Sei sulla pagina 1di 12

CORROSION ENGINEERING

Erosion and Erosion-Corrosion of Metals


A.V. Levy*

ABSTRACT
The loss of material from the surface of a metal being impacted by small, solid particles carried in a flowing gas occurs by a sequence of mechanical actions.The present work, which was presented as the plenary lecture at CORROSION/95, describes the mechanisms that occur on ductile and brittle materials and relates these mechanisms to surface loss rates. The ability of these mechanisms to be represented by mathematical models is assessed. The effects of material properties on erosion behavior is discussed. Combined erosion-corrosion surface degradation mechanisms are explained. KEY WORDS: brittle scales, ductile metals, erosion, erosioncorrosion, oxidation, solid particles

simultaneous oxidation. This behavior is affected by the characteristics of the target materials and the impacting particles, as well as by the conditions under which the surface degradation is occurring (i.e., flow velocity and direction, temperature, gas composition, the quantity of particles in the flowing gas stream, and the streams chemical composition and reactivity). The ability to analytically describe and predict surface degradation rates also is discussed.

ANALYTICAL MODELING
Efforts expended over the past 40 y by a rather large number of investigators seeking to model the erosion process have produced nothing that could be used to predict erosive behavior in any circumstance other than the one used in the research to develop a specific model. Even in those instances, the liberal use of constants in equations was necessary to even come close to having predicted erosion losses match experimental losses. The basic problem in modeling erosion revolves around the sequence of events, rather than a single event, that occurs when a surface is struck by successive small, solid particles. The microscopic mechanisms that occur, in sequence, have been well documented in microscopic studies and are understood easily. Representation of the series of events that occurs in equations which incorporate the variables that are active in the ongoing process has been elusive to date. K. Ludema, a prominent tribologist from the University of Michigan, recently bemoaned the fact

INTRODUCTION
Surface degradation of metals by the combined actions of small, solid-particle erosion and oxidizingsulfidizing corrosion in the elevated temperature fluid-flow environments of energy generation systems is an important consideration in their design and operation. In this paper, the results of many years of research to define and understand the mechanisms of these destructive phenomena are presented. This work concentrates on the physical behavior that occurs when small, solid particles strike surfaces that are chemically inactive or are undergoing
Submitted for publication May 1995; in revised form, June 1995. Presented originally as the plenary lecture at CORROSION/95, March 1995, Orlando, FL. * Lawrence Berkeley Laboratory, Berkeley, CA, 94720.

872

0010-9312/95/000197/$5.00+$0.50/0 1995, NACE International

CORROSIONNOVEMBER 1995

CORROSION ENGINEERING

that, throughout the field of tribology and not just in erosive wear, there were no successful models that could be incorporated into computer design programs to predict surfaces losses under any form of wear environment.1 It was his assessment that wear was not being treated properly in design in good part because, in todays computer design world, wear could not be computerized. He encouraged a much renewed, creative research effort to overcome the current tribology modeling deficiency. Ludema used an analysis of erosive wear to point up weaknesses and even ludicrousy of the modeling of the erosion process now in the literature. He reviewed 98 mathematical erosion models gleaned from the literature. He discarded all but 28 because he assessed that they were not amenable to conclusive scrutiny. In the 28 remaining equations, he identified 33 different variables plus constants used by the authors. Many patterns of variables were used in the various models. Seven models used seven variables, while nine used three or less, and one model used only one. No variable was used in all of the models, and the number of constants in the models varied from one to 10. There was no consistent pattern of use of the variables. Twelve of them were used only once. Some authors used the same variable in the denominator that others used in the numerator. None, according to Ludema, could predict material loss rates accurately. Reasons for this inability to represent the erosion process analytically in an effective, consistent manner can be explained based on the physical evidence that describes the mechanism that occurs. A brief analysis of the most utilized basis for erosion modeling will explain some of what has been wrong in its modeling. Modeling of a physical behavior that occurs on a microscopic scale generally involves a simplification of the process so that the mathematics can, in effect, cover the mechanism. Almost all of the early and many of the later erosion models that were based upon them used the concept of microcutting.2 In this concept, a theoretical erosion particle strikes a target surface at a shallow angle, and its tip acts like the tip of a cutting tool, cutting away a swath of metal in front of it as it translates across the microsurface. Using an equation of motion to describe the path of the particle and the cross section volume cut out by its leading edge geometry, the amount of material to be removed can be calculated. Then, if only a relatively small number of the total particles in the flow (represented by a constant) are assigned to be actual cutting particles, a volume or weight of material can be calculated to be removed over the entire impacted surface in a period of time. Figure 1 shows curves for the calculated and experimental material losses from a series of experiments carried out to verify the model.2 Significant

