M. R. Duignan and S. Y. Lee
Westinghouse Savannah River Company
Aiken, SC 29808

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy.

Available for sale to the public, in paper, from:  U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161,  phone: (800) 553-6847,  fax: (703) 605-6900,  email:  orders@ntis.fedworld.gov   online ordering:  http://www.ntis.gov/support/ordering.htm

Available electronically at  http://www.osti.gov/bridge/

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062,  phone: (865 ) 576-8401,  fax: (865) 576-5728,  email:  reports@adonis.osti.gov

No work is done in a vacuum and even a simple literature search involves the help of many generous folks. We would like to thank all the people we bothered, cajoled, and harassed to obtain the information contain herein. While West Valley did not have a lot of information to contribute we appreciate the time that both Steve Barnes and Larry Petkus gave to answer our questions. From the Pacific Northwest National Laboratory, Monte Elmore was very helpful to understand the several tests he and other carried out, and both Pete McGrail and Leela Sasaki were generous enough to send documentation of past abrasivity studies; we are grateful. The Savannah River Site (SRS) contributed in various ways: to understand past material studies we thank Ken Imrich and Phil Zapp; to recall past flow-loop work for the Defense Waste Processing Facility we thank Chris Randall and Charles Nash; to obtain past documentation on flow-loop work for the Consolidated Incineration Facility we thank Dan Burns. We would also like to thank: the manager, Lee Whitlock, of the Georgia Ironworks, Inc. (GIW) Hydraulic Testing Laboratory of Grovetown, Georgia for allowing us to take a tour of the facilities, to see where past flow-loop work was done for the SRS, and Graeme Addie, V.P. of Engineering and R&D of GIW for allowing us to rifle through past reports on erosion studies. We thank Zafar Qureshi of SRS for allowing us to reference the Ph.D. dissertation by Dr. Wu, even though it has yet to be published. Any worthwhile document needs to be critically reviewed and we thank our technical reviewer, Hector Guerrero, for a sharp eye. Lest we forget, we would like to thank Susan Isaacs, Sue Stallings, and the rest of the SRS library staff for answering all of our questions and for making any library search much less cumbersome. Finally, we would like to thank Mike Johnson of CH₂M Hill Hanford Group Inc. and Paul Townson of BNFL, Inc. for initiating this work and for providing requested documentation and answering questions promptly.

Keywords: Hanford River Protection Project , Pretreatment, Filtration, Evaporator, Erosion, Corrosion, Stainless Steel, Hanford Waste

Tests are planned to measure the wear rates in scaled flow loops that represent full-scale systems in Pretreatment sections of the Waste Treatment Plant to be built as part of the Department of Energy (DOE) River Protection Project. Those tests are to be done in the Experimental Thermal Fluids Laboratory of the Savannah River Technology Center at the DOE Savannah River Site. To make the most of those tests, and to build on work done in the past that may relate to such tests, a literature search was performed. Current literature, both inside and outside of the DOE Complex, may contain data and insights to estimate wear in the flow systems of concern. In the very least, available literature will have information to assist in the design of scaled flow loops. Those loops would mimic important aspects of the cross-flow filtration and evaporator systems to include pipes, filter tubes, bends, valves, pumps, other equipment, and simulated wastes to measure the wear rates.

A literature search was done to review slurry wear studies from known sources in the DOE-Complex and from selected appropriate non-DOE-related research completed during approximately the last 20 years. This search uncovered a large body of data that exists with respect to slurry wear. Unfortunately, that literature information is not sufficient to avoid doing an experimental test to elicit accurate results on wear rates in a specific flow loop with a specific slurry. The number of parameters to consider, the complexity of the wear mechanisms, and the lack of comprehensive computational tools to estimate wear, make wear estimates without testing questionable. The state of wear knowledge is not at the point where a series of parameters like particle size, shape, hardness/material, velocity, etc.; slurry chemistry, solids concentration, flow rate, temperature, etc.; and flow loop orientation, material, internal surface finish, etc., can be used to estimate a wear rate without actually doing a physical experiment. When data do exist from an experiment, which includes measurements of many of the needed parameters, they are only specific to that specific test setup; extrapolation to different environments is risky at best. Data are available for some very general geometries like a straight pipe, but with very specific slurry flows and conditions. Besides, straight pipe is expected to have the least amount of wear with respect to other flow loop equipment. Moreover, it is rare to have flow systems similar enough to have any confidence in the extrapolation of wear test results.

It is concluded that combining the knowledge gained from the literature, a flow loop with a prototypic slurry that is employed under prototypic flow conditions, can be used to accurately estimate wear, especially, if the amount of wear and wear rate expected are low. The goal of the tests will be to determine if the wear rate is greater or less than 0.0025 inch/year. However, care must be taken when extrapolating wear rates to full scale. While the computational evaluation of wear rate without experimental benchmarking is not recommended, it can be used effectively to evaluate similar flow regimes at different scales. Such an evaluation that is benchmarked to experimental data will increase confidence in full-scale wear-rate estimations.

Part of the River Protection Project is to build a vitrification plant at the Hanford Site to stabilize the radioactive waste currently stored in large tanks at that site. The pretreatment system that prepares the waste for vitrification includes separation technologies like filtration and evaporation. These systems will contain continuous flows of solids-liquid mixtures that will cause wear to the pipes and associated equipment. (In this document the word wear is used to indicate the loss of material from a surface due to the flow of slurry. That loss can be caused by any of the mechanisms of erosion, corrosion, and the synergistic effect of both together. From a fundamental point of view to understand wear, knowledge of all three slurry mechanisms is important and there are many studies that attempt to quantify them individually. However, the goal of this work is to quantify the loss of material from walls of planned flow systems independent of how the material is removed.) To have a system in good working order over the designed plant lifetime of 40 years, where maintenance must be minimized because of a radioactive environment, the rates of wear need to be quantified. Once the wear rates are known, design requirements and maintenance schedules can be developed to insure continuous safe plant operation. As a precursor to experiments to measure wear rates, the current state of the wear knowledge needs to be evaluated to find out if the proposed flow systems have been studied in the past, especially within the DOE complex. Moreover, by searching through current literature, it may be possible to obtain data and insights to estimate wear in the flow systems of concern. In lieu of available data, the literature may have information that will assist in the design of scaled flow loops. Those loops would mimic important aspects of the cross-flow filtration and evaporator systems to include pipes, filter tubes, bends, valves, pump, other equipment, and simulated wastes to measure the wear rates. The scaled tests would be designed, fabricated, tested, and analyzed by the Experimental Thermal Fluids Laboratory, with assistance of other groups, of the Westinghouse Savannah River Company.

The pretreatment system being planned for the RPP vitrification plant will have a large range of fittings, flow changes, temperatures, velocities, and slurries. The hardware will contain 90° bends, expansions, contractions, plenums, valves, pumps, and instrumentation. The slurry flow conditions will include temperatures from 25°C to 85°C, velocities up to 5 m/s (~16 ft/s), and pressures up to 605 kPa (88 psia). The slurries themselves will be caustic (pH ~ 14), have a high insoluble solids concentration (up to 20 wt%), contain solid particles with sizes that range from submicron to tens of microns, have the non-Newtonian rheological characteristics from a Bingham plastic to a time-dependent pseudoplastic and, of course, be radioactive.

While the pretreatment flow systems have many similarities to existing industrial flow systems, the combination of fittings, filters, and slurries make the systems unique. This uniqueness makes the extrapolation of existing data from other wear studies difficult, if not impossible.

A search through various sources of information and communication with other DOE sites did not reveal data that can be used to directly evaluate the Hanford RPP pretreatment flow systems. However, some of the following information may be used indirectly to better estimate the knowledge needed to study slurry erosion.

The West Valley Demonstration Project (WVDP) has some similar pretreatment systems to prepare waste for vitrification. However, no known wear studies exist (Barnes, 2001; Petkus, 2001). This is understandable because that project had a very short life span and wear from flowing slurries was assumed to be insignificant.

Hanford Site

The DOE Hanford Site was, and is, faced with waste treatment plants with much longer time spans than the WVDP. In evaluating the existing wastes (e.g., Hodgson, 1995) the abrasive aspects of the slurries was recognized (McGrail, 1991). As Hanford prepared for an eventual waste remediation plant, some attention was given to the wear slurry could produce in the storage tanks and in the transfer piping systems.

