Telepresence – remote handling comes of age at JET30 July 1998
Following last October’s record breaking experimental run by the JET fusion project, the Tokamak’s hundreds of divertor panels have been replaced with a new configuration. To accomplish this work, a “telepresence” capability was used. This involves a fully teleoperational methodology using a highly adaptable “man-in-the-loop” controlled remote handling system which allows the skilled operators to complete the task as if it were being done manually inside the torus.
The Joint European Torus (JET) project is the world’s largest experiment to study the physics and technology of energy generation using nuclear fusion. The experimental machine, a Tokamak, comprises a toroidal vacuum vessel of 3 m major radius inside which a plasma is heated to temperatures up to 450 million degrees. The Tokamak stands 13 m high and weighs 4000 t.
A future fusion reactor will use a mixture of deuterium and tritium (D-T) as its fuel. The JET Tokamak has been in operation since 1983 using primarily hydrogen and deuterium in order to avoid producing a large number of fusion neutrons which would cause significant activation of the torus and a loss of manual access. This policy was adopted in order to ensure maximum flexibility for experimentation with the Tokamak mechanical configuration and components.
However, during October 1997 a series of deuterium-tritium experiments were performed in the JET machine which produced record breaking fusion power and energy. These have resulted in an increase in its radioactivity to a level which renders the torus inaccessible to personnel for around one year.
Looking at the inside of JET, the divertor system can be seen as a continuous channel formed by tile surfaces in the lower part of the torus.
The JET programme required that in early 1998 this divertor be replaced with an alternative configuration and the activation levels dictated that this modification was done using only remote handling methods.
It should be added that the experiments conducted in D-T were highly successful giving real confidence in the design of ITER, the next fusion machine that is being planned as a world-wide project.
A suite of remote handling equipment was originally specified, designed, built and commissioned to satisfy a general repair and maintenance function for JET during a final operational phase when significant amounts of D-T were to be used. The experimental nature of the JET machine and its progressive modification and development demanded that the remote handling system be as adaptable as possible and be able to undertake repair and maintenance tasks at short notice. To meet this requirement a fully teleoperational methodology was adopted and a system based on man-in-the loop control using bi-lateral, anthropomorphic, force reflecting servo-manipulators was implemented. The original remote handling system requirements did not include a major modification to the JET torus, but the inherent adaptability of the teleoperation philosophy has allowed its application with little modification.
Going beyond this rather dry statement, the reality of “man-in-the-loop” has been made clear. We have seen our operators reach, quite unconsciously and naturally, for tools which they know are “on the floor beside them”!
The MkIIa divertor comprised a ‘U’ shaped water cooled mechanical structure onto which were bolted 1444 MkIIa divertor modules. These had to be removed and replaced with 192 MkIIgb divertor modules of similar size but different shape. In addition to this primary task it was necessary to clean four beryllium evaporators, vacuum clean the divertor region, replace a number of small diagnostic components, inspect the torus protective tiles and undertake a 3-D dimensional survey of the divertor support structure using videogrammetry techniques.
The D-T operations made use of a 20 gm site tritium inventory and activated the torus to an in-vessel dose rate of 4000 µSv/hr at the start of our remote operations.
The JET Tokamak was designed for operations with deuterium-tritium as fuel and therefore took the requirements for remote maintenance into account from the start.
The Mk IIA and MkII Gas Box divertor modules, the support structure and the associated diagnostics were designed to enable the remote exchange in a short time. This is exemplified by the concept of tiles attached to tile carriers which are installed and exchanged in one piece. This has the added advantage that alignment of the tiles is achieved automatically by the correct installation of the carriers. Special attention was paid to the design, the clearances and tolerances, the lifting features and camera viewing, the possibilities for jamming and seizing, the materials used and the possibility for removal if damaged.
An in-depth study was made of the optimum design of the remote handling fixing bolts and threads including the exchange of seized bolts/threads.