FIGURE 1. Erosion loss of aluminum, measured and calculated.

loss occurred at the higher impingement angles, where the microcutting model predicted no material loss. It is not shown in the figure, but the peak of the predicted curve had to be moved to coincide with that of the experimental curve. The basic problem with this much-used microcutting model is that the erosion of ductile metals, which the model was intended to depict, just does not occur in that physical manner, except for a few particles in a flow impacting at shallow angles (< 15). In the next section, the observed physical description of the erosion process is presented. It involves several different but common metal deformation processes occurring sequentially.

EROSION MECHANISM OF DUCTILE METALS3


With scanning electron microscopy (SEM) and its great depth of field and high magnification, which were not available to developers of the microcutting model, the actual mechanism by which material is removed from an eroding surface, as evidenced by surface topography, can be observed. With its use in a careful, interrupted test manner, the sequence of material removal can be determined. It is not necessary for each step in the process to occur at a distinctly different time. In some instances, the whole process occurs in one continuous action. The process is a straightforward one that involves common metal deformations including extrusion, pancake forging, bending, and fracture.

CORROSIONVol. 51, No. 11

873

CORROSION ENGINEERING

FIGURE 2. Platelets at edge of primary erosion zone on Al 1100-0 (700-m diam steel spherical shot, v = 62 mps [200 fps], = 60).

Figure 2 shows the outer edge of an Al 1100-0 (UNS A91100)(1) specimen that was impacted by 700- m steel spherical shot at a velocity (v) of 62 m/s (19 ft/s) and an impact angle ( ) of 60 at room temperature. A number of flattened platelets can be seen over the hilly, center portion of the eroded area, with some bent and/or partially cracked. The hillocks, or moguls, are known to occur at steep impact angles. This low-magnification micrograph taken early in our erosion research program was convincing evidence that microcutting was not the principal mechanism of erosion. Subsequent research consistently reinforced this idea. The results of a very precise, interrupted erosion test on Al 7075-T6 alloy (UNS A97075) are shown in Figure 3. The test consisted of impacting the specimen with small increments of erodent particles in a zone that was marked with microhardness indentations around its periphery to allow a return to it. After each increment, the microtarget area was observed using SEM, and locations in it were photographed. Figure 3(a) shows an eroded area with an extruded crater marked with curved striations made by grooves on the eroding particle. Figure 3(b) shows a single platelet that was extruded out of the crater on the right and flipped over the top of the crater shown in Figure 3(a), covering the striations. In Figure 3(c), a particle has struck the platelet and pancake forged it out, over part of the crater from which it was extruded. Other sides of the platelet also are extended out. The platelet is bound to the base metal only in an area alongside the crater from which it was extruded. In Figure 3(d), impacting
(1)

UNS numbers are listed in Metals and Alloys in the Unified Numbering System, published by the Society of Automotive Engineers (SAE) and cosponsored by ASTM.

