.
- Multiple operators were trained in the use of the
Scarifier, jet pump, and blower.
Within the conveyance system
- Working parts were shielded.
- Each power machine was assigned an operator
with an e-stop; each operator was empowered to shut
down in the event of any unexpected issues.
- Redundant ear protection was provided for
operators and observers; flashing lights and warning
signs were used to limit access to the test area
during conveyance.
A NIST safety officer reviewed these measures and
added comments prior to operation.
Conclusion. All tests were highly successful
and issues raised prior to testing have either
already been resolved or their resolution is
underway.
[Engineer's Companion home]
[Back to Ron's page]