Mascot, a bi-lateral, anthropomorphic, force reflecting servo-manipulator, has a 20 kg handling capacity, a 100 gm tactile sensitivity per arm and a winch which allows it to lift and lower components up to 50 kg. It operates in master-slave configuration, ie it duplicates the movement of the master unit being manipulated by the operator in the Remote Handling Control Room distanced from the JET vessel. It has additional features for force scaling, gravity compensation and a teach/repeat capability all provided by the digital control system. The mascot slave is configured with four cameras: one attached above the arms in the middle of the shoulder; one on a separate movable “camera arm”; and one attached to each wrist on a passive mount which can be directed wherever required by using the gripper on the other arm.
The JET Articulated Boom is a 10 m long transporter consisting of up to six sections flexibly joined together. It has 18 degrees of freedom and can carry up to 600 kg. It can reach any point within the vessel to within better than 10 mm. The boom can be pre-programmed to automatically execute any number of motions at the same time and to approach a pre-determined point.
The nature of the Mascot manipulator is such that in principle it can, with minor adaptations, use any tool including power tools designed for manual operation. This facility is particularly useful when required to deal with unplanned situations
A set of special tools was designed to minimise the complexity of the tasks, whilst maximising the speed and safety of the remote operations. The tools attached to each of the tile carriers had features to guide the adjacent tools, thereby eliminating the danger of damage to the tile edges when installing or removing tile carriers. As the in-vessel tasks consist in large part in bolting and unbolting operations performed directly using the manipulator, a specially designed ratchet driver was used to run-in the bolts. The same tool could also be used for unbolting by simply using alternate gripping features on the tool. Additional tools were designed for other tasks such as: the removal/installation of the inner wall, poloidal limiter and divertor entrance tiles; the cleaning of the beryllium evaporators heads; the vacuum cleaning etc.
The removal of the MkIIa modules was planned to be done by first removing the 48 Inner modules followed sequentially by the Outers and the Bases (see below). The design of the components required that this removal sequence be interrupted at two points for removal of three small radio frequency (rf) waveguides and the replacement of the diagnostic connector sockets on the divertor structure. Also, in order to remove the 48 Outer modules it was first necessary to remove and temporarily store inside the torus 96 rectangular carbon tiles spaced equally all around the torus.
The removal of all 144 MkIIa divertor modules was successfully achieved using the pre-planned techniques. The 480 attachment bolts were successfully released with the unfastening torque found to be typically within 20% of the installation fastening torque. The worst case was one bolt which required a torque 90% greater than the fastening torque. One unexpected aspect was the partial sticking of some of the inconel modules to the inconel divertor structure. Post removal inspection suggests this to be caused by partial diffusion bonding. Not all of the modules suffered from this effect, but those that did were more difficult to remove and required the Mascot operator to impart a twist or shearing motion on the module at the same time as lifting it with the winch.
After removal of the MkIIa modules it was necessary to clean the four beryllium evaporator heads and the divertor support structure using a remote handling compatible vacuum cleaner and then to undertake a full 3-dimensional digital photogrammetry survey of the divertor structure and other selected in-vessel components. These operations were implemented as planned with no problems.
The installation of the MkII gas box divertor modules was planned as a sequential operation starting with the Bases followed by the Outers, the Inners and finally the 48 Septum modules. A new set of RF waveguides were planned to be installed at the appropriate point in this sequence. These operations were implemented as planned with minor changes to the order in which modules were installed which was necessary to accommodate some unexpected tasks as discussed later. In addition by the end of these operations three of the Outer modules and three of the Base modules had been installed, removed and then re-installed.
The remote handling operations were facilitated by the installation into the torus of an eight camera viewing system and a range of temporary storage trays. These systems were installed and connected in the torus as soon as possible after gaining access and then were removed at the end of the operations with no problem.