particles have struck the platelet in a manner that has fractured its attachment stem to the base metal, and it has been removed from the surface completely. The striations in the crater shown in Figure 3(a) can be seen again. This sequence is the mechanism of erosion of ductile metals and has been observed countless numbers of times. A classic-shaped platelet is shown in Figure 4. Its surface has been cupped slightly by an impacting particle. Evidence of pancake forging cracks can be seen around its periphery. Subsequent impacts could break off pieces along these cracks or remove the whole platelet by fracturing the attachment stem located somewhere beneath its surface. Figure 5 shows evidence of a cracked attachment stem on another eroded specimen of type 310 (UNS S31000) stainless steel (SS). The curves in Figure 6 provide insight into the sequence of the erosion process. The curves show the target materials weight loss rate for increments of erodent of 60 g (2.1 oz). It can be seen that the first 60 g of silicon carbide (SiC) particles caused considerably less material loss than subsequent 60-g increments. If microcutting were the mechanism of erosion, the most material would have been removed by the first increment because the metals surface would have been in its least worked, lowest strength condition. The second 60-g (2.1 oz) increment brought the loss rate for all three alloy heat treat conditions up to a steady-state level that remained constant for each subsequent increment of 60 g of SiC erodent. Combining this information with microstructural observations led to the description of the eroded surface area and the mechanism of erosion presented below. It also is interesting to note in Figure 6 that the lowest strength, most ductile condition of the C1075 steel (UNS G10750), sheroidized, had the lowest erosion rate. The immediate surface of the eroding region, 5 m to 15 m thick, consists of craters and platelets at various stages of generation as the result of large plastic strain deformation by a sequential extrusion and pancake forging mechanism. Beneath the platelet zone is a somewhat thicker work-hardened zone that develops because the energy of the impacting particles is more than that required to form and deform the surface area into platelets. Beneath the work-hardened zone, which was measured in cross section using low-load microhardness tests, is base metal at its initial condition. The following sequence occurs: platelets are formed by impacting particles extruding metal out of craters, which then are pancake forged to an edgecracked, distressed condition. This leads to their fracture and removal from the surface. At the same time, early in the erosion process, the excess energy of the impacting particles cold works the region

874

CORROSIONNOVEMBER 1995

CORROSION ENGINEERING

(a)

(b)

(c)
FIGURE 3. Sequence of platelet formation and removal on Al 7075-T6.

(d)

below the surface platelets. The first material loss occurs prior to the full hardness of the cold-worked zone being achieved. When the complete target metal surface is filled with mature platelets and the coldworked zone is at its maximum hardness, steadystate material removal conditions occur. The coldworked zone acts as an anvil, and the erodent particles act as hammers. Their maximum efficiency in forming and removing platelets of metal coincides with steady-state erosion at its highest rate, as is shown in Figure 6. This sequence progresses through the metal as surface material is removed. The sequence of physical events that results in surface material removal by the erosion process is well documented. The mechanism is considerably more complex than simple microcutting and has not been represented successfully in an erosion model to date. When corrosion is added, the mechanism becomes still more complex. It still can be comprehended readily as a mechanism of surface degradation from a physical standpoint, but it is even more

difficult to represent mathematically. That is the dilemma of erosive wear modeling.

EROSION MECHANISM OF BRITTLE SCALES ON METALS3


To determine the basic mechanism of erosion of brittle oxide scales formed on base metals under somewhat controlled conditions where erosion and corrosion were not occurring simultaneously, commercially pure nickel was oxidized in a static air furnace where a duplex nickel oxide (NiO) scale formed of varying thicknesses. In all instances, the outer scale was a 20- m thick, dense, columnar grain scale, while the inner scale was a much finer, equiaxed grained scale that was porous and of different thicknesses (from 0 m to 80 m), depending on the exposure time. The mechanism of erosion was identical to that which occurs when an ice pick is used to break up a block of ice. As shown in Figure 7, the first particles

CORROSIONVol. 51, No. 11

875

CORROSION ENGINEERING

FIGURE 4. Erosion platelet on Al 7075-T6.