To begin the process of stabilizing stored waste, the waste would have to be removed from the tanks by resuspending the solids before pumping; therefore, the first concern was the wear slurry would have on the tanks as it was being mixed with rotating jet mixer pumps. Smith and Elmore, 1992 studied the corrosion and erosion of the carbon steel (A-537) storage tanks by setting up a small-scale test that attempted to replicate the slurry jet action on the tank surfaces. On the average, they found that the wear rate was 4 mil/year using average jet velocities from 3.6 to 55 ft/s at temperatures of up to 103°C. The two different slurries (simulants of neutralized cladding removal waste and neutralized current acid waste) used in the test had pH~13, densities ~ 1.25 g/ml, and insoluble solids ~30 wt% that were made primarily of oxides of Al, Fe, and Zr. For one part of the test, a metal coupon was subjected to a continuous, perpendicularly oriented, stream of slurry, i.e., the jet formed a 90° angle with the metal coupon. See the impingement coupon in Fig. 1 (this figure is Fig. 6.1 from Smith and Elmore, 1992). To simulate the slurry stream from the rotating jet pump, another part of the test alternated metal coupons in and out of the slurry jet so that one coupon experienced the direct impingement of the jet for 90% of the time and the other for 10% of the time. However, the coupon that was not in the direct stream was still exposed to the slurry stream; it was in the path of the slurry leaving the impingement coupon, Fig. 1. That is, the coupon, which only experienced the direct impingement slurry jet for 10% of the time, was also attacked by the slurry at oblique angles for the other 90% of the time. Besides experiencing oblique-angle attacks by the slurry, there could have been secondary flows. In fact, the authors expressed surprise that the sample with only 10% direct-jet impingement showed the most wear (4 mil/year).

Significant wear at oblique angles is not surprising when considering the published data, which discusses the importance of the direction of a solid particle towards a surface (see the Particle Direction section below). Carbon steel, like most steels, has been shown to be a ductile eroder, which means the erosion rate is the highest at particle-to-surface angles of around 30°. Moreover, the erosion rate for ductile eroders is the lowest when the particle-to-surface angle is 90° (materials that erode the fastest at 90° are referred to as "brittle," like ceramics (Finnie, 1995)). From the setup used for this test, it is clear that the metal coupon that was held in the off-jet position 90% of the time received considerably more low angle particle attacks then the direct impingement coupon; thus it had more wear. This information is very useful in planning an erosion test because now it is known that attention must be directed towards surfaces which will be subject to low angle particle attack.

The second wear concern in the preparation of the Hanford waste remediation plant was the wear slurry would have on waste transfer piping. Due to the difficulty in extrapolating bench-top slurry abrasiveness test results to indicate the slurry wear that occurs in flow loops, a scaled flow loop erosion test (Elmore, 1994) was done. The test had two stainless steel (304L) simple flow loops, one circulated at 7.5 ft/s (in a 1-inch diameter pipe) and one at 13 ft/s (in a 1.5-inch diameter pipe), where the slurry solids concentration was tested at 15 wt% and 30 wt%. Two slurries were used: one was water and ground glass frit and the other was made from water and ground silica with particle sizes from 74 to 177 microns. The results showed a high wear rate (~5 mils/80hour or about ½ inch/year) in an elbow. This test was extremely conservative since the wear rate was based on sharp and large particles, and the liquid was plain water which is much more corrosive to steel than the strong caustic wastes in the storage tanks. Even with the flow loop test results, Elmore, 1994 clearly spelled out the difficulties in relating measured wear rates and to actual flow systems. He listed the many parameters that could affect the results.

Further work on evaluating wear/corrosion/erosion was not done at Hanford until recently, when Danielson and Pitman, 1999 did scoping studies with non-radioactive slurries simulants for the RPP vitrification plant. Two metals 316L stainless steel and Hastelloy C-22 were tested as possible candidates for the plant piping and vessels. As a subset of those tests an erosion test was done (Elmore, 2000) to evaluate the abrasivity of a waste stream to those two metals. This erosion test followed an ASTM standard (Miller and Miller, 1993) and determined that the waste stream was abrasive to the two metals tested. However, during the test the slurry had to be diluted because of its thick nature and therefore the actual slurry consistency was not known, making the test results less useful. Unfortunately, there is no way to relate those results to the actual wear a flow system would experience. At a minimum, a scaled flow-loop test is needed to show the magnitudes and location of slurry wear.

Savannah River Site

At the DOE Savannah River Site (SRS) some corrosion/erosion tests were either planned or done, but none can be directly useful for RPP waste treatment plant. The closest match was a planned wear test (Burns et al., 1991) for a cross-flow system to be used to filter a recirculated slurry. A slurry was created from scrubbing offgas of particulates from the Consolidated Incineration Facility (CIF); the mixture was composed of NaCl salt and (CIF) flyash. The test was to be done on a single-tube cross flow filter that was made of Hastelloy C-276 metal, had a 2-micron pore rating, was 60 inches long, and had an inside diameter of ½ inch. Unfortunately, that test only got to the point of visually benchmarking the filter’s new condition, using a fiberscope. The test was run for approximately 50 hours at a slurry velocity of 15 ft/s, but budget cuts in the program resulted in the test not being completed; wear results were not obtained.

At SRS many studies were made in preparation of the construction of another waste treatment plant called the Defense Waste Processing Facility (DWPF). However, tests to evaluate wear from slurry flow in the plant are almost non-existent. In the planning stages of DWPF, studies were done (Carstens, 1982; 1979; Lewis, 1982) with simple sand and water slurries to determine the hydraulic characteristics of the flow in the designed piping system. One goal of those studies was to analyze the relative abrasivity of the slurries, but to date that analysis has not occurred. Fortunately, wear problems that have occurred in DWPF could be fixed, since most of the equipment is remotely accessible. Gee et al., 1997 and more recently, Imrich et al., 1999, investigated wear in the Slurry Mix Evaporator (SME) tank, Fig. 2, on the impellers, Fig. 3, and cooling coils, because of equipment failure due to accelerated wear from the very aggressive environment of glass frit and waste sludge at high temperatures (~100°C).

In most cases, the many pieces of equipment of the DWPF plant are lasting their design lives, but certain pieces, like the lower radial-flow impeller in the SME tank, needed replacement after 18-24 months instead of the planned 5 years. Even after going through a series of different metals from Hastelloy C-276, to Stellite 6B, to Ultimet, the agitator life is still only about 3 years.

The unexpected reduced life of the SME tank agitator was caused by the combination of an aggressive mixing environment and equipment design. That mixing tank contains an oxidizing slurry with a 50 wt% concentration of glass frit (with a mean diameter of 100 microns). This slurry caused expected wear on the equipment, but due to trailing vortices near the sharp protrusions on the lower impeller blade, the wear was accelerated. Figure 4 shows waves of wear on the backside of the steel blade caused by the square protrusions of the blade bracket and the end of the blade itself. This phenomena was initially noticed in 1986, Fig. 5, on a full-scale test blade. However, at that time it was thought that the protruding bolts caused the increased wear; therefore the actual impeller was made as a single welded unit. Obviously, the bolts were not the only protrusions and the wear continued. The wear evidence shows why experimentation must be done to find out where unexpected wear may occur. The wear experience in Figs. 4 and 5 was not predictable through past unrelated wear tests, nor could it have been predicted computationally with the current state of wear modeling. This type of wear is a very clear example of why a testing is necessary to indicated potential wear problems. While the test done in 1986 did not resolve the wear problem, it clearly indicated where attention needed to be focused to avoid serious problems.

Due to the limited database on slurry wear from studies within the DOE Complex it is prudent to peruse the engineering community for the current scientific knowledge to better understand the state of art on wear. At a minimum, if studies do exist that closely match the needs of the present task that information would save considerable money and add confidence to predicting wear rates in a plant yet to be built. What becomes immediately obvious on searching for previous studies in slurry wear is the vast amount of information available. The study of wear by either erosion, corrosion, or a combination of both has occupied many researchers for a long period of time. During the last 50 years there have been literally hundreds, if not thousands [especially if tribology is included (Blau, 1997)], of studies carried out by an equal number of researchers (for example see: Blau, 1997; Finnie, 1995; Clark, 1990; Humphrey, 1990). It would be impractical due to time and cost to list all of those studies here. However, upon scanning the recent literature for the last decade or so, ideas can be formed about the state of knowledge on wear due to slurry flows and if wear is predictable.

The first impression one gets in scanning through available literature on wear is the complexity of the subject. There are at least eight different mechanisms (Shook and Roco, 1991; p. 157) for a particle to erode a surface and there are an equal number of environments that such wear can take place. Moreover, as the solids concentration increases the particle-to-particle interactions must be taken into account (Xie et al., 1999; Humphrey, 1990) which generally make the problem intractable for computational analyses. The complexity is further compounded when trying to understand the wear contribution due to corrosion and the synergistic effect of corrosion and erosion (Stack et al., 1997; Aiming et al., 1996; Zheng et al., 1995; Umemura et al., 1987).

