Waterjetting 7d - High-pressure Waterjet cleaning over sandblasting paint

Over the years I have been caught up in “discussions” with several folk about how good high-pressure and ultra-high pressure waterjet streams were as a surface cleaning tool, in contrast with chemical and abrasive use in removing paint and other surface layers. One debate was about cleaning some particularly toxic chemicals from various surfaces. The point that often comes up in these discussions is that of “how clean is clean?” And in this particular case it was stated that the surface could never be completely cleaned. The rationale for that position was because the chemicals would enter into any cracks and flaws in the paint, and could therefore be retained either in the top coat, or the underlying primer. My answer to that was to take a small sample and clean the surface over the first quarter, raise the pressure and remove the top coat on the second quarter, raise the pressure further and remove the primer down to bare metal on the third quarter, and then, after adding a small amount of abrasive to the water, remove a thin surface coat of metal from the sample. It seemed to be a convincing demonstration, though I will come back to one problem in a later post, and for this post I will discuss taking the paint off.

It is now reasonably well known that high-pressure water can be cost effective as a way of removing paint, particularly from large structures such as bridges, and ship hulls, but it took a while for some of the benefits to become evident.


Figure 1. It was originally estimated that it would save some $1.75 Canadian per square foot to clean the Quebec Bridge with ultra-high pressure waterjets, rather than sandblasting. That increases to $4.50 per sq. ft. were hand tools the alternative (WJTA Jet News, March 2000)

There are 8-million square feet of surface in the bridge. As I noted at the end of the last post, the historic method for cleaning surfaces, and removing deteriorated paint has been to suspend abrasive particles in an air stream, and to use those particles to abrade and erode the paint from the surface. When the paint, rust and other coatings have been removed the job is often considered finished when the surface is restored to a nice shiny surface finish. There is, however, a snag, when one does this. The numbers that I was once given were on the order of: from the time that a railroad wagon was put into service, it would take 5 years before it would require stripping and repainting. After that first treatment, however, the paint would deteriorate more quickly and often within another 18-months the wagon would have to be taken back for repainting.

So why is this, and why does high/ultra-high pressure paint removal help extend the life of that second paint coating? I, and the industry, are deeply indebted to Dr. Lydia Frenzel who did a lot of the pioneering work in helping to define the benefits of the technology, and then spread the word about them. The problem begins as the surface begins to corrode, and I will continue to use the wagon as the example, though the result holds true for many surfaces. As the rust and damage continues to eat through the paint and into the underlying metal, that surface is not attacked evenly, but, instead small pockets of corrosion develop, where the metal is eaten away more in the middle or along the sides of the pocket.

By the time that the surface is ready to be painted it is no longer, therefore, smooth, but rather is pitted and covered in corrosion.


Figure 2. Exaggerated illustration of the condition of the surface, with the overlying corrosion shown in green.

When the surface is cleaned with an abrasive, typically driven using an air stream to sandblast the surface, the particles will impact and distort the surface. Thus while the majority of the corrosion will be removed by the impact and scouring action of the abrasive, some will not. Further the impact of the abrasive particles will bend over the weaker structures on the surface as well as peeling over some of the metal on the surface.


Figure 3. Electron microscope picture of a piece of metal on the edge of a pass by an abrasive laden stream, so that the action of the individual particles in cutting into and plowing the surface can be seen. Note that this peels over metal edges, for example at the arrows.

The peeling over of the surface, and the flattening of it give the shine that used to be the sign that the job had been effectively done. There are, however, two disadvantages to this. The first is that by distorting the surface, the bending over of the metal traps small pockets of corrosion within the surface layer of the metal.


Figure 4. Representation of the metal surface after it has been cleaned with abrasive. Note the folding over of metal to trap corrosion products. The abrasive particles are also not small enough to penetrate into the smallest tendrils of corrosion migrating into the metal, and these pockets (green) also are trapped.