to strike the scales surface made plastically deformed individual indentations. Figure 8 is a view of one such indentation from the lower center area of Figure 7. Radial cracks emanated out from the walls of the indentation. This is exactly what happens at the point of impact of an ice pick on ice. Subsequent impacts caused patterns of cracks to form that led to pieces of the scale being able to be knocked off the surface. Figure 9 shows the condition that exists when the outer, more dense scale has been eroded off and the porous, inner scale is eroding. Figure 9(a) shows a location along the edge of the hole that is being eroded in the scale. The outer surface and outer and

inner layers of the scale, as well as the base metal with a crater and the beginning of a platelet, all can be seen. Figure 9(b) shows the inner scale with typical Hertzian cone cracks dispersed in it. Figure 10 is a sketch of the stress fields and crack distributions that occurred beneath the sites where the initial particles plastically deformed the surface. Planar cracks formed along the interface between the outer and inner scales when the area was unloaded after the erodent particle rebounded from the surface. This resulted in the cones seen in Figure 9 having flat tops. Thus, both radial and planar cracks formed in the scale early in the erosion process, breaking up the surface region into a horizontal and vertical crack network. Figure 11 is the result of a very precise erosion test sequence using small quantities of particles and interrupted tests. The microstructural state of the outer and inner scales can be related to the rate of material removal. E1 is the plastic indentation period in which essentially no material is removed. E2 is the removal of the dense, columnar, more erosionresistant outer scale. The higher loss rates of E3 are primarily porous inner scale being removed readily as the eroded hole increases in overall diameter, and E4 is the lower loss rate caused by the exposure of only the scale on the sides of the hole, where the impact angle was very shallow and the projected scale area quite small. The base metal was in the early process of forming platelets, and no material had been removed yet. All brittle scales formed in situ on ductile metals are removed by this cracking and chipping mechanism. When the scale formation is occurring at the

(a)
FIGURE 5. Cross sections of type 310 SS eroded surface (710C,

(b)
= 30).

876

CORROSIONNOVEMBER 1995

CORROSION ENGINEERING

FIGURE 7. Nickel oxide scale surface after a few angular SiC particle impacts.

FIGURE 6. Incremental erosion of C1075 steel to steady state.

same time as the particle impacts, certain circumstances cause pieces of different sizes to be removed, with different resulting loss rates, but the basic loss mechanism remains the same. In some instances, the scale that forms during combined erosioncorrosion can protect itself and the substrate metal from material loss.

EFFECT OF STRUCTURES AND PROPERTIES OF METALS ON EROSION RATES


The properties and structures of most commonly used metal alloys have relatively small effects on their erosion resistance.4 Table 1 shows that several heat treat conditions of C4340 steel (UNS G43400) with their relatively large differences in strength, ductility, hardness, and toughness had a relatively minor effect on erosion rate. A hardness difference of 300% resulted in an erosion rate difference of ~ 10%, with the softer material having the lowest metal loss. This behavior is somewhat different from other kinds of wear, such as sliding wear, where higher hardness results in lower wear rates. In a series of erosion tests on a group of nickel- and cobalt-based alloys variously heat treated and cold worked to six different conditions with Vickers hardnesses numbers (VHN) ranging from 243 VHN to 465 VHN, the erosion rates ranged from 1.07 x 105 cm3 g1 to 1.17 x 105 cm3 g1, less than 10% difference for a hardness difference of almost 200%.5 It did not make any difference in erosion behavior whether the alloys
FIGURE 8. Single impact crater on NiO columnar grain surface.

were in the annealed, aged, or cold-reduced condition. There were reasonably large differences in erosion behavior between some alloys that were related to their plastic flow characteristics. Figure 12 compares the incremental erosion weight loss rates for C1020 steel (UNS G10200) and type 304 SS (UNS S30400).6 The two steels had similar yield strengths and hardnesses, but the stainless steel was more ductile than the mild steel and, being face-centered cubic (FCC) compared to body-centered cubic (BCC), had more slip systems with which to deform. It eroded at half the rate of the C1020 steel. This comparison indicated that when an alloy is more capable of dissipating the force from an impacting erodent particle by local plastic deformation without fracture, its erosion loss rate is reduced.