In studying erosion, research is primarily divided into two categories: 1. highly controlled impingement of a well defined particle or group of particle onto a surface to better understand an erosion mechanism, or 2. measuring wear on specific surfaces that are subjected to a slurry flow which have a varied mix of particle types and sizes, like in a flow loop. These studies lead to very good information on: erosion under controlled particle-surface conditions but are not related to any flow loop situation, or erosion of a specific flow-loop configuration, with slurries that have relatively uncontrolled particle-to-surface conditions. The information of the former can not generally be applied to most slurry flow loops because of not knowing the particle-to-surface conditions and the latter is only applicable to the specific flow loop and slurry tested. The challenge is to understand the particle-to-surface flow condition in any flow loop so that the information obtained in the former is applicable. While progress has been made, to date, this challenge has not been overcome, except under a very select set of circumstances, i.e., wear in a straight pipe (Gupta et al., 1995; Shook et al., 1990; Colwell and Shook, 1988; Nasr-El-Din et al., 1987; Roco and Shook, 1983; Sharp and O’Neill, 1971; Shook et al., 1968) or an elbow (Mishra et al., 1998) for known particles, particle distributions, particle concentrations, slurry chemistry, pipe materials, temperature, pressure, etc. More importantly, research is ongoing to relate data obtained in benchtop slurry testers (jets, pot, Miller, Coriolis, etc.) to actual flow conditions. However, that research (Pagalthisvarthi et al., 1998; Gupta et al., 1995; Pagalthisvarthi and Helmly, 1992; Miller and Miller, 1993; Elmore, 1994) is very limited and the extrapolations of predictions of wear rates to flow systems, other than what were used in those studies, would be questionable. In fact, Humphrey, 1990 states that "innovative experimental and theoretical approaches (to studying wear) are sorely needed," and he was referring to gas-solids flows. This is especially true when the carrying fluid is a liquid where the increased viscosity will have a significant effect (Clark, 1990).

Slurry wear research, with a slurry defined as the mixture of solids in a fluid, can be divided into Liquid-Solids and Gas-Solids. These investigations can be further broken down into subcategories: Flow loops, jets, testers, and tests on specific equipment, like pumps (the specific references are not listed here but the Reference section is broken down into subject areas to facilitate use.) For the task at hand the liquid-solids investigations would be of more interest. However, most of the available fluid- and solids-wear research deals with gas-solids studies which (while they cannot not be compared on a one-to-one basis with liquid-solid studies, Sargent et al., 1982) are useful to understand the state of wear research and are therefore included. In fact, a large percentage of liquid-solids studies are recent and have been done since 1990; much of that research has been guided by the earlier gas-solids research.

All of the available published research makes clear the large number of variables that must be considered to evaluate wear due to slurry flow. The three main pieces of the puzzle are the particles, the affected surfaces, and the carrying fluid. Table 1 lists some of those variables. In fact, Miller and Miller, 1993 mentions 33 variables to consider for their slurry tester and Clark, 1990 lists 21 variables of importance for slurry erosion.

As explained in the Introduction Section, this literature search was made as part of a program to estimate wear that will occur in a planned flow system. The system will consist of off-the-shelf 304L and 316L stainless steel parts, which will contain a liquid-and-solids slurry that, will:

• be oxidizing and highly caustic (pH~14),
• have small solid particles (the majority < 60 microns),
• have a high solids concentration (maximum of 20 wt%),
• have moderate velocities (2-5 m/s) and temperatures (up to 85°C),
• have borderline to moderate turbulence (Reynolds numbers from 3000 to 250,000) and
• have the non-Newtonian rheological characteristics of either a Bingham plastic or a time-dependent pseudoplastic.

The particles will have a variety of shapes (sharp and round), hardnesses (2.5-7.5 Mohs), and be made of many materials (Al, Fe, Si, Zr, U, and other metal oxides and hydroxides - for non-radioactive simulant testing purposes U will be substituted with W).

Considering the flow parameters listed above, the following discussions are highlights of pertinent information found in the included reference sources. This information is intended for guidance in evaluating wear, or to design a system to evaluate wear. The reference and explanations are by no means exhaustive but hopefully sufficient to develop a path forward in estimating potential wear.

Many researchers have considered particle size important to wear, but only after the average size becomes greater than some threshold. At one extreme, Zhong and Minemura, 1996 determined that erosion rate increases with size but the effect is "small" until a particle reaches 1000 microns. Another paper (Iwai and Nambu, 1997) determined that erosion rate only becomes independent of particle size above 300 microns. Some (Mishra and Finnie, 1981) have found that independence kicks in for particles larger than 100 microns. While others (Mills and Mason, 1981) say that the cut off occurs at 50 microns. Finally, Gandhi et al., 1999 found that the erosion rate is always affected by particle size, albeit "weakly." However one study (Finnie, 1995) quantified the relative effect of particle size on erosion rate and states that a 10-micron particle is only 25% effective as a 100 micron size. This wide range of results seem to be confusing, but in the context of the present need, all the studies seem to imply that for particles of less than 60 microns a change in size will affect the erosion rate. (See also Fig. 8-6 in Shook and Roco, 1991) This particle-size dependence makes using results from other studies less appealing.

A further complication that is not addressed by most studies is the fact that in real systems the transported particles have a nonuniform size distribution. Usually stated is the mean particle size, without giving information on the actual distribution of particle sizes within a group. However, a group of particles having all the same size and another, which has a wide range of sizes - with a mean size equal to the uniform-size group, may give different wear results. In fact, wear results from the two different groups of particles would not be expected to match because the interaction energy and the rate of material removal are nonlinear with particle size. Some experiments performed for particles with a broad quasi-logarithmic size distribution suggest that the "equivalent wear diameter" for slurry pipes or pumps is larger than the mean particle size (Roco and Cader, 1990; Roco and Minani, 1989). The "equivalent wear diameter" refers to a particle diameter that is assigned to a group of particles, which has a range of sizes, that would cause the same wear rate as a group of particles that all have that same (assigned) size.

The effect of particle shape on the erosion rate is not as well investigated as size, but some information exists. Sundararajan and Manish, 1997 showed that friable particles tend to cause high erosion rates because newly broken particles contain sharp edges that cut more easily into surfaces. Elmore, 1994 showed that for a closed slurry loop the erosion rate decreases as the particle become rounded from frequent collisions. However, Zhong and Minemura, 1996, claim that the effects of particle velocity, concentration, size, and shape on the rate of erosion is still a relatively unexplored field.

A harder particle does lead to an increased wear rate but it is the relative difference of the hardness of the particle to that of the wearing surface, which is important. Sundararajan and Manish, 1997 state that if a particle is more that twice as hard as the eroding surface, then the erosion rate is independent of the particles hardness. For the case under study the particles’ hardnesses are 2.5 to 7.5 Mohs. The two possible stainless steels to be used have a hardness of 4-5 Mohs. In this case, for any existing database to be useful the particle hardness would have to be considered.

When investigating the effect of velocity, care must be taken to know which velocity is being discussed. While the velocity of particles is generally always mentioned, many times it is the velocity of the slurry and not the particle that is measured. It is not clear that all authors are cognizant of that fact, which can lead to confusion. Iwai and Nambu, 1997 found that the erosion rate with a jet of sand-water slurry was related to velocity by (v-v₀)ⁿ where v₀ is a critical velocity of the jet, which is dependent on particle size and below which they could not measure the erosion rate (but their threshold was very high, i.e., 5 microns/hour = 1.7 inch/year). (When the slurry jet velocity, v, is less than the critical velocity, v₀, this equation is indeterminate and no longer valid.) The critical velocity was given as 10 m/s and 8.5 m/s for 91 and 323 micron-size particles, respectively. The exponent was stated to be independent of particle size but is dependent on the eroding surface material (it was given as 2.7 for 304 stainless steel). For alumina particles in water, Stack et al., 1997, showed that at 3 m/s and a concentration of 20 wt%, corrosion plays just as large role in the wear of 304 and 316 stainless steels as erosion. Umemura, et at., 1987 found that for 40-micron alumina particles in an acidic solution with velocities from 1.5 to 5 m/s onto a 304 stainless steel surface, the corrosion played less of a role to wear at the higher velocities. That is, at 1.5 m/s, corrosion caused 1/3 of the total measured wear, but at 5 m/s the contribution of corrosion dropped to 20%. For pumps in acidic solutions van Bennekom et al., 2001 states that there is ‘some’ threshold velocity (called the breakdown velocity) above which the substrate of an abraded surface is exposed. This exposed surface now will have a high wear rate due to corrosion and as the density of particle increases the breakdown velocity decreases. Unfortunately, no breakdown velocities are given. Gandhi et al., 1999 measured the erosion rate of steel plating by a jet of sand and water. The particles ranged from 200 to 900 microns and the solids concentration from 20 to 40 wt%. Using velocities from 3 to 8 m/s they determined that the erosion rate = ƒ(velocity^2.6). However, when changing the material of the eroding surface the exponent of particle velocity also changes. Further, Stack and Pungwiwat, 1998 found that the exponent varies from 0.5 to 5.4 for surfaces that ranged from polymers and ceramics to metals.