With corrosion already embedded in the surface, before it is painted, that will develop immediately and thus the relatively short time before it undercuts the paint and causes it to fall off. There is also another reason for this. As air pressure is increased to speed up the cleaning, and give that “shinier” surface it smooths the surface and makes it more difficult to anchor the paint on the metal. This was shown by F.W. Neville (and is quoted in the book “Blast Cleaning and Allied Processes, by H.J. Plaster) with this table:


Figure 5. Relative paint pull strength as a function of the pressure of the air driving the sandblasting stream in pre-cleaning the surface of the old paint, prior to repainting.

As the table shows, the higher the air pressure then the smoother the surface, and the poorer the bond made with the paint.

Now consider what happens when a high-pressure jet cleans the surface. The water does not have the power to distort the metal, but rather does have the ability to penetrate all the cracks and pits on the surface, and flush them clean. As a result the surface is left rough (to give a good paint bond) and corrosion free.


Figure 6. Illustration of the relative condition in which a high-pressure waterjet will leave the surface.

One of the difficulties that early proponents such as Lydia had in getting the technique accepted, however, lay in the cleanliness of the surface. Because the metal had not been distorted back into a smooth upper surface, it does not reflect light in the “shiny” manner that an abrasive cleaned surface does. Thus to those trained to the latter, it did not appear clean. There had to be a considerable amount of demonstration, explanation and training before it was accepted that this “grey” surface was actually cleaner. And there are now standards, issued by the Steel Structure Painting Council, that recognize this.


Figure 7. A primer coated plate (left) that has been cleaned to white metal (right) using a high pressure waterjet.

Note that actual microphotos of abrasive and waterjet cleaned metal surfaces can be found in the paper by Howlett and Dupuy (Howlett & Dupuy, NACE Corrosion/92, paper No. 253; Mat. Perf, Jan. 1993, p. 38, the waterjet pressure was 30,000 psi).

Waterjetting 7a - An intro to jet structure

Once a waterjet starts to move out of the nozzle with any significant speed, as the pump pressure begins to build, it becomes more and more difficult to look at the stream of water and get any realistic idea of its structure. Mainly what is seen is the very fine mist that surrounds the main body of the jet, and while some idea of the structure can be obtained by making cuts through material, it can be quite expensive to actually see within that structure. Part of the problem is that though the mist is very fine, it is also moving at speeds in the range of a couple of thousand feet per second. The human eyeball isn’t quite that fast. But we can use a very high-speed flash (in this case it was on for two millionths of a second) which has the effect of “freezing” the motion.


Figure 1. 40,000 psi jet issuing from a 0.005 inch diameter orifice, front lit.

However this mist still hides the solid internal structure of the jet and does not change much in relative structure, even when the internal jet conditions can be quite different. Fundamentally the internal structure was described by Yanaida at the 1974 BHR Group Waterjet Conference, and his description has been validated by many studies since.


Figure 2. The break-up pattern of a waterjet (Yanaida K. “Flow Characteristics of Waterjets,” 2nd BHRA Conf. 1974, paper A2.)

This structure holds for jets across a wide range of pressure and flow volumes, but it is difficult to determine the exact transition points of that structure conventionally. And this can lead to very unfortunate results. I have twice seen people back a nozzle away and then move their hand in front of the jet to show that even high-pressure jets (these were being used to cut paper products and had no abrasive in them at the time) could be “safe.” If both cases the individuals were very lucky to escape injury (water can penetrate the pores of the skin and lacerate the internal parts without any surficial signs of injury, and, as I showed last time, if the nozzle is too close it will slice through flesh and bone). I thought to take today’s post to show, though the use of photographs, why that was such a stupid action.

The photos were taken down at Baxter Springs, KS in the early 1970’s and involved the use of what was then a MacCartney Manufacturing Co intensifier, to shoot jets of varying pressure, and nozzle diameter along a path, so that we could see how coherent the jets were. As I mentioned above, the problem with looking directly at the jet is that the internal structure is hidden by the surrounding mist. To overcome that part of the problem we shone the light along a ground glass screen (to diffuse it) that was placed behind the jet, so that we could see the outline of the internal structure.


Figure 3. Arrangement for taking photographs of a high-speed jet.

This more of the downstream mist from the photograph, and a much better idea of the internal structure of the jet, and where the solid section ended could be measured.


Figure 4. Backlit, 30,000 psi jet issuing from a 0.01 inch diameter nozzle, the distance across the photograph is 6 inches.

