Waterjetting 33d - Nozzle oscillation along a contour

In the last post I reviewed, in part, Dr. Shunli Xu’s work on oscillating nozzles, and how they can, on occasion, almost double the penetration while also improving surface finish. The problem with that basic technique, however, is that the nozzle is oscillating perpendicular to the direction of the cut, which is fine when the nozzle is cutting a long straight pass, but becomes more of a problem when the cut is in the form of an intricate contour. Given that the nozzle assembly does not usually rotate to follow that contour, the motion that gives the correct direction of oscillation when the head is moving laterally won’t when the cut is being made into the depth of the piece, i.e. parallel to the oscillation direction.

The answer to this is to provide the nozzle with an orbiting motion similar to that which, for example, Clark Barker used to provide a circular motion to a cutting head back when we needed to cut a path ahead of a drill head at a time when high-pressure rotary couplings lasted about 14 minutes. (We knew this because we had bought a model of every one we could find in the country and run them to failure, and the best lasted that long, provided you kept it cool by playing a hose on it. That was some thirty-five years ago, and things have got a lot better, and cheaper in that time).

Figure 1. Mechanism for orbiting a nozzle (after Barker). In the above figure the outer sleeve rotates, driven by an external motor, and with a flexible connection to the nozzle, the nozzle moves in a circular path, without itself rotating as it moves, as it moves in an orbital path as the outer sleeve turns.

In the model for Clark Barker’s device the intent was to drill a hole in coal some 6-inches in diameter, and the tool worked well in being able to do this. (It was used in a tool that turned from a vertical well to drill out horizontally within a 9-inch turning radius. The device was proved in the field by Sandia Labs, who drilled out into a coal seam from a vertical well using the second generation of the tool that was developed.

In many ways the use of an orbiting mechanism for cutting is a lot simpler to develop, given that, as noted in the earlier post, the angle that the jets must move through is very small (around 8-degrees). With the nozzle constrained so that it remains pointing only slightly off-axis, there is no need for the more complex tool required to advance a drill tens of feet into a coal seam (and deal with all the debris that was flowing back out of the hole at the same time).

I have described John Shepherd’s Wobbler tool in an earlier post and it is worth returning to that design and our study for a little further analysis.

The object of our study was to examine how the tool could be used in milling pockets in material, and more specifically how best it could be used to create a relatively flat floor to the pocket, while maintaining relatively sharp corners to the pocket walls, a capability that conventional mechanical tool milling does not allow. (Unfortunately I can’t at the moment produce any of the figures from that work, though they can be found in the paper we gave at the 17th Waterjet Symposium in Mainz in 2004.)

When it came to the assessment of performance, it is perhaps of note that Dr. Zhang’s study found an optimal oscillation speed at around 8 Hz. It would appear from our study that the optimal oscillation to achieve greater depth was just below 8HZ, whereas that which gave the greater volume removal rate was at around 10 Hz, which would both lie close to the optimum suggested by Dr. Shunli Xu.

The assessments were admittedly for different overall phenomena, Dr. Xu was interested in achieving a greater and cleaner cut, while Dr. Zhang was more focused on achieving a milled surface, typically to be achieved with a single overall pass, nevertheless the relative agreement on an optimal parameter is significant.

Further, in order to achieve a smooth floor for the pocket, Dr. Zhang was incrementing the nozzle between passes as a function of the width swept out by the jet. The initial overlap of the jets provided an uneven floor to the pocket that was removed when the jets were further apart. (In most cases with a 120% spacing between the passes a smoother surface was achieved).

However, with the generalized conclusion being that the optimal basic operating parameter (oscillation/rotation speed) was in the same range for both studies, and with the angle that the jet swings through on the order of 6 degrees, again of similar range in both studies, would appear to validate the cross-transfer of information.

The path that the Wobbler makes is shown, in exaggerated form, in Figure 2, and a typical result for pockets cut in glass and steel are shown in figures 3 and 4.

Figure 2. John Shepherd's Wobbler and the path it drives the jet along as it moves over a target.

Figure 3. Pocket of varying depth and contour milled from glass using the Wobbler.

Figure 4. Lettering and the map of Missouri cut into metal using the Wobbler.

Waterjetting 13d - Pocket milling considerations

If I were to hand you a cake and suggest you take some, you would likely take a knife, and make two cuts to define a wedge of the cake, which you could then easily remove with your fingers, and easily eat.

Figure 1. Conventional cutting of a piece of cake.

However if one takes the approach used in many machine shops, and in many of the processes that are used to excavate soil and rock, then the process and product are a lot different. For, instead of just making two cuts that make it easy to remove and handle the product, instead the operator takes the knife and chops up the piece of cake into small particles.