There are many reasons why the erosion rate’s dependence on velocity varies. Probably the main reason is not knowing true particle velocity. As already stated, many studies simply assume that a particle velocity is the same as the fluid velocity. Humphrey, 1990, stated that "for systems containing fluids it is incorrect to assume that the particle’s velocity at the instant of impact is equal to the surrounding fluid’s velocity." Another reason for the large variation in the velocity dependence of erosion rate is from the synergistic effect of both erosion and corrosion. Stack et al., 1997 stated that "…the velocity dependence for (steels) varied significantly," due to "different erosion-corrosion regimes…" Some studies did not attempt to determine the actual particle velocity and simply based experimental results on a measured slurry velocity. One research project (Mills and Mason, 1981) points out that particle velocity is indeed the important parameter to measure but then measured slurry velocity because it is "more appropriate to industry." Their results showed a wide range of velocity exponents to estimate erosion. Clark et al., 2000 discuss a rotational velocity and relate a slurry wear rate to an exponential function of that velocity. (This rotational velocity is not a particle velocity, it is the angular velocity of a spinning disk in a Coriolis tester. That tester works by forcing a slurry through two, oppositely oriented, small channels that are on a spinning disk with centrifugal force and uses the Coriolis force to direct particles towards a surface to be eroded. Even so, the particle velocity is dependent on the rotational velocity of the tester. See Clark et al., 2000 for more details.) In trying to correlate the rotational velocity to a wear rate they did not obtain a single value for the velocity exponent and attributed the lack of a constant value to a changing environment in the slurry flow. That is, when one parameter is changed, like velocity, other parameters change too. For instance, they specifically mentioned how the thickness of the particle bed near the eroding surface changes, which affect the velocity-wear rate relationship. Clark et al., 2000 conclude that unless all other parameters are held constant, a single relationship between velocity and wear rate is difficult to obtain.

Some authors have tried to relate slurry velocity to wear rates. In measuring slurry velocity Shook et al., 1990 found no dependence of erosion rate on velocity. They used sand particles from 160 to 900 microns in water at velocities from 1 to 2.5 m/s at a concentration of 20 wt%. As the slurry velocity increased, more solids are suspended in the slurry flow increasing homogeneity. This increase in solids suspension minimized the heavier concentration of solids at the bottom of pipes which in turn decreases the erosion there. The net result was that the overall wear rate was not affected by slurry velocity.

Like velocity, it is important to differentiate between the movements of solids particle and the slurry. When direction is studied/discussed, it generally refers to a specific direction of the solids particle to the surface being affected. Most studies that deal with the direction of a particle towards an eroding surface fall into two categories: particle trajectory angles that cause ductile wear or brittle wear. Those types of wear will not be discussed here but they refer to the way material is removed from a surface. Ductile wear is defined when a surface has the highest wear rate at an impingement angle of about 30° and brittle wear is at an angle of about 90°. Finnie, 1995 states that ductile wear occurs between 20-30°, but he adds that it is always the predominate type of wear when particles are less the 10 microns in diameter and move at "slow" velocities. Using 304L stainless steel Burstein and Sasaki, 2000 state that sand particles attacking surfaces at oblique impingement angles (40-50°) remove the passive oxide layer more effectively than at 90°. This result in not surprising because many other papers indicate that stainless steels generally result in a ductile type wear (at ~ 30°). In fact, Foley and Levy, 1983 showed that 304 stainless steel does indeed wear fastest with a particle angle of 30°. Singh et al., 1991 showed that both 304 and 316 stainless steels have the same rate of wear when impinged with an air jet containing SiC particles that were 160 microns in diameter, and had angular shapes. Both metals wore the fastest when the impingement angle was at 30° and it was the slowest at 90°. This information is very useful when designing a test because it indicates where attention must be directed to evaluate the maximum wear locations. That is, wear measurement must not be concentrated only at a section of a flow loop where the flow makes an abrupt 90° change. (This was well demonstrated in the previously explained experiment where Smith and Elmore, 1992, Fig. 1, studied the wear on a steel specimen from perpendicularly (90°) oriented slurry jet, only to find that another steel specimen, which only received oblique-angle particle attacks, unexpectedly showed more wear.) The conclusion is, for ductile eroders, like stainless steel, measurements should be made where particles hit a surface at lower angles, e.g. at the sides of a T fitting as opposed to its back portion where the flow comes to almost a halt.

There is not a lot of information on the effect of changing the temperature of the eroding surface on the wear rate. After a severe piping rupture in 1989 Ting and Ma, 1999, evaluated the wear rate in a Taiwanese nuclear plant and they found that it increased exponentially with temperature. (During several scheduled plant outages they measured the wear ultrasonically at many locations in over 300 pipe components to include: 90° and 45° elbows, expanders, reducers, tees, and straight pipe sections). At 100°C the wear rate was 2.5 mil/year and it increased to 40 mil/year at 180°C. However, most of their wear came from bubbles collapsing in two-phase flow and not from a true slurry. The wear rate does increase with surface temperature but it generally occurs due to secondary reasons. For instance, Tuerberg, 2000 explains that above 50°C stainless steels like 304 and 316 are more susceptible to stress corrosion cracking for slurries that contain compounds like chlorides and flourides. As a surface weakens due to cracking it, more than likely, is weakened further and faster from a constant stream of solids particles.

Two aspects of the material of an abraded surface have already been mentioned: The erosion rate for surfaces that are twice as soft as the particles in a slurry is independent of the surface hardness (Sundararajan and Manish, 1997), and the erosion rate will be affected if the surface material is susceptible to chemical attack by the slurry (Tuerberg, 2000). Stack and Pungwiwat, 1998 state that ceramics abrade faster than steels when the particle impingement angle is 90°, that is, when a particle strikes a surface perpendicularly. They also conclude that erosion rate of surfaces like Teflon are dependent of a particle’s composition more than for stainless steels. That is, the erosion rate of stainless steel is less sensitive to the material of an impinging particle. However, in general, Stack and Pungwiwat, 1998 conclude that independence of a wear rate on particle size or particle velocity cannot be stated without considering the "particle/target interaction." Zheng et al., 1995 showed the importance of the synergistic effect of corrosion and erosion to a material worn by a slurry. They used an acidic slurry (10 wt% of H2SO4 + 15 wt% of corundum sand with particle sizes from 250-325 microns) against 316L and 321 stainless steel surfaces. For the 316L stainless steel, at a slurry velocity of 2.5 m/s, there was no measurable wear when only erosion was available and almost no wear by just corrosion (0.2% of the total). However, when both wear mechanisms were present there was significant wear; it was 99.8% of the total measurable wear. When the slurry velocity was increased to 5 m/s the measured wear was 38.6% from erosion, none from corrosion, and 61.4% from the synergistic effect of both.

See the preceding section on a particle’s direction. As already explained in that section, many researchers set up tests to obtain very specific angles of attack of a particle against a surface to determine the wear-rate dependence. Surfaces that wear fastest at impingement angles of 60-90° are considered brittle/hard materials like basalt and at angles of 0-30° are considered ductile/soft, like elastomers (Shook and Roco, 1991). Of interest is stainless steel and some have found that 304 (Foley and Levy, 1983) wears the fastest at 30° and others found 304L (Burstein and Sasaki, 2000) wears the fastest at 40-50°. This indicates more of a wear response between ductile and brittle materials. However, stainless steels (e.g., 304, 316, 410) tend more to the ductile type of wear than the brittle type, as shown by others (Stack et al., 1997; Singh et al., 1991).

Many materials are subject to processes that affect their surfaces. For instance, metals that are cold worked or heat treated will have different properties on the surface than in their interior. Case hardening or work hardening will make many metallic surfaces harder, but this is usually at the expense of becoming less ductile and more susceptible to erosion. These treatments may have been intentional, done at the factory, or be the result of material’s use, e.g., when a surface is subjected to a continual pounding which leads to work hardening. With respect to wear, especially from erosion, the condition of an abrading surface may have an effect on the rate at which the surface wears. Elmore, 1994 postulated that the 304L surface in the flow loop became work hardened with time because the wear rate decreased. However, he also showed that the particles’ ability to cause wear reduced with time (shown with a separate wear tester), which complicates the question as to what was the true mechanism of the reduced wear rate. Contrary to that evidence, Abbade and Crnkovic, 2000 found that changing the surface hardness did not significantly affect the wear rate. They used a low carbon, niobium-titanium, steel (API 5L X65) in annealed and heat treated conditions, which doubled the metal’s yield and tensile strengths. Against this metal they directed a jet of slurry at 4.5 m/s, which was made of water that contained 3 wt% of sand (the particles were rounded and had sizes from 150 to 300 microns). The fastest wear occurred at an angle of 30° and it was slowest at 90°, which implied a ductile-type wear, but both the heat-treated and non-heated surfaces displayed the same wear rate, for the same particle direction. Another study (Dasgupta et al., 1997) showed a slight reduction (~30%) in wear rate when a steel substrate was hardfaced with another steel, which increased its hardness by a factor of 4. However, that result, with a water and sand (30 wt%) slurry, was only obtained at one velocity; at other velocities the data seemed to be inconclusive. Others (Xie et al., 1999; Imrich et al., 1999), who coated stainless steel with harder materials, like Ultimet or Stellite, have shown a 20% to 50% (or more) decrease in wear rates; therefore, it seems that if a surface starts out harder a reduction in the rate of wear is possible, but then again, those coated surfaces were not actually made hard, they were made different, which complicates the matter. Once again, existing data seem to be at odds.