The benefit of the technique is perhaps more evident when nozzles at different pressures and diameters and different chemistry are compared. First consider the change with an increase in jet diameter. From the front-lit view there is little difference in the jets. From the backlit, it is clear that the smaller diameter jet only reaches 3-inches across the screen, while the larger jet barely reaches the end of the range.


Figure 5. The effect of doubling the orifice diameter at the same jet pressure on jet range, the photo length is 6 inches.

One of the parts of the study we were carrying out in 1974 was to examine the effect that adding different long-chain polymers had on jet structure. The ones that we were looking at include some that are now used in the oil and natural gas industry to make the “slick water” that is used in the fracking industry to improve production from shale reservoirs. But it also has an advantage in “binding” the jet together. And so, in the study, Dr. Jack Zakin and I tested a wide range of different polymers to see which would be give the best jet.

There were a number of different things we were looking for. In cutting paper, soft tissue and water sensitive material for example, the polymer can bind the water sufficiently well as to further lower wetting to the point where it doesn’t have an effect. It also can improve jet cutting under water – but I’ll cover those in a few post on polymer effects that will come to later in the series.

The effect of a polymer (in this case an AP273) is shown in two tests where the only change was to add the polymer to the water for the lower one.


Figure 6. Jets with an orifice diameter of 0.01 inches at a pressure of 20,000 psi, the range is 6 inches, and the lower jet has had the polymer AP273 added to the water.

The narrower stream in the lower frame is the effect that we were looking for. Putting change in diameter and the better polymers together gave, as an example, the following:


Figure 7. The effect of changing jet pressure, nozzle diameter and polymer content on jet cohesion.

It might be noted that the jet in the bottom frame has as much relative concentration (and power) at the end of the range as the top jet had at the beginning of the range.

Now it all depends on what you want the jet to do, as to which condition you wish to achieve. Inside abrasive mixing chambers the object is much different than it is when the object is to cut a foot or more of foam with high quality edges. And there have been some interesting developments with different polymers over the years, but I’ll save those stories for another day.

But bear in mind that those individuals who could slide their fingers under the jet in the top frame of figure 7 would have had them all cut off if the jet had been running instead under the conditions of the bottom two frames, and in all three cases, to the naked eye the jets looked the same.

Waterjetting 3d - High-pressure pump flow and pressure

When I first began experimenting with a waterjet system back in 1965 I used a pump that could barely produce 10,000 psi. This limited the range of materials that we could cut (this was before the days when abrasive particles were added to the jet stream) and so it was with some anticipation that we received a new pump, after my move to Missouri in 1968. The new, 60-hp pump came with a high-pressure end that delivered 3.3 gpm at 30,000 psi. which meant that a 0.027 inch diameter orifice in the nozzle was needed to achieve full operating pressure.

However I could also obtain (and this is now a feature of a number of pumps from different suppliers) a second high-pressure end for the pump. By unbolting the first, and attaching the second, I could alter the plunger and cylinder diameters so that, for the same drive and motor rpm, the pump would now deliver some 10 gpm at a flow rate of 10 gpm. This flow, at the lower pressure, could be used to feed four nozzles, each with a 0.029 inch diameter.

Figure 1. Delivery options from the same drive train with two different high-pressure ends.

The pressure range that this provided covers much of the range that was then available for high-pressure pumping units using the conventional multi-piston connection through a crankshaft to a single drive motor. Above that pressure it was necessary to use an intensifier system, which I will cover in later posts.

However there were a couple of snags in using this system to explore the cutting capabilities of waterjet streams in a variety of targets. The first of these was when the larger flow system was attached to the unit. In order to compare “apples with apples” at different pressures some of the tests were carried out with the same nozzle orifice. But the pump drive motor was a fixed speed unit which produced the same 10 gpm volume flow out of the delivery manifold regardless of delivery pressure (within the design limits). Because the single small nozzle would only handle a quarter of this flow, at that pressure (see table from Waterjetting 1c) the rest of the water leaving the manifold needed an alternate path.

Figure 2. Positive displacement pump with a bypass circuit.

