As the nation's power stations reach the end of their working lives the problem of their demolition has prompted engineers to look at implementing novel techniques in many aspects of nuclear decommissioning.
Using one laser, configured in two different ways, TWI's team has shown that both tube cutting for size reduction and concrete scabbling for removal of contaminated surface layers can be conducted in a safe, remote and efficient manner.
In March 2009, the UK's Nuclear Decommissioning Authority awarded TWI a contract to develop prototype equipment for demonstrating the twin processes of concrete scabbling and tube cutting, and how these technologies might be implemented for remote use in nuclear decommissioning environments. The goal of the project was to allow Site Licence Companies and supply chain companies to evaluate the technology in terms of both process capability and operating costs, mindful that the underlying technical issues had already been addressed.
Contaminated concrete and pipework present major decommissioning challenges in terms of the huge volumes of material to be treated, the radiation levels present and the number of facilities affected. The topics are highlighted on many occasions in Lifetime Plan (Technology Baselines and Underpinning R&D) documents for Sellafield, Dounreay and Magnox North for example. Several concrete decontamination techniques have already been evaluated and whilst water jetting or mechanical scabbling are favored options each has drawbacks. They both present significant secondary wastes and require extensive control and deployment systems.
Concrete decontamination by means of laser scabbling has the potential to avoid many of the above drawbacks. However, whilst the technique has already been demonstrated at a laboratory scale, to date, no representative scale demonstration has been provided which would give industry confidence in the technique.
Although pipe cutting has been performed on numerous occasions, most of the techniques used are slow to operate or are not suitable for remote deployment in highly active cells. Lasers for cutting are well suited to remote deployment due to their lightness and compact process heads. There is no reaction force between the head and the tube and they generate limited fume. However, as with scabbling, the process needs to be adequately demonstrated before active deployment will be seriously considered. Industrial lasers
A key parameter in most laser processes is the power density in the beam applied to the surface of the material in question. The two processes of concern in this article are unusual in that laser cutting requires a very high power density in the beam, whereas laser scabbling requires a relatively modest power density.
An industrial fibre laser was chosen for several reasons; it had to be suitable for both roles, it needed to be robust and compact, and appropriate in remote applications using optical fibre delivery of the laser beam power. The laser chosen has an output power of 5kW, adequate to demonstrate both processes, but the same type of laser is commercially available in powers up to 30kW.
In a fibre laser, the laser light is generated inside a small diameter optical fibre, some tens of metres in length. This fibre is connected to the beam delivery fibre, which is of the 'plug and play' type and easily interchangeable. The delivery fibres are well protected in a flexible metallic armored sleeve. Such fibres can be manufactured up to several hundred metres in length, without appreciable losses in delivered power.
The fibre laser produces light with a wavelength of about 1micron, so it is invisible to the human eye, being in the near infra red part of the spectrum. The performance of the laser was monitored using a laptop computer, which also provided detailed information about the operating status of the laser. Control of the laser was from the controller of the deployment system in use, in this case an articulated arm robot.
Single-sided laser cutting
Laser cutting is a very well established manufacturing process which accounts for the largest use of high power lasers. The majority of work performed involves cutting material up to about 20mm thick, with exceptional quality of the resulting edge.
Tube cutting is also performed commercially, but almost all of these systems rotate the tube under a stationary laser beam. For single sided tube cutting with a laser beam, a definite requirement for decommissioning activity, alternative systems are required.
The laser light arriving at the cutting head down the optical fibre first expands as it leaves the fibre and is then made parallel by a lens. Below this lens, a second lens then focuses the laser light to a very small spot to create the power density needed for cutting. The system used in this work is unusual in that its focusing lens had a focal length of 500mm. The effect of this was to produce a very narrow beam of light, with a large depth of focus.
This large depth of focus makes a major contribution to the process of single sided tube cutting. The laser beam is enclosed by a cutting nozzle and a nozzle tip with an exit diameter of about 5mm.
In contrast to conventional laser cutting, for tube cutting, the laser beam focus is positioned about 90mm below the tip of the nozzle, allowing tubes up to 170mm in diameter to be cut from one side. The cutting process is assisted by a high pressure jet of air, which leaves the nozzle concentric to the laser beam. This compressed air is necessary to blow away material in the kerf of the cut melted by the laser beam. It is particularly important for single sided tube cutting in achieving separation of the tube.
This cutting system was also equipped with a video camera which looks directly through the cutting nozzle. It is focused at approximately the same point as the laser beam. This is useful for remotely positioning the cutting head above the tube to be cut. For the work described here, the cutting head was manipulated by an articulated arm robot. All movement of the process head, and hence the laser beam, switching of the compressed air and control of the laser, was achieved through the single robot controller.
Using this equipment various options for single sided tube cutting were possible. Stainless steel tubes from 25mm diameter to 170mm diameter, with a range of wall thicknesses from 1.5mm to 11mm, were cut using single pass, two pass and multiple pass techniques.
Generally speaking, a two pass technique proved the most efficient. From examination of the cut edge of a tube it is clear that the quality of the cut at the side closest to the cutting head is much cleaner than the opposite side. This is because, on the first pass, most of the energy in the laser beam, and the assist gas, are used to cut material originally contacted by the beam.
Only laser energy and gas which have passed through the upper section of the cut are available to address the lower section and this is cut less effectively. For the second pass, a kerf has been previously opened in the top section and now the majority of the laser energy passes through this and acts more effectively on the lower section of the tube.