CORROSIONVol. 51, No. 11

877

CORROSION ENGINEERING

(a)
FIGURE 9. Micrographs of eroding NiO showing sections down to base metal.

(b)

FIGURE 10. Sketch of cross section of eroding scale.

The effect of the force of impacting particles on the platelet structure and resulting metal loss is shown in Figure 13.7 At the lower particle velocity of 30 m/s (9.1 ft/s), the craters and platelets shown in the micrograph on the left were relatively small, and the resultant loss rate of the Al 1100 alloy was 1.6 x 105 cm3 g1. When the velocity of the particles was 90 m/s (27.4 ft/s), the microstructure on the right side resulted. The craters and platelets were several times the size of the lower velocity test specimens, and the metal loss rate was 13.3 x 105 cm3 g1, over eight times greater. Another difference in erosion behavior that was determined to be due to microstructural variations in families of carbide-containing materials is shown in Figure 14.8 Three groups of carbide-containing

FIGURE 11. Incremental erosion rate of NiO during first 0.5 g of erodent impact.

alloys/composites are plotted. The group on the lower left was composed of ingot iron and two carbon steel alloys. The materials in the upper, steep curve were alloys 6 and 6B, TS 1, 2 and 3, and a chromium-molybdenum white iron. The composites in the curve on the right side were all Kenemetal carbidemetal composites used primarily for cutting and forming tools. The iron groups behavior primarily was a function of ductility with the increasing carbide content up to the C1080 steel reducing the ductility and causing an increase in the erosion rate. As the

878

CORROSIONNOVEMBER 1995

CORROSION ENGINEERING

TABLE 1 Effect of Ductility, Strength, Toughness, and Hardness on Erosion Behavior of 4340 Steel(A)
Ultimate Tensile Strength (ksi) 307 273 182 ~ 100 Klc in ksi (in) 34 58 62 Charpy Impact Strength (ft/lb) 10 16 12 Steady-State Erosion (mg) 1.03 0.97 0.97 0.90

Heat Treat Condition As-quenched 200C tempered 500C tempred Spherodize annealed
(A)

Hardness (Rc) 60 53 39 ~ 19

Elong. (%) 8 11 14 ~ 25

Statistical average of weight loss per 30-g load of 140-m Al2O3 particles at steady-state erosion ( = 30, v = 30 mps, and T = 25C).

carbide particle content increased in the materials in the steep curve, the plastic deformation capability of the material was decreased, and the erosion rate increased. The carbide particles impeded the ability of the metal matrix to flow under the impact force of eroding particles, and voids and cracks opened that led to loss of pieces of the material. Even though these materials were very hard and could resist sliding wear, the forces of erodent particle impacts could defeat them readily. The ingot iron and the two carbon steels all had considerably greater erosion resistance than the alloys 6 and 6B and the TS 1, 2, and 3 materials. The carbide composites in the curve on the right side of Figure 14 generally had increasing erosion resistance with carbide content, particularly at carbide contents 80%, where they formed a complete carbide skeleton that could cause impacting erodent particles to shatter. Below the 80% carbide skeleton, structure, the carbide particles detered plastic deformation of the matrix metal, and the resultant erosion rates were higher than those of the ductile ingot iron and carbon steels. There is no simple deterrent to erosion in the properties and structures of materials. Certainly, the commonly used property, hardness, does not relate to erosion behavior, except perhaps in a negative manner. The selection of materials for a particular design environment is more a function of the environment than it is the material, in the absence of elevated temperature corrosion conditions. The properties of the impacting particles that have a significant effect on their erosiveness are their physical integrity upon impact (shatter resistance) and their shape. To erode a surface, the erodent must maintain its size and shape as it strikes the surface. When its strength/toughness is great enough to prevent shattering and its shape is angular to be able to concentrate its force on the surface, its erosiveness is at its greatest. Further increases in strength/toughness do not increase the erosiveness of the particle. Thus, angular-shaped sand (S1O2) with a hardness of 700 VHN is just as erosive as SiC