Foley and Levy, 1983 performed an erosion test with as-rolled and annealed 304 stainless steel. They found that the harder "as rolled" stainless steel was 25% less abrasion resistant than the annealed steel. In other words, the more ductile a surface is the more abrasion resistant it is. To complicate the matter further, Mishra and Finnie, 1981 discuss tests that used air with silicon carbide particles against many different materials. They claim that the particle size relative to the thickness of the hardness layer is important; the hardness layer is approximately 20 microns. (Imrich et al., 1999 found a hardness layer for Ultimet close to 75 microns.) This thickness mitigates erosion from particles that are less than 100 microns. However, larger particles have sufficient momentum to break through the thickness of the hardness layer to increase the wear rate.

Another form of surface heat treatment occurs when its temperature changes from releasing heat due to the plastic deformation that occurs as particles plough, dig, and displace material, however the effect is not well understood (Mishra and Finnie, 1981). A recent study (Sasaki and Burstein, 1996) looked at the effect of surface roughness on wear. They used 304L stainless steel with different roughness determined by the silicon carbide grit paper used to prepare a surface. The specimens were subjected to a 3.8-m/s jet of a slurry of 0.6 M NaCl with 13.2 wt% of sand particles that ranged from 425 to 600 microns in size. They found that for surfaces prepared to a smoothness finer than what can be obtained with a 600-grit paper, those surfaces became rougher when subjected to the slurry flow. Further, for surfaces that were made rougher than that produced with 600-grit paper, those surfaces became smoother by the slurry flow. They also determined that after the slurry flow was initiated, the entire metal specimen surface (with either a rough or smooth finish) only needed 20 seconds to be completely filled with scratches, and after 60 minutes, the original surface roughness was completely obliterated by the flowing stream of particles. Likewise, after 60 minutes the surface hardness increased to an asymptotic value of 60% above its original value. However, the increasing hardness does not significantly affect the surfaces’ potential to pit and therefore wear. Moreover, the rougher the surface the more susceptible it is to wear. Since oblique impingement angles of particles to a surface cause more roughness on ductile materials, the wear rate is higher for these materials then when they are attacked by particles at angles closer to 90°.

The temperature of a slurry affects erosion by causing the eroding surface to become more susceptible to wear. For instance, Tuerberg, 2000 discusses how certain stainless steels have good corrosion resistance to chlorides and flourides, as long a certain temperature is maintained, e.g., for 304 at room temperature and 316 below 50°C. However, above these temperatures the metals may experience stress corrosion cracking. Combined with the erosion from solid particles, the weaker metals will wear faster. The slurry temperature also changes the slurry viscosity, which affects the erosion rate. Clark, 1990 discusses how the erosion rate decreases as the viscosity increases. Conversely, as the temperature increases the slurry viscosity decreases, leading to more wear. One reason that this increase in wear may occur is because the drag force between the liquid and solids decreases, and therefore the particles will have more of a tendency to hit the surfaces.

As already stated in the preceding Particle Velocity Section: Flow rate of a slurry is differentiated from the particle velocity because the two parameters are generally distinct. Only under very specific conditions will they be the same. Despite this fact, there is a large body of research that makes that assumption. Some of those studies are discussed below:

Many studies (see the Particle Velocity section) have quantified some functional relation between erosion rate and particle velocity and it is generally Vⁿ_particle where n~2, but the exponent can be many other values too. For those (Aiming et al., 1996; Gupta et al., 1995; Shook et al., 1990; and Clark, 1990) who have studied the relationship between slurry flow rate (or velocity) and erosion rate, the wear rate depended on what they measured. Aiming et al., 1996 relates erosion rate to a flow velocity and states there is the "well known" exponential relationship. However they used a "pot" tester where the metal specimen is spun in a container of slurry and the velocity is really of the specimen against the particles suspended in a pool of slurry. Clark, 1990 discusses the wide range of exponential values given for the "particle" speed and states that one of the reasons for this variable range is the difficulty in relating the rotational speed of a pot tester to the true particle speed. Shook et al., 1990 found that the erosion rate "did not correlate with slurry (mean) velocity," in contrast to other studies that claim a dependence on "velocity raised to an exponent of two or more." In fact, they show the lack of correlation between velocity and erosion rate comes from that fact that as the slurry velocity increases the particulate distribution becomes increasingly homogeneous. The heavier particles that were on the bottom of a pipe become more entrained in the overall flow stream which changes the erosion environment; increasing the erosion rate in certain locations and decreasing it at others. The net effect is that the erosion rate’s dependence on slurry velocity is complex and cannot be separated from other factors like slurry solids concentration and particle size. However, Shook, et at., 1990 (see their Eq. 1) imply a direct proportional dependence of erosion rate on slurry velocity.

Turbulence is the general flow regime of slurry flows. Laminar flows are possible but the solid particles must be light enough to be carried by the fluid. However, in most industrial applications the particles are usually heavier than the fluid and the flow must have enough energy to maintain them in motion. Humphrey, 1990 states that despite the importance of flow motion, little attention has been given to clarify its influence, especially in the turbulent flow regime. Probably because of the complexities of turbulence he further states that, "many researchers continue to interpret and attempt to understand particle impact erosion almost exclusively in terms of the material properties." For example, on the importance of turbulence, Dosanjh and Humphrey, 1985, show, through a computational study of a particle-laden jet impinging against a flat plate, how the location of maximum erosion is displaced toward the stagnation point as the turbulent intensity increases.

Turbulence effects have been studied for some slurry flows, but they seem to be limited to straight channels. Even though Sharp and O’Neill, 1971 studied turbulent slurry flow in a horizontal pipe, the turbulence was only discussed in terms of the homogeneity of the turbulent core in a pipe. Their work was then limited to this core region where "the simplest flow environment existed." They concluded that the solids-particle gradient only occurs vertically; the particles were equally distributed horizontally. Shook et al., 1968 go further into trying to understand slurry flow from the point of view of turbulence in a straight pipe. They evaluated a term, e_s, called the eddy diffusivity of the mass-exchange of particles and "the lifting and settling tendencies at a point in the flowing mixture." Their investigation concluded that e_s "to be of the same order of magnitude as, e_m , the eddy diffusivity for momentum and mass in a single-phase fluid," however to use the Reynolds analogy to obtain e_s from e_m is not recommended without more work on the effects of high-solids concentrations. Further, they found that the effect of higher solids concentrations is not to ‘damp’ the turbulence, as others claim, but does increase the suspending capacity of the fluid."

As explained in the preceding section, 3.2, in conjunction with Figs. 1, 4, and 5, turbulent secondary flows, especially trailing vorticies, are very effective in creating an environment where the erosion rate can be increased, many times, unexpectantly. The better turbulence is understood, the better a flow regime can be understood and predicted. The better the effect that a flow regime has on the movement of solid particles in flowing slurry, the better that wear and wear rates can be estimated. With the current state of knowledge on how a flow regime affects the rate of wear, only experimental measurements can provide accurate information.

In the presence of a combination of a flowing liquid and suspended solid particles, both mechanical wear and chemical attack to a surface may be important. In fact, it is clear from the literature that to understand slurry wear, both the effects of corrosion and erosion are important. The synergistic effect of the two generally have a more significant effect than either effect individually. Aiming et al., 1996 studied the effects of a pH=1 solution with a 74-micron gypsum particle loading at 28.6 wt%. For a 6 m/s slurry jet, they found the wear of a stainless steel specimen, similar to 316, at 40°C to be comprised of 7.2% from just corrosion, 47.5% from just erosion, and 45.3% from the synergistic effect of both. For the same setup at 80°C the percentages were 35.1%, 29.7%, and 35.2% respectively. Zheng et al., 1995 did a similar evaluation using 316L and another stainless steel, 321, similar to 304. Here, their slurry was made of another strong acid (10 wt% H2SO4) with 15 wt% of 250-315 micron-size particles. For "flow" velocities from 2.5 to 10 m/s, they found the wear of 316L to be similar to that of 321 stainless steel. At 2.5 m/s the wear was 0.2% to 0.6% from just corrosion, 0% to 3.8% from just erosion, and 94.6% to 99.8% from the synergistic effect of both. At 5 m/s, the wear was 0% to 0.5% from just corrosion, 36.7% to 38.6% from just erosion, and 61.4% to 62.8% from the synergistic effect of both. Umemura et al., 1987 did a study at a particle impingement angle of 30° with 304 stainless steel and a slurry with a pH that ranged from 2 to 7 with 40-micron solid particles of alumina. They concluded that as the slurry velocity increased from 1.5 m/s to 5 m/s the corrosion contribution to wear dropped from 33% to 20%. These numbers do not seem to conform to the two preceding studies, which show a much lower contribution from just corrosion of stainless steels.