Figure 2. Milling pockets into a piece of cake, with a pile of crumbs.

Now while it is true that this might produce nice clean pockets in the cake, it does leave the extracted cake in small pieces that cost a lot more energy to produce, are a lot more difficult to collect and handle, and are relatively worthless without being reprocessed back into a more useful form.

In some cases, where the cuts are made to the edge of the piece, it is not that difficult to adapt waterjet technology to emulate the knife in the above example. Cuts can be made along the edges of the pocket in such a way that they intersect within the piece to define a block of material that can be removed.

Figure 3. Piece of metal trimmed from a block in a form that allows it to be used for something else.

The savings can be considerable and are not restricted to small pieces such as this. Consider if you wish to create a tube, 2.5-inches long in Hastelloy. One could mill out the central volume to give the required shape:

Figure 4. Tube made from a block of Hastelloy

Or one could merely cut around the desired shape in the block, using a high-pressure abrasive waterjet, which would produce the central block as a single piece that could then be used as feed stock for another part.

Figure 5. Removed block of material from the piece shown in Figure 4. (For this initial demonstration the quality of the side cut was not considered important, it can be significantly improved for both internal and external surfaces, as discussed below).

Life becomes a little more complicated when the pocket must be generated without through cutting the part, and where the depths of the individual cuts must be precise, at the same time as the wall quality must my produced to final surface specifications.

There are a number of techniques that can be used in this case, depending on the material being cut, and on the depth that the pocket must be cut to. As mentioned in a previous piece, when higher quality surfaces are required then the cut must be made only with particles that have not cut, bounced and recut the work piece. Rather they must leave the cutting zone after the first cut, without further damage.

When the abrasive cutting conditions are controlled accurately, then the precision control in the depth of cut can be high, so that, for example, channels can be precisely cut into engine parts as needed.

Figure 6. Milled channels into a surface (note that the channels were deepened with three passes of the cutting head). (Courtesy Ormond)

In the above case the channels were narrow enough to be removed by the full width of the cutting jet, however, when larger pockets must be cut then this precision must be maintained over the full depth.

A problem that can then arise comes from the shape of the cut as it gets deeper. If the nozzle is not advanced into the cut (and this is not normally possible given the relative widths of the two) then there will normally be a slight taper into the cut as it deepens. With relatively thin work structure this taper will often be slight enough that it does not fall outside the tolerance required for the cut. Quality control can be achieved, therefore, with a series of rapidly moving nozzle, shallow individual cut passes. Omertz, for example, showed that a much better control of cut depth and wall quality can be achieved by twelve passes at a higher rate than a single pass over the surface with the same residence time on the part. However the optimal condition will depend on the tolerance on asperities in the floor, and the total depth of cut to be achieved, as well as the material being cut. In deeper parts the sides of the wall can narrow to interfere with the jet structure and create ripples that take the deeper wall surfaces beyond this acceptable limit.

Figure 7. Schematic of the jet structure as it moves away from the nozzle.

One solution to this is to return to the technology that we demonstrated with John Shepherd’s Wobbler namely to move the nozzle in such a way that the jet oscillates over the surface, not only making multiple passes along the floor of the desired cut at a high enough speed to achieve good depth control, but also by slightly widening the cut there is less chance of undesirable bounces from particles on the edge of the jet from causing out-of-tolerance rippling in the lower parts of the cut.

Figure 8. Exaggerated movement schematic to show how oscillating the head widens the cut, allowing the jet better access to the bottom of the cut without sidewall interference.

This particular aspect of nozzle motion has also been studied in Australia and Italy. The Australian results, for example, (Patel, Chen and Siores, 6th Pacific Rim Conference, Brisbane, 2000) looked at oscillation both along the cut and transverse to it. They found that surface quality was improved three-fold on the walls of the cut, and that, depending on oscillation angle and frequency, the smooth zone cut depth could be extended further into the work piece. The optimal angle of oscillation was found to be three degrees.

Figure 9. The effect of nozzle oscillation frequency on the depth of smooth cutting in the walls of the piece. (Patel, Chen and Siores, 6th Pacific Rim Conference, Brisbane, 2000)

How one breaks out the floor of the pocket, after defining the walls remains a work in progress.

Waterjetting 13b - Milling without a Mask

As abrasive waterjets have developed they have been used to both cut through materials, and, in more recent work, have been used to mill pockets within the internal part of the piece.