FIGURE 12. Effect of ductility on the erosion of C1020 steel and type 304 SS (RB = Rockwell B hardness scale, Sy = tensile yield strength, and n = strain hardening coefficient).

with a hardness of 3,200 VHN. Hardness can be used as an indicator of a particle materials physical integrity. If the particle shatters, the resulting small pieces do not have the kinetic energy necessary to cause maximum plastic deformation of the target metal. If the particles are round or become rounded when shattering occurs, they do not have the ability to concentrate their force to cause maximum surface deformation. For example, round steel shot causes only 25% of the erosion that angular steel grit does at the same particle size and erosion conditions.

EROSION-CORROSION OF STEELS AT ELEVATED TEMPERATURES


When small, solid particles are carried to the surfaces of metals in elevated temperature air streams such as occur in fossil fuel combustion

CORROSIONVol. 51, No. 11

879

CORROSION ENGINEERING

(a)
FIGURE 13. Cross section of Al 1100 eroded by 250-m SiC particles at

(b)
= 30 and: (a) v = 30 m/s and (b) 90 m/s.

FIGURE 14. Erosiveness of materials as a function of their carbide content.

chambers, simultaneous erosion-corrosion material degradation occurs. The rates of material loss can be increased or decreased compared to bare metal loss rates, depending on the nature of the scale formation process and its characteristics. The basic mechanisms of erosion are the same as described earlier for ductile and brittle materials. However, the actual material that is being removed is one that is being created on the surface at the same time and by the same forces that are removing it. This materials

chemical composition and physical properties are a function of the base metal and gas compositions, as well as the composition and shape of the erodent particles and the exposure conditions. The ability of the impacting erodent particles to increase the effective corrosion temperature of a steel as indicated by the morphology of the scale that forms is shown in Figure 15.9 The micrograph on the left shows the thin, adherent, essentially chromium oxide (Cr2O3) scale that formed on a 9% Cr-1% Mo steel under dynamic corrosion conditions. In this test, air heated to 750C was blasted at the target surface at a velocity of 60 m/s (18.3 ft/s) with no erodent particles in it. When 130- m aluminum oxide (Al2O3) particles were added to the 750C air, the scale morphology shown in Figure 15(b) occurred. The scale consisted of domains of iron oxide (Fe2O3) separated by a crack network. The dynamic corrosion surface of the steel at a test temperature of 900C is shown in Figure 15(c). It is comparable in its morphology to that of the erosion-corrosion surface from the 750C test. Thus, the presence of erodent particles in the gas stream effectively increased the Fe2O3 scale forming conditions by 150C. The morphology of the scale that forms on the target surface during combined erosion-corrosion has a major effect on the loss rate of the material.9 Increasing the velocity of the rounded Al2O3 erodent particles impacting 9% Cr-1% Mo steel specimens at = 90 and a test temperature of 850C caused distinct changes to occur in the morphology of the scale formed on the target surface. Figure 16 shows these changes over a particle velocity range from 10 m/s (3.05 ft/s) to 70 m/s (21.3 ft/s). The most notable difference occurred between a particle velocity of 30 m/s (9.1 ft/s) and 45 m/s (13.7 ft/s). At 10 m/s and 30 m/s velocities, the scale appeared to be segmented and cracked and had a larger oxide crystallite size. At 30 m/s velocity, the beginning of consolidation or densification of the

880

CORROSIONNOVEMBER 1995

CORROSION ENGINEERING

(a)

(b)

(c)

FIGURE 15. Scale morphology of 9% Cr-1% Mo steel at: (a) 750, vair = 60 m/s (b) 750, vpart = 70 m/s, and (c) 900C, vair = 60 m/s.