Contrary to the previous studies which used acidic slurries, the work for the present study involves caustic slurries. Any simulated slurry is expected to have pH~14 (Eibling and Nash, 2000; Elmore, 2000). According to Tuerberg, 2000, the higher the pH, the better stainless steels can resist pitting corrosion. Mickalonis, 2001, showed that 304L and 316L stainless steels have considerable resistance to pitting (< 0.1 mil/year) for a slurry at pH=14. Stack et al., 1997 showed the effect of pH by using a water and alumina-particle slurry (100 to 500 micron). They adjusted the pH, slurry jet velocity, and slurry concentration. For pH>7 and "particle" velocities greater than 0.2 m/s, they found that erosion dominates slurry wear for mild steels.

There are many slurry wear studies that discuss, or at a minimum measure, a slurry solids concentration. However, an erosion rate’s dependence on a slurry mean solids concentration is not clear; some claim a dependence, other claim none, and others are somewhere in between. The following information is a summary of some of the available research:

Studying solids-laden gas flows in pumps, Zhong and Minemura, 1996 state that cast iron and stainless steel erosion decreases with particle concentration. However, the dependence is not strong. Evaluating the effects of various liquid slurry flows in pumps, van Bennekom et al., 2001 qualitatively state that "there is a linear increase in the rate of material damage with increasing concentration of solids up to a certain level," after which the rate of damage is slower at "high concentrations." These results seem to be in conflict with those from a study of sand and water flows in pumps by Walker and Bodkin, 2000. For particle sizes from 150 to 1000 microns and impeller tip speeds of 18.8 and 23.4 m/s, they show that with solids concentrations from 10 to 34 vol% the erosion rate is independent of the concentration. However, Walker and Bodkin, 2000 admit that their measurement uncertainty was too large to quantify a dependence, if it existed. In justifying their work they state that the dependence of wear rate on "the different variables involved is not generally well understood," and that studies of wear in pumps are limited.

Andrews and Horsfield, 1983 did a test with a jet of gas with solids particles to study the mechanics of an eroding surface. They stated that increasing the particle concentration causes a decrease in the erosion rate due to the interference of the particles themselves. That is, the particles already at the surface serve as a buffer against the following particles. Xie et al., 1999 determined that for very dilute slurries, where the solids concentration is less than 1 vol%, or when the particle-to-particle distance is greater than 20 times a particle diameter, the effect of the solids concentration can be neglected, because the particles do not interfere with each other. Clark, 1990 suggested that experiments show a solids concentration of greater than 5 wt% is necessary to have significant particle-to-particle interaction.

In determining a standard to measure erosion, Miller and Miller, 1993 (see their Fig. 8) showed that erosion rate increases rapidly as the slurry concentration increases to 10 wt%, but after 20 wt% the erosion rate dependence is relatively unaffected by further increases in concentration. (This is a standard of the American Society for Testing and Materials: ASTM G75-89, "Standard Test Method for Determination of Slurry Abrasivity (Miller Number) and Slurry Abrasion Response of Materials (SAR Number)." The standard is the result of a test apparatus developed by John E. Miller "in 1967 to determine the degradation of slurry abrasivity in a closed loop pump test of Savage River Iron Slurry." Today many companies used the standard to indicate slurry abrasivity to help in the selection of slurry handling equipment, especially pumps.) In fact, after reaching 40 wt% the erosion is independent of the slurry solids concentration. However, the slurry in their tester (which maintains a metal specimen in a fixed pool of slurry) is in a completely different environment than a slurry flowing in a pipe where the particles are free to diffuse into varying particle distributions. For instance, using a sand and water slurry, with large particle sizes (> 200 microns), solids concentrations of greater than 20 wt%, and moderate velocities (3 to 8 m/s), Gandhi et al., 1999 showed that erosion rate is only weakly dependent on solids concentration and that the dependence was also a function of the slurry velocity.

Care must be taken when discussing the slurry concentration in pipe flow and relating it to something like a wear rate because, when the solids are of a different density than the carrying fluid, the particle distribution is generally not uniform. Shook et al., 1990 and Shook et al., 1968 measured the particle distribution in a flow of sand and water in a pipe. The distribution was not homogeneous and higher particle concentrations occur in the bottom of the pipe. That is, for a fixed mean solids concentration, as the slurry velocity increases the particle distribution becomes more uniform, which results in less wear at the bottom of pipe and more near the top and sides of a pipe. The distribution phenomenon is true even when the particles have a density close to the carrying fluid, which was shown by Sharp and O’Neill, 1971 with plastic particles (specific gravity = 1.035) in water. For another example, Roco and Shook, 1983, showed that a slurry of water and sand in a 2-inch pipe, with a mean particle size of 165 microns, a mean velocity of 4 m/s, and a solids concentration of 28.6 vol%, the vertical concentration ranged from 25 vol% at the top of the pipe to 35 vol% at the bottom. For a 10-inch inside-diameter pipe with a mean slurry velocity of 3.5 m/s and a mean concentration of 26.8 vol%, the solids distribution ranged from 10 vol% at the top to 38 vol% at the bottom. While there is a defined vertical distribution of solids in horizontal pipe flow, which depends on the slurry velocity, solids are evenly distributed horizontally from the vertical centerline (Nasr-EL-Din et al., 1987).

Finally, the non-uniform particle distribution in horizontal pipes affects how long a flow takes to become fully established; heavy particles need less time. Colwell and Shook, 1988 showed a 4-m/s flow with 900 micron particles (specific gravity = 2.65) needs 50 pipe diameters to become fully established, but for a slurry which contains lighter particles (specific gravity = 1.05) the particle distribution is still developing after 185 pipe diameters!

The solids concentration is a complicated parameter for erosion rate. There does not seem to be clear method of determining wear rates from knowing the concentration. However, knowing how solids behave as the concentration changes in a flow is necessary to understand where wear occurs.

Clark, 1990 summarized studies that looked at the effect of viscosity on erosion rate and comments on how the rate is much more dependent upon viscosity for liquids, than it is for gases. This claim is not surprising but he goes on to say that this dependence comes from two sources. Due to the large drag forces set up by the more viscous carrier fluids, many of the particles do not hit surfaces. This effect may reduce the erosion rate by "two or three orders of magnitude" over gas-carrier systems. Another reason why the higher liquid viscosity reduces the erosion rate is due to the viscous environment in the boundary layers. These viscosity effects are alluded to by Xie et al., 1999 who talk about particles being dragged and not being able to reach a surface as a slurry concentration increases, which can lead to higher slurry viscosities. This effect is also mentioned by Shook et al., 1990 who talk about fluid movements between a surface and a particle causing a repelling force, which leads to less wear.

It is clear from the preceding section that erosion rate depends on many factors, many of which are not well known, understood, complex, or have multiple dependencies. Computationally estimating hydraulic parameters for even most single phase and all two-phase turbulent flows is still a very difficult task. When adding the complication of corrosion/erosion variables the estimation capabilities of numerical schemes is even more challenged. There have been considerable efforts to predict the effect of erosive flows. However, to date numerical wear predictions are generally based on semi-empirical formulas that are "limited by the experimental base used…and therefore extrapolations are not recommended." (Shook and Roco, 1991). For instance, Dosanjh and Humphrey, 1985, had some success in relating numerical calculation to their study of a particle-laden jet against a flat plate, however those results could only be used to predict the erosion for the specific setup they used. There have been some successful numerical applications with using bench-top wear testers (e.g., Pagalthivarthi et al., 1998; Zhong and Minemura, 1996) which then are related to specific pieces of equipment like a pump of interest. Further, there has been work in the prediction of erosion in heat exchangers from fluid-solids systems using three-dimensional models, but those efforts continue (Lyczkowski et al., 1994). Humphrey, 1990 states that "we can increasingly expect to see phenomenological modeling approaches being complemented by direct numerical simulation techniques capable of more accurate representations of particle-laden turbulent flows." Notwithstanding Humphrey’s statement made more than ten years ago, modeling of particle-laden flows is still in its infancy. For an example, a modeling effort as late as 2001 (Wu, 2001), using a dilute mixture of gas and solid particles that was flowing in a 750-micron round channel, attempted to track the movement of 750 to 4000 particles. Those particles were spheres, they all had the same size, and due to the no slip boundary condition they stuck to the channel wall once contact was made. That model takes approximately 50 hours of CPU time to converge on an established flow pattern. (This model was actually a series of models which combined "a mechanistic Eulerian-Lagrangian model, based on Reynolds-Averaging Navier-Stokes turbulence models," with "particle dispersion models applicable to microscopic particles.") To then try to determine a particle’s interaction with an eroding surface, along with its interaction with all the other particles seems to be a very difficult, if not impossible, problem. A better approach would be to treat wear process stochastically (Shook and Roco, 1991). By studying the particle interaction with a wall probabilistically there is a better chance to develop usable computational models in the near future. There is definite progress ongoing, albeit at a slow pace.