As an example of performance, Figure 1 shows the dependence of the maximum cutting speed at which the tube is severed as a function of laser power, for a tube of 155mm diameter and 1.5mm wall, during two pass cutting. Note that the cutting speed appears to be linear with applied power, at least up to 5kW.
The optimum assist gas pressure was about 8bar. Figure 2 shows cut sections from 60mm diameter tube, with wall thicknesses from 1.5 to 11mm, again for two pass cutting. Process parameters are given in the figure caption. The largest tube to be cut in this work had a diameter of 170mm and a 7mm wall. However this is not believed to be a limit for the technique. Using a three pass technique this tube was severed in a time of 7min, using 5kW of laser power. Another possibility demonstrated was the cutting of concentric tubes. For example, a 25mm diameter tube located inside a 60mm diameter tube. In this case a two pass technique was effective in severing both tubes at once.
To demonstrate the tube cutting process a demonstrator was assembled at TWI consisting of a closely packed array of 25-150mm diameter tubes, mounted in various orientations, using conventional fixturing. See Figure 3. This array was demolished in 15 minutes using over 50 separate cuts through both tube and fixtures.
In the laser scabbling process, the laser beam is applied to the surface of the concrete and its energy is absorbed, heating the concrete matrix and the concrete aggregate. Expansion of residual water vapor, probably in both the matrix and aggregate and differential expansion between aggregate and matrix, causes the concrete to break up in a highly energetic fashion, leaving a rough scabbled surface, consisting of matrix and aggregate.
In any effective use of this process for decontamination, clearly the laser beam must move with respect to the concrete surface and the ejected debris must be contained. In this work, the former was achieved by the use of an articulated arm robot and the latter by enclosing the process and using a large pump and filtration system to recover the debris. The scabbling head showing its major components can be seen in Figure 3.
In the scabbling system, the laser light was fed, via an optical fibre, to a set of optics similar to that used for laser cutting, although in this case the focal length of the lens used was much shorter. The laser light is brought to a focus at a small diameter aperture and then allowed to diverge to a diameter of about 60mm at the base of a debris recovery tube. This tube, about 150mm in diameter, was terminated round its circumference by a steel wire brush, in contact with the concrete surface. The aperture and the region through which the beam passed below the focusing lens were both protected by jets of compressed air.
On this system the air pressure and any possible contamination of the optical elements were continuously monitored. If contamination occurred a warning signal was automatically generated. If the compressed air failed, the laser beam could not be released.
The top of the aluminium tube was connected to a long flexible hose and then to a pumping system which removed the concrete debris as it left the surface of the material. The complete scabbling head was mounted on the arm of an articulated robot, which was itself mounted on a linear gantry some 6m in length. The scabbling process and effective debris removal requires the process head to be at all times, roughly perpendicular to and at a constant distance from, the concrete surface.
The six axes of motion offered by the robot allow this to be achieved. However, the scabbling head was also equipped with its own vision system. A combination of low power lasers and a camera were mounted on the side of the scabbling head. The information recorded by the camera is interpreted by software and the results fed back automatically to the robot controller. In this way, once a scabbling start point has been set and a scabbling area defined, the vision system and its feedback to the motion controller of the robot, automatically maintain both the attitude of the head perpendicular to the concrete surface and a constant stand off distance, as the scabbling process proceeds.
A 16kW motor powered the vacuum system which removed the concrete debris. Air is sucked in at the base of the scabbling head, through the wire brushes. This air draws the concrete debris into the flexible tube and down to the first stage of an enclosed separation process. Concrete particulate matter was deposited in a first container and concrete dust was collected via a filter, in a second container. The body of the pumping unit also contained two additional filter housings capable of containing HEPA filters. The efficacy of the debris removal system was high, with hardly any scabbled material remaining on the concrete regardless of its orientation.
For a given laser spot size on the concrete, the main process parameters are the laser power and the travel speed. Work performed has indicated that removal rate is proportional to laser power, at least up to the 5kW of power available with the laser being used. At 5kW power, this system has removed a square metre of material in a time of 110min. A single pass of the process results in a scabbled 'trough', lenticular in section. This shape is related to the energy distribution in the incident laser beam, which at the concrete surface, is Gaussian in form. A slower process speed will generally result in a deeper scabbled section.
For concrete containing limestone aggregate, the deepest section has been measured at 22mm, using a laser power of 5kW and a travel speed of 100mm/minute. For removal of large surface areas, a track overlap of 50% proved to be the most effective for producing a uniform scabbled profile. Re-scabbling over an existing track is possible, and does result in an increased removal rate. However, in multi-pass processing of the same track, the amount of concrete removed was seen to drop at each successive pass. For example, at 5kW laser power and 300mm/minute travel speed, the maximum depth of scabble recorded for three successive passes of the beam was 10mm, 18mm and 22mm, respectively. Surface contaminants such as grease and paint (Figure 4) had no effect on the scabbling process.
For concrete with a limestone aggregate, a 5kW laser will remove a square metre of surface to a minimum depth of 10mm in under two hours. Coverage can be increased by either reducing the depth of removal or by increasing the laser power. For other types of aggregate, which show less reaction to the laser than limestone, scabbling is not as effective and further work is required to optimise performance. A very effective and efficient system for cutting of stainless steel pipes and other fixtures/fittings has been developed. The cutting head is both lightweight and has a significant stand-off tolerance and so is relatively simple to deploy and operate remotely.
The authors are grateful to the Nuclear Decommissioning Authority for funding the work reported in this paper and for giving permission for its publication. The assistance of Matt Spinks and Paul Fenwick in conducting the trials is also acknowledged.