(a)

(b)

(c)

(d)

FIGURE 16. Effect of particle velocity on scale morphology of 9% Cr-1% Mo steel at: (a) 10 m/s, (b) 30 m/s, (c) 45 m/s, and (d) 70 m/s (nozzle tester temperature = 850C, = 90, t = 5 h, 130-m Al2O3).

scale could be seen. It became more pronounced at 45 m/s and 70 m/s velocities. At 45 m/s and 70 m/s, the scale appeared to be essentially continuous. No divisions or cracks were discernible, and individual crystallites could no longer be seen at the magnification used. At 70 m/s, embedded Al2O3 could be seen in the surface, primarily along the crests of the indentations. This change in scale morphology with velocity (and test temperature) occurred at = 90 but not at = 30. The metal thickness loss vs particle velocity curves are shown in Figure 17 for 90 and 30 impact angles. The 90 impact curve had a distinct transition at ~ 30 m/s velocity, going from low metal loss rates to high metal loss rates. No such transition occurred in the 30 impact tests, which had a continuous increase in loss as the increasing velocity resulted in higher impact energy particles striking the surface. The 0 m/s velocity data point in the 90 impact test curve was that for dynamic corrosion (no particles in the flowing gas). The type of S-shaped curve for the 90 impact tests is indicative of a major change in the mechanism of any behavior that is being plotted. This major change in the mechanism

of erosion-corrosion that occurred was the change in scale loss from a low loss rate mechanism of cracking and chipping of small pieces of scale to a high loss rate mechanism of periodic spalling of much larger pieces of scale. The loss of the scale at different rates translated to the loss of the base metal as it was oxidized to replace the eroding scale. The spalling was thought to be due to the continuous nature of the scale, which did not have stress-reducing cracks in it, as occurred in the scales formed at the lower particle velocities. Thus, stress levels that could result in spalling built up. Periodically these stresses, aided by the force from impacting erodent particles became sufficiently high to break off relatively large pieces of scale. The micrographs of the scales shown in Figure 16 show scales at 45 m/s and 70 m/s that had not reached the spalling stress level yet. The change in metal thickness loss rates at the higher particle velocities were attributed to the change in the mechanisms of scale formation and removal rather than the change in kinetic energy of the particles or some indefinable, synergistic combined erosion-corrosion mechanism. Kinetic energy increased uniformly by the square of the particle

CORROSIONVol. 51, No. 11

881

CORROSION ENGINEERING

FIGURE 17. Metal thickness loss vs vpart for 30 and 90 impact angle tests.

FIGURE 18. Effect of impact angle on metal loss at four particle velocities.

removal method than cracking and chipping of small individual pieces, as occurred at the lower impact velocities. Figure 18 shows the different types of behavior of the scales in a more dramatic manner. At the higher velocities where the consolidated scale spalled in the 90 impact tests, the material loss was greater at 90 than at 30 for the same particle velocity. However, at the lowest particle velocity of 25 m/s (7.6 ft/s) where consolidation did not occur and the loss mechanism was similar at both 30 and 90, the greater erosion resistance that brittle materials are known to have at steep impact angles compared to shallow angles resulted in a lower thickness loss at 90. The effect of the scales morphology on the erosion-corrosion loss rate is also a function of test temperature.10 Figure 19 shows the surfaces of specimens tested at different temperatures in air using 130- m Al2O3 erodent particles. The scale is segmented in the 750C test but consolidated in the 900C test. The metal loss curve as a function of test temperature was similar in shape to the transition portion of the 90 impact test curve in Figure 17. While the oxide scales mechanical properties changed very little over the test temperature range, its morphology went from a segmented to a consolidated configuration with a resulting thickness loss increase of greater than five times. Thus, in combined erosion-corrosion, the characteristics of the scale that is being formed and removed simultaneously is the primary determinate in establishing the metal loss rate. In addition to the morphology factor discussed above, other factors such as scale thickness (it is quite thin but protective on stainless steels)11 and particle composition (the presence of calcium causes a cementing action to occur in the scale that makes it protective)12 can play important roles in determining the material loss rate. The use of protective coatings13 to modify or prevent the formation of a scale on the surface has been demonstrated to be a very effective means to deter erosion-corrosion material losses in fossil fuel combustion boilers.