This was provided through a bypass circuit (Figure 2) so that, as the water left the high-pressure manifold it passed through a “T” connection, with the perpendicular channel to the main flow carrying the water back to the original water tank. A flow control valve on this secondary circuit would control the orifice size the water had to pass through to get back to the water tank, thereby adjusting the flow down the main line to the nozzle, and concurrently controlling the pressure at which the water was driven.

Thus, when a small nozzle was attached to the cutting lance most of the flow would pass through the bypass channel. While this “works” when the pump is being used as a research tool, it is a very inefficient way of operating the pump. Bear in mind that the pump is being run at full pressure and flow delivery, but only 25% of the flow is being sent to the cutting system. This means that you are wasting 75% of the power of the system. There are a couple of other disadvantages that I will discuss later in more detail, but the first is that the passage through the valve will heat the water a little. Keep recirculating the water over time and the overall temperature will rise to levels that can be of concern (it melted a couple of fittings on one occasion). The other is that if you are using a chemical treatment in the water then the recirculation can quite rapidly affect the results, usually negatively.

It would be better if the power of the pump were fully used in delivering the water flow rate required for the cutting conditions under which the pump was being used. With a fixed size of pistons and cylinders this can be achieved, to an extent, by changing the rotation speed of the drive shaft. This can, in turn, be controlled through use of a suitable gearbox between the drive motor and the main shaft of the pump. As the speed of the motor increases, so the flow rate also rises. For a fixed nozzle size this means that the pressure will also rise. And the circuit must therefore contain a safety valve (or two) that will open at a designated pressure to stop the forces on the pump components from rising too high.

Figure 3. Output flows from a triplex (3-piston) pump in gpm, for varying piston size and pump rotation speed. Note that the maximum operating pressure declines as flow increases, to maintain a safe operating force on the crankshaft.

The most efficient way of removing different target materials varies with the nature of that material. But it should not be a surprise that neither a flow rate of 10 gpm at 10,000 psi, nor a flow rate of 3.3 gpm at 30,000 psi gave the most efficient cutting for most of the rock that we cut in those early experiments.

To illustrate this with a simple example: consider the case where the pump was used configured to produce 3.3 gpm at pressures up to 30,000 psi. At a nozzle diameter of 0.025 inches the pump registered a pressure of 30,000 psi for full flow through the nozzle. At a nozzle diameter of 0.03 inches the pump registered a pressure of 20,000 psi at full flow, and at a nozzle diameter of 0.04 inches the pressure of the pump was 8,000 psi. (The numbers don’t quite match the table because of water compression above 15,000 psi). Each of these jets was then used to cut a slot across a block of rock, cutting at the same traverse speed (the relative speed of the nozzle over the surface), and at the same distance between the nozzle and the rock. The depth of the cut was then averaged over the cut length.

Figure 4. Depth of cut into sandstone, as a function of nozzle diameter and jet pressure.

If the success of the jet cut is measured by the depth of the cut achieved, then the plot shows that the optimal cutting condition would likely be achieved with a nozzle diameter of around 0.032 inches, with a jet pressure of around 15,000 psi.

This cut is not made at the highest jet pressure achievable, nor is it at the largest diameter of the flow tested. Rather it is at some point in between, and it is this understanding, and the ability to manipulate the pressures and flow rates of the waterjets produced from the pump that makes it more practical to optimize pump performance through the proper selection of gearing, than it was when I got that early pump.

This does not hold true just for using a plain waterjet to cut into rock, but it has ramifications in other ways of using both plain and abrasive-laden waterjets, and so we will return to the topic as this series continues.

Waterjetting 3b - Pumps, Intensifiers and Cannons

When we say a rock is hard it means something different, in terms of strength, to the meaning when we say that we want an egg hard boiled. Terms have to be, and usually are defined through the way in which they are used. At the same time each trade, industry or profession has certain terms that it adopts for its own with more specialized meanings than those which we, in the general public, are familiar.

Ask someone on the street what level of pressure they consider to be “high” and they might answer with numbers that range from 100 psi to perhaps 2-3,000 psi. And yet, within the industry those pressures are really quite low, relative to those most commonly used in cleaning and cutting. High-pressure systems are now available that will generate streams that run continuously at 100,000 psi, and the highest pressure jet that we generated in the MS&T Laboratories was at around 10 million psi.