Computational tools will definitely become increasingly better, but to date the complexities in slurry flow still presents barriers that put accurate slurry wear prediction beyond reach without experimental benchmarking, except the most elementary flow situations. Where computational assistance has always been invaluable is in its flexibility to do parametric studies of simplified flow situations. For instance, scaling of results is always a complex and difficult task. To save time and expense experiments are commonly done at small scale, but the path to relating small-scale results to large-scale effects is only as good as the engineering judgement used in doing the extrapolation. For experimental wear results that indicate only small changes in an eroded surface, e.g., less than a few percent of a pipe wall thickness, or a wear rate that is either linear or slowly increasing, e.g., not exponentially, then computational assistance would greatly increase the confidence in extrapolating results. This would be especially true if the experimental results were done at more than one small scale, so the computational models could be benchmarked and appropriate extrapolations could be developed.

Despite the large body of data that exists in the literature, there seems to be no way to avoid doing experimental tests to elicit accurate results on wear rate in a specific flow loop with a specific slurry. The number of parameters to consider, the complexity of the wear mechanisms, and the lack of comprehensive computational tools to estimate wear, makes wear estimates without testing questionable. As already mentioned, some work has been done to relate bench-top measured wear rates to wear that is seen in the field, but most of that was relating very specific pieces of equipment (Pagalthisvarthi et al., 1998; Gupta et al., 1995; Pagalthisvarthi and Helmly, 1992; Miller and Miller, 1993). The state of wear knowledge is not at the point where a series of parameters like particle size, shape, hardness/material, velocity, etc.; slurry chemistry, solids concentration, flow rate, temperature, etc.; and flow loop orientation, material, internal surface finish, etc., can be input and out comes a wear rate. While there does exist enough data to estimate the wear expected in many straight pipe slurry flows under varying conditions, flow systems are always more complicated and, in general, wear in straight pipes is the least of all the flow loop components; making that information less useful.

A flow loop with a prototypic slurry, employed under prototypic flow conditions, should be used to accurately estimate wear. Care must be taken if wear rates are obtained at scales other than full size. However, if the wear rates are low, then many parts of the flow loop could be subjected to computational modeling that would be very useful in estimating full-scale results. At the very least, computational assistance could determine similar flow patterns in small- and large-scale loops, which would indicate how small-scale flow loops need to operate to set up similar flow patterns, and subsequently, similar erosion patterns.

Lastly, Figs. 1, 4 and 5 make it abundantly clear that erosion may occur where not expected. Care must be given to any flow loop constructed to estimate wear so that unusual protrusions are included into the flow stream. Those obstructions to flow may be the locations where the largest amount of wear occur.

Literature indicates that wear from the flow of slurry is a complex process, dependent upon many parameters. Many of these are still not well understood today. In fact, it appears that there are many inter-dependencies among some of those parameters, making a fundamental understanding very challenging. In light of this information the best way to quantify wear and wear rates is to perform a full-scale test, which uses prototypic slurry under prototypic flow conditions. Conversely, unless existing data came from an experimental setup that is very similar to a flow system of interest, they should not be used to estimate wear. (A " very similar" flow system cannot be strictly quantified, but good engineering judgement is needed to make such a determination.) However, those data are very valuable in designing a test flow loop, and for knowing where and how to make measurements.

• A scaled test should use a slurry that is prototypic with respect to particle size, particle concentration, and overall slurry chemistry. If the flow loop is to contain many different slurries, then one can be selected that would be the most abrasive. This can be determined by a bench-top wear tester, e.g., Miller tester (Miller and Miller, 1993).
• A scaled test should use prototypic flow conditions. However, obtaining similar conditions at different scales may not be obtained by a simple linear scaling of all aspects of a flow system, e.g., geometries, slurry velocities, temperature, etc. It is very important to look at each piece of a flow loop individually and attempt to understand the movement of the slurry and its particles at each scale. This is where the use of available literature and computational tools is extremely important. Past experiments can indicate where wear may occur and turbulent modeling can compare flow regimes at different scales. Both sources of information give valuable insight into the proper scaling of equipment and flow conditions.
• A flow loop should be designed such that there is more than one scale, for similar pieces of equipment, so that computational modeling predictions can be compared to available experimental data. These comparisons will allow an accurate benchmarking of models to experimental data, which in turn will produce a better estimate of wear and wear rates at larger scales.

1. Barnes, S.M., 2001. West Valley Demonstration Project. Personal communication on January 9, 2001.
2. Burns, D.B, Hope, M.C., and Occhipinti, J.E., 1991. Mott crossflow filter test – Phase 1 & 2. Westinghouse Savannah River Company Inter-office Memorandum No. WSRC-TR-91-131, (March 20), addressed to M.G. Looper.
3. Carstens, M.R., 1979. Slurry tests in 3-in. diameter pipe. Internal report of the Georgia Ironworks Laboratory of the Georgia Ironworks Industries, Inc. (Grovetown, GA) (November).
4. Carstens, M.R., 1982. Flow characteristics of simulated slurries. Internal report of the Georgia Ironworks Laboratory of the Georgia Ironworks Industries, Inc. (Grovetown, GA), also document under cover letter: DuPont Savannah River Site Report No. DPST-82-955 (September 1982).
5. Danielson, M. and Pitman, S., 1999. Results of the non-radioactive corrosion test for BNFL. Attachment of Battelle Pacific Northwest National Laboratory Transmittal Letter No. 005220 from E. Morrey to M. Johnson dated July 28, 1999.
6. Eibling, R.E. and Nash, C.A., 2000. Hanford waste simulants created to support the research and development on the River Protection Project – Waste Treatment Plant. Westinghouse Savannah River Company Report No. WSRC-TR-2000-00338.
7. Elmore, M.R., 1994. TWRS characterization program: slurry abrasion testing with a slurry pipeline pilot plant facility. Battelle Pacific Northwest National Laboratory Report (not published).
8. Elmore, M.R., 2000. Simulant erosion testing. Battelle Pacific Northwest National Laboratory Report No. WTP-RPT-001, October.
9. Gee, J.R., Chandler, C.T., Daugherty, W.L., Imrich, K.J., and Jenkins, C.F., 1997. Erosion/Corrosion concerns in feed preparation systems at the Defense Waste Processing Facility. Presented at the NACE Annual Conf. – Corrosion ’97 Symposium Proc. At New Orleans, 3/9-14/1997, (also listed under the Westinghouse Savannah River Company Report No. WSRC-MS-96-0363).
10. Hodgson, K.M., 1995. Tank characterization report for double-shell tank 241-AZ-101. Westinghouse Hanford Company Report No. WHC-SD-WM-ER-410 (July 26).
11. Imrich, K.J., Sides, B.K., and Gee, J.T., 1999. Corrosion/erosion resistance of ULTIMET^â R31233 in a simulanted feed for a radioactive vitrification facility. Westinghouse Savannah River Company Report No. WSRC-MS-98-00655 (May 21).
12. Lewis, D.P., 1982. Hydraulic testing of simulated DWPF waste slurries at the Georgia Irons Works Hydraulic Laboratory: summary report. DuPont Savannah River Site Report No. DPST-82-954. (Transmitted as an attachment to the letter: Sludge Slurry Hydraulic Data. Number DPST-82-954-TL, dated April 15, 1983.)
13. McGrail, B.P., 1991. Results of the analysis of the large chunk of material and the measurement of the Miller number for DST-101-AZ Core #3 Waste. Battelle Pacific Northwest National Laboratory Transmittal Letter No. 9101055 from B.P. McGrail to L.M. Sasaki on February 11, 1991.
14. Mickalonis, J.I., 2001. Pitting potential of 304L and 316L stainless steel in a Sr/TRU supernate. Westinghouse Savannah River Company Inter-office Memorandum No. SRT-MTS-2001-20007, to M. R. Duignan on March 22, 2001.
15. Petkus, L., 2001. West Valley Demonstration Project. Personal communication on January 9, 2001.
16. Smith H.D. and Elmore, M.R., 1992. Corrosion studies of carbon steel under impinging jets of simulated slurries of neutralized current acid waste (NCAW) and neutralized cladding removal waste (NCRW). Battelle Pacific Northwest National Laboratory Report No. PNL-7816 (also UC-721), (January).
    