CONCLUSIONS
y Mathematical models of the erosion of ductile and brittle material surfaces that can predict material loss rates have not been developed successfully to date. y Ductile metals are eroded by a mechanism that sequentially extrudes, pancake forges, and finally fractures platelets of material off the target surface. y Brittle materials are eroded by a mechanism that initially creates a crack network in the target material surface and then chips pieces off. y Target material properties generally have relatively minor effects on erosion by small, solid particles.

velocity. This was the primary cause for the loss increases at the 30 impact angle. At the 90 impact angle and particle velocities > 30 m/s, the impacting particles consolidated the oxide by a hot-pressing type of action, which did not occur at the more shallow, 30 impact angle. This resulted in the formation of the continuous scales which could develop high enough thermalmechanical stresses over large enough areas to cause spalling and make it the primary scale removal mechanism. Spalling is a much more effective

882

CORROSIONNOVEMBER 1995

CORROSION ENGINEERING

(a)

(b)

(c)

FIGURE 19. Effect of test temperature on scale morphology in 70 m/s particle velocity test (a) at 900C, (b) at 850C, and (c) at 750C (nozzle tester time = 30 min, = 90, 130-m Al2O3).

y The shatter resistance and shape of erodent particles have the greatest effects on their erosiveness. y Combined erosion-corrosion of metals occurs by a simultaneous formation and removal of scale from the surface by a cracking and chipping mechanism or by a spalling mechanism.
REFERENCES
1. H.S. Meng, K.C. Ludema, Wear Life Equations for Mechanical Designers: State of the Art, 1993 Int. Wear of Materials Conf. San Francisco, California, 1993 (Lausanne, Switzerland: Elsevier, 1994).

2. I. Finnie, The Mechanism of Erosion of Ductile Metals, Proc. 3rd U.S. Natl. Cong. Appl. Mechanics (New York, NY: American Society of Mechanical Engineers, 1958), p. 527-532. 3. A.V. Levy, The Erosion of Metal Alloys and Their Scales, Proc. NACE Conf. Corrosion-Erosion-Wear of Materials in Emerging Fossil Energy Systems, Berkeley, California, 1982 (Houston, TX: NACE, 1982), p. 298-376. 4. A.V. Levy, Wear 68, 3 (1981): p. 269-287. 5. A. Ninham, Wear 121, 3 (1988): p. 307-324. 6. T. Foley, A. Levy, Wear 91, 1 (1983): p. 45-64. 7. A. Levy, M. Aghazadeh, G. Hickey, Wear 101, 1 (1986): p. 23-42. 8. A. Ninham, A. Levy, Wear 121, 3 (1988): p. 347-361. 9. A. Levy, Y. Man, Wear 111, 2 (1986): p. 135-160. 10. A. Levy, Y. Man, Wear 111, 2 (1986): p. 161-172. 11. A.V. Levy, Y.F. Man, Wear 131, 1 (1989): p. 39-52. 12. G.Q. Geng, B.Q. Wang, A.V. Levy, Wear 150, 1-2 (1991): p. 125-134. 13. B.Q. Wang, G.Q. Geng, A.V. Levy, Surf. Coat. Technol. 43-44, 1-3 (1990): p. 859-874.

CORROSIONVol. 51, No. 11

883

Potrebbero piacerti anche