Within that very broad range some simple divisions make it easier to group the ranges and applications of the different tools that are now common within different parts of the industry. At the same time, over the period of my professional life, the technology has moved forward a long way. Consider that when I wanted to run at test at 50,000 psi back around 1970 I had to use this particular set of equipment.

Figure 1. MS&T Water Cannon firing 12 gallons of water at 50,000 psi.

The water cannon was made by cutting the end from a 90-mm howitzer, and threading a one-inch nozzle on the end. Smaller orifices could then be attached beyond that to give different flow combinations. The pressure to drive the cannon was generated by putting 2,000 gm. of smokeless powder in a cartridge, and then loading and firing the cannon. We had been given the mount, which rotates around two axes by the then McDonnell Douglas (now Boeing), who had used it to hold and move the Gemini spacecraft while they were being inspected.

The pressure divisions which were debated and agreed by the Waterjet Technology Association back in the mid 1980’s broke the pressure range into three separate segments, which described the industry at the time.

The first range is that of the Pressure Washers. Operating pressures lie at and below 5,000 psi.

Figure 2. A small pressure washer being used to clean a drain. (Mustang Water Jetters)

These are the types of unit which are often found in hardware stores for use in homes, and while I won’t get into this until some later posts on safety, and on medical applications, it should be born in mind that it is possible to do serious injury even at these relatively low pressures. (Folk have been known to use the jets to clean off their shoes after work . . . need I say more – a waterjet cuts through skin at around 2,000 psi, and skin is tougher than the flesh underneath). At pressures below 2,000 psi these are often electrically powered. A gasoline motor is often used to drive the portable units that operate above that pressure range.

High-pressure units are defined as those that operate in the pressure range from 5,000 psi to 35,000 psi. The drive motors are usually either electrical or use a diesel drive, and units of over 250 horsepower are now available. Many of these units are known as positive displacement pumps. That is taken to mean that the pump, being driven by a motor at a constant speed, will put out the same volume of water, regardless of the pressure that it is delivered at (up to the strength of the drive shaft).

To ensure that the pressure does not rise above the normal operating pressure several safety devices are usually built into the flow circuit so that, should a nozzle block, for example, a safety valve would open allowing the flow to escape. Different flow volumes can be produced in larger units by placing a gear box between the pump and the motor. As the motor speed changes, for the same piston size in the pump, so the volume output changes also. However, because the pump can only deliver at a certain power the size of the pistons can also be changed so that, at higher delivery pressures the same motor will produce a lower volume of water. I’ll go into that in a little more detail in a later piece.

Figure 3. Section through a high-pressure pump showing how the crankshaft drives the piston back and forth in the cylinder block, alternately drawing low pressure (LP) water in, and then discharging high pressure (HP) water out.

Normally there are a number of pistons connected at different points around the crankshaft, so that as it rotates the pistons are at different points in their stroke. The evens the load on the crankshaft, and produces a relatively steady flow of water from the outlet. (Which, in itself, is a topic for further discussion).

As the need for higher pressures arose the first pumps in the ultra-high pressure range (that above 35,000 psi) were intensifier pumps. These pumps are designed on the basic principal that the force exerted on a piston is equal to the pressure of the fluid multiplied by the area over which it is applied. Thus with a piston that is designed with two different diameters can produce pressures much higher than those supplied.

Figure 4. The basic elements of an intensifier.

Fluid at a pressure of perhaps 5,000 psi is pumped into chamber C. As it flows in the piston is pushed over to the left, drawing water into chamber B. At the same time the water is chamber D is being pushed out of the outlet channel, but because of the area ratio, the delivery pressure is much higher. If, for example, the ratio of the two areas is 12:1 then the pressure of the water leaving the pump will be at 12 x 5,000 = 60,000 psi.

Over the years the materials that pumps are made from, and the designs of the pumps themselves have changed considerably, so that pressure ranges are no longer as meaningful as they were some 25-years ago when we first set these definitions, but they continue to provide some guidance to the different sorts of equipment, and the range of uses of the tools within those divisions, so I will use these different pressure range definitions in the posts that follow.