The remote handling operations were carefully planned to ensure the safety of both the JET in-vessel components and the remote handling equipment itself. Accordingly, it was necessary to temporarily remove a small number of vulnerable tiles from the torus and to install temporary protection covers over other vulnerable components. These operations were also successfully completed with no problem.
Planned visual and radiological surveys were conducted using the remote handling cameras and Mascot to deploy a Gamma monitor. The video survey was designed to be an automatic sweep of every visible part of the torus internal elements but interrupted at various locations to enable manual control of the camera to be used to perform detailed close up inspections. The video survey was recorded and subsequent inspection using the video revealed further regions of the torus where a revisit for close up inspection was required. These operations were successfully completed over a 3 day period at the start of the shutdown and the automatic video survey was repeated at the end of the shutdown. The gamma radiation monitoring was successfully achieved using a hand-held monitor in the Mascot gripper and with the dial reading visible from the remote handling cameras. To obtain the maximum sensitivity of readings at various locations in the torus it was necessary for the Mascot operator to change the dial readout scale setting by rotating a plastic knob on the meter.
A number of unexpected events occurred during the shutdown which required immediate assessment and new operations to be implemented. In each of these cases, whilst the necessary tooling, procedures, Boom teach files and Mascot moves could be created and visually checked off-line, no practical mock-up trials or operator training was feasible before the actual task was performed inside the torus.
For example, the first major alteration to the planned operations occurred when it was discovered that the Outer waveguide of the RF system was unable to be unbolted even at a torque close to its expected yield. Unfortunately the waveguide mechanically trapped one of the MkIIa Outer modules in situ and after an assessment of the situation and a further unsuccessful attempt to dislodge the waveguide it was decided to prepare two mechanical levering tools, an inspection boroscope, an inspection lamp and a electric shear capable, if necessary, of cutting the waveguide into two. After three days the equipment was ready and taken into the torus. The waveguide fixing bolt is some 200 mm below the visible surface of the divertor structure and so the boroscope was used to confirm that the waveguide mounting bracket was distorted and thereby imposing a large strain on the waveguide. One of the mechanical levers was then used with Mascot to move the waveguide to relieve the strain. The bolt was then able to be unfastened and the waveguide was successfully removed using the second lever device and Mascot fingers. The operations inside the torus during analysis and final removal of the waveguide resulted in a loss of 1 day.
In total there were five unplanned tasks and a number of minor events. Of vital importance to finding a solution for all of the unexpected events described above and some other cases was the ability of the JET remote handling system to be used as if a man were inside the torus thereby providing the operations engineers with the tools for ad-hoc manipulation, inspection and diagnosis. For example the analysis of the mechanical interference problems was only possible because we were able to test various hypotheses by offering up the components in a free form way and at the same time inspecting the module and the infrastructure. Similarly, the decisions to proceed with new tasks was taken only after detailed inspection and analysis of the relevant parts of the torus.
The first JET remote handling shutdown has been successfully concluded on time.
Over 450 individual components have been remotely handled within the JET torus during a 15 week period starting in February 1998.
The 38 pre-planned tasks were implemented with minimal operational problems or procedural variations.
The JET remote handling equipment functioned and performed almost flawlessly with only two significant faults occurring over the entire period. The total operational time lost due to equipment problems was less than 3 days.
As found in previous manual JET shutdowns, a number of new tasks had to be included in the operations at short notice. All of the unplanned tasks were successfully completed and, whilst they took longer to implement inside the torus than if they had been fully prepared using the mock-up, the total interruption to the shutdown was insignificant due to time saving in other well-rehearsed tasks.
The successful conclusion of the pre-planned shutdown tasks was the result of detailed and rigorous preparation of the equipment, procedures and training.
The successful completion of the additional un-planned tasks was due to the adaptability of the man-in-the-loop based JET remote handling system. The telepresence capability provided by the system, coupled with the skill of experienced operations staff, allowed the work to be implemented as if it were being done manually inside the torus. Without this capability any of the unplanned tasks would not have been possible to perform without significant delays to the shutdown programme.