Figure 1. Milled Pocket in Glass

In the early parts of pocket milling simple linear cuts were made adjacent to one another across the space where the pocket needed to be created. However, with the need to slow the head down and reverse direction, the edges of the pocket were being cut deeper than the inside floor, and this could cause some problems with part life and utility.

The first step to overcome this problem was to provide a mask, cut to the size of the pocket to be cut, but made out of a harder material, such as steel. By placing the mask over the piece, and setting the machine so that the cuts were made at constant speed over the pocket, a flat floor could be cut. All the slowing and reversing of the head takes place over the mask, so that it is destroyed fairly quickly. But if it survives one milling, then for some parts this provides a process that cannot be achieved in other ways.

Consider, for example, the sheet of glass cut in figure 1. The corners of the pockets are relatively sharp and of consistent radius all the way down the wall, which is relatively straight. A conventional mechanical milling tool transmits high levels of force between the part being milled and tool holder. Therefore, to prevent vibrations, the tool diameter must be no less than a quarter of the tool length. This means that the radius of the pocket wall cannot be less than one-eighth of the pocket depth. That restriction does not exist with an abrasive waterjet milled pocket, where the radius can be much tighter.

This is a critical issue in the milling of parts, where the milling is to get weight out of the component. In many parts that are made for the aircraft industry the part can be designed so that much of the internal volume is not needed for strength, and can be removed to lower the weight. But with conventional tools there are limits to how much can come from a single pocket, not with the AWJ system.

As the above figure shows, and masking and other techniques allow, the radius of the corner can fall below a tenth of an inch even when milling pockets more than eight-inches deep.

There remain, however, a number of problems with the use of the masking technique. It takes time to make and install the mask, and it costs an additional expense that makes the process less competitive. One of the problems that arise with the use of masking comes with rebound of the abrasive from the mask. Dr. Hashish has illustrated this problem with a diagram.

Figure 2. Abrasive rebound from a worn mask (Dr. Hashish)

If the mask is not shaped properly, or if it has been used before and is worn, then it may have a chamfered edge. When the abrasive waterjet stream strikes the curved surface it can be reflected back onto the work piece, giving an unwanted erosion shadow along the edge of the pocket.

Another problem can arise if the speed of the nozzle, and the distance that the nozzle moves between passes is not controlled to ensure a smooth and even cut over the pocket surface. As mentioned in an earlier post, the roughness of the cut increases if the abrasive particles are allowed to bounce and make a second cut within the piece. To ensure quality, as a result, the nozzle should be moved, relatively quickly, over the workpiece.

Yet the inertia of the cutting head, and the drive assembly in the table motion controller make this difficult to do at relatively high speed. John Shepherd at PIW Corp came up with an answer to this problem, that coincidentally did away with masking.

Figure 3. The Wobbler showing the nozzle motion.

The concept behind the device is that, by slightly oribiting the motion of the focusing tube around an axis, the jet will sweep out a circular path on the workpiece. Because it is only the end of the focusing tube that is moving the forces involved are small, and easily provided through a small motor on the device. The relative speed with which the nozzle moves over the surface is now much higher, while the speed of the main arm remains relatively low. The device was studied at MS& T and the parameters that controlled the depth and quality of cut were found by Dr. Shijin Zhang as part of his doctoral research.

As with the control of single passes of a non-oscillating nozzle, the distance between adjacent passes is critical to the satisfactory performance. If the distance is too great then ridges will be generated in the floor that are almost impossibly to remove using abrasive waterjets alone. Dr. Hashish, in an early paper on milling, for example, showed that if the upper layers of a pocket are aggressively milled with higher pressures and larger grit sizes, that this floor roughness cannot be later removed by using finer grit sizes. This is because the finer grit, while removing some surface asperities will still erode the surface relatively evenly, so that the roughness pattern shown in figure 4, cannot be later removed entirely.

Figure 4. Rough floor to the pocket where the distance between adjacent passes is too great. (Dr. Zhang)

On the other hand it is not always necessary to have a high quality surface for the pocket. For example MS&T have made a number of plaques where metal plates, cut and lettered with the AWJ are then inset into pockets in polished samples of marble or granite. Since these are not strength-bearing, and the plates are glued in place, the pocket floor does not have to be of that high a quality.

Figure 5. Milled pocket in the shape of the United States, Note the edge sharpness and the narrow cutting radii.

On the other hand, where a smooth surface is required then this can be equally well achieved through programming the path of the overall head movement, so that the nozzle sweeps the floor evenly. The glass plate in Figure 1 was also milled with the Wobbler.

Figure 6. Pocket cut into metal without a mask, using the Wobbler. Note the smooth floor.

I will come back to this topic next time.