    

General and Miscellaneous Studies

17. Blau, P.J., 1997. Fifty years of research on the wear of metals. Tribology International 30, No. 5, 321-331.
18. Clark, H.McI., 1990. Slurry erosion. Proc. Corrosion-Erosion-Wear of Materials at Elevated Temperatures, Berkeley, CA (Jan. 31 – Feb. 2), (10-1)-(10-14).
19. Finnie, I., 1995. Some reflections on the past and future of erosion. Wear 186-187, 1-10.
20. Humphrey, J.A.C., 1990, Fundamentals of fluid motion in erosion by solid particle impact. Int. J. Heat and Fluid Flow 11, No. 3, 170-195.
21. Lyczkowski, R.W., Bouillard, J.X., Ding, J., Chang, S.L., Lottes, S., and Burge, S.W., 1994, State-of-the-art review of computational fluid dynamics modeling for fluid-solids systems. Argonne National Laboratory Report No. ANL/ES/CP—82070 and as a conference paper for the International Symposium on Parallel Computing in Multiphase Flow Systems Simulations, 1994 Winter Annual Meeting of the American Society of Mechanical Engineers, Nov. 6-11, 1994, Chicago, Illinois: Paper CONF-941142--32
22. Sargent, G.A., Spencer, D.K., and Sagues, A.A., 1982. Slurry erosion of materials. Proc. Corrosion-Erosion Wear of Materials, NACE, Berkelely, California.
23. Shook, C.A. and Roco, M.C., 1991. Slurry Flow. Publ. Butterworth-Heinemann (ISBN 0-7506-9110-7).
24. Tuerberg, J.C., 2000. A stainless steel primer: Part 3: Selection of the proper alloy. Flow Control VI, No. 10, 28-36.
25. Wu, X., 2001. Monte Carlo modeling of turbulent dispersion of small particles in channels. Doctoral Dissertation from the Department of Mechanical Engineering of the Georgia Institute of Technology. (To be issued in June 2001).



Liquid-Solids

Flow Loops (liquid-solids)

26. Colwell, J.M. and Shook, C.A., 1988. The entry length for slurries in horizontal pipeline flow. Can. J. Chem. Eng. 66, 714-720.
27. Gupta, R., Singh, S.N., and Sehadri, V., 1995. Prediction of uneven wear in a slurry pipeline on the basis of measurements in a pot tester. Wear 184, 169-178.
28. Mishra, R., Singh, S.N., and Seshadri, V., 1998. Study of wear characteristics and solid distribution in constant area and erosion-resistant long-radius pipe bends for the flow of multisized particulate slurries. Wear 217, 297-306.
29. Nasr-El-Din, H., Shook, C.A., and Colwell, J., 1987. The lateral variation of solids concentration in horizontal slurry pipeline flow. Int. J. Multiphase Flow 13, No. 5, 661-670.
30. Roco, M.C. and Cader, T., 1990, Energy approach for wear distribution in slurry pipelines", Jap. Journal of Multiphase Flow 4, 2-20.
31. Roco, M.C. and Shook, C.A., 1983. Modeling of slurry flow: The effect of particle size. Can. J. Chem. Eng. 61, 494-503.
32. Sharp, B.B. and O’Neill, J.C., 1971. Lateral diffusion of large particles in turbulent pipe flow. J. Fluid Mech. 45, part 3, 575-584.
33. Shook, C.A., Daniel, S.M., Scott, J.A., and Holgate, J.P., 1968. Flow of suspensions in Pipelines: Part 2: Two mechanisms of particle suspension. Can. J. Chem. Eng. 46, 238-244.
34. Shook, C.A., McKibben, M., and Small, M., 1990. Experimental Investigation of some hydrodynamic factors affecting slurry pipeline wall erosion. Can. J. Chem. Eng. 68, 17-23.
35. Ting, K. and Ma, Y.P., 1999. The evaluation of erosion/corrosion problems of carbon steel piping in Taiwan PWR nuclear power plant. Nuclear Engineering and Design 191, 231-243.



Jets (liquid-solids)

36. Abbade, N.P. and Crnkovic, S.J., 2000. Sand-water slurry erosion of API 5L X65 pipe steel as quenched from intercritical temperature. Tribology International 33, 811-816.
37. Iwai, Y. and Nambu, K., 1997. Slurry wear properties of pump lining materials. Wear 210, 211-219.
38. Stack, M.M., Corlett, N., and Zhou, S., 1997. A methodology for the construction of the erosion-corrosion map in aqueous environments. Wear 203-204, 474-488.
39. Burstein, G.T. and Sasaki, K., 2000. Effect of impact angle on the slurry erosion-corrosion of 304L stainless steel. Wear 240, 80-94.
40. Sasaki, K. and Burstein, G.T., 1996, The generation of surface roughness during slurry erosion-corrosion and its effect on the pitting potential. Corrosion Science 38, No. 12, pp. 2111-2120.
41. Stack, M.M. and Pungwiwat, N., 1998. A note on the construction of materials performance maps for resistance to erosion in aqueous slurries. Wear 215, 67-76.
42. Umemura, F., Matukura, S., and Kawamoto, T., 1987, Electrochemical study of erosion-corrosion in carbon stainless steels. Corrosion Engineering 36, 569-578.



Pumps (liquid-solids)

43. van Bennekom, A., Berndt, F. and Rassool, M.N., 2001. Pump impeller failures – a compendium of case studies. Engineering Failure Analysis 8, 145-156.
44. Pagalthivarthi, K.V. and Helmly, F.W., 1992. Applications of material wear testing to solid transport via centrifugal slurry pumps. ASTM-STP, Philadelphia, 114-126.
45. Roco, M.C and Minani, L.K., 1989. Effect of particle size distribution and gravitation on wear in centrifugal pump castings", American Society of Mechanical Engineers, Paper No. 89-FE8.
46. Walker, C.I. and Bodkin, G.C., 2000. Empirical wear relationships for centrifugal slurry pumps Part 1: side-liners. Wear 242, 140-146.



Testers (liquid-solids)

47. Aiming, F., Jimming L., and Ziyun, T., 1996. An investigation of the corrosive wear of stainless steels in aqueous slurries. Wear 193, 73-77.
48. Clark, H.McI., Tuzson, J., and Wong, K.K., 2000. Measurements of specific energies for erosive wear using a Coriolis erosion tester. Wear 241, 1-9.
49. Dasgupta, R., Prasad, B.K., Jha, A.K., Modi, O.P., Das, S., and Yegneswaran, A.H., 1997. Wear characteristics of a hardfaced steel in slurry. Wear 209, 255-262.
50. Gandhi, B.K., Singh, S.N., and Seshadri, V., 1999. Study of the parametric dependence of erosion wear for the parallel flow of solid-liquid mixtures. Tribology International 32, 275-282.
51. Miller, J.D. and J.E. Miller, 1993. The Miller Number – a review. Proceedings of Hydrotransport 12 BHR Fluid Engineering held at Brugge, Belgium, 175-189.
52. Pagalthisvarthi, K.V., Ramanathan, V., and Visintainer, R.J., 1998. Finite element prediction of free surface flow in Coriolis Wear Tester. Am. Soc. Mech. Eng. Proceedings of the Fluids Engineering Division Summer Meeting Conference, June 21-25, Washington, DC
53. Xie, Y., Clark, H.McI., and Hawthorne, 1999. H.M. Modelling slurry particle dynamics in the Coriolis erosion tester. Wear 225-229, 405-416.
54. Zheng, Y., Yao, Z., Wei X., and Ke, W., 1995. The synergistic effect between erosion and corrosion in acid slurry medium. Wear 186-187, 555-561.



Gas-Solids

Flow Loops (gas-solids)

55. Mills, D. and Mason, J.S., 1981. Conveying velocity effects in bend erosion. J. Pipelines 1, 69-81.
56. Jets (gas-solids)
57. Andrews, D.R. and Horsfield, N., 1983, Particle collisions in the vicinity of an eroding surface. J. Phys. D: Appl Phys. 16, 525-538.
58. Dosanjh, S. and Humphrey, J.A.C., 1985. The influence of turbulence on erosion by a particle-laden fluid jet. Wear 102, 309-330.
59. Foley, T. and Levy, A., 1983. The erosion of heat-treated steels. Wear 91, 45-64
60. Mishra, A. and Finnie, I., 1981. On the size effect in abrasive and erosive wear. Wear 65, 359-373.
61. Singh, T., Tiwari, N., and Sundararajan, G., 1991. Room temperature erosion behaviour of 304, 316, and 410 stainless steels. Wear 145, 77-100.
62. Sundararajan, G. and Manish, R., 1997. "Solid particle erosion behaviour of metallic materials at room and elevated temperatures. Tribology International 30, No. 5, 339-359.



Pumps (gas-solids)

63. Zhong, Y. and Minemura, K., 1996. Measurement of erosion due to particle impingement and numerical prediction of wear in pump casing. Wear 199, 36-44.