Conference Paper CONFERENCE PAPER - ADVANCED MATERIALS FOR PUMP SEALS COMPANY WIDE CW-127520-CONF-001 Revision 0
Prepared by Rédigé par
Reviewed by Vérifié par
Approved by Approuvé par
2013/02/13 2013/02/13 UNRESTRICTED ILLIMITÉ
Atomic Energy of Énergie Atomique du Canada Limited Canada Limitée
Chalk River, Ontario Chalk River (Ontario) Canada K0J 1J0 Canada K0J 1J0 CW-127520-CONF-001
Proceedings of the 2012 20th International Conference on Nuclear Engineering collocated with the ASME 2012 Power Conference ICONE20-POWER2012 July 30-August 3, 2012, Anaheim, California, USA
ICONE20POWER2012-54685
ADVANCED MATERIALS FOR PUMP SEALS
Timothy Sykes Marc Pinault Atomic Energy of Canada Limited Atomic Energy of Canada Limited Chalk River, Ontario, Canada Chalk River, Ontario, Canada
Austin Jackson Nick Potvin Robby Baidwan Atomic Energy of Canada Atomic Energy of Canada Atomic Energy of Canada Limited Limited Limited Chalk River, Ontario, Canada Chalk River, Ontario, Canada Chalk River, Ontario, Canada
next outage or, in the worst case, causing a forced outage of Abstract the reactor [1]. The effectiveness and durability of high performance pump Work is underway within the Fluid Sealing Technology seals in the broad range of conditions in which they operate is Branch at AECL’s Chalk River Laboratories to both better often limited by the behaviour of the materials from which understand the mechanism causing failures such as seal components have been designed and manufactured. degradation of nickel-bound cemented tungsten-carbide seal Aging management of active components, such as pumps and faces and to assess the performance of alternative seal face valves, requires a better understanding of the behaviours of materials for these applications. these materials in dynamic conditions in order to extend component life and to reduce losses associated with A suite of erosion and wear tests have been performed on a sub-component failures, such as those associated with seals. diverse range of carbon-graphite composites and ceramic seal Pump seal designs employ several types of materials, two of face material combinations, which includes the following: which have behaviours that are not well understood at their Steady-state wear tests at aggressive limits: carbon-graphite composites and wear-resistant pressure-velocity conditions, ceramics. An ongoing research program is underway at Accelerated wear-rate tests at low hydraulic pressure, AECL’s Chalk River Laboratories to both better understand Erosion tests at typical hot standby conditions, and the behaviour and assess the performance of new seal face Highly accelerated wear tests at dry-running materials, which have been developed using advanced conditions. technologies. This paper presents the results of this work. Introduction Seal Face Materials Mechanical seals are used in the pumps of a broad range of nuclear power plants systems. The mechanical pump seal This phase of the advanced materials test program required industry has been challenged to extend the service life and that AECL consider commercially-available seal face reliability of these seals such that they are replaced less materials—both the commonly used materials and the latest frequently and only at planned outages. A material commonly materials that have been produced using advanced used for rotors in mechanical seals is nickel-bound cemented technologies. Criteria such as mechanical properties and tungsten carbide. Although this material has performed well corrosion resistance were then used to select a subset of under some of the conditions in which it operates, it has also materials for comparative testing. caused repeated failures in other conditions. Failure Stator Materials investigations have shown that selective dissolution of the nickel binder weakens the tungsten carbide matrix leading to Several factors were considered when selecting the stator erosive wear of the rotor and ensuing high leak rates, materials for the test program - mechanical properties such as ultimately requiring replacement of the mechanical seal at the elastic modulus, thermal conductivity, and flexural strength as
1 Copyright © 2012 by ASME well as other characteristics such as dimensional stability, wear resistance, porosity and chemical composition. Carbon Graphite Potential stator materials were selected based upon a comparison of their properties to a specific grade of Raw Material carbon-graphite, which is referred to here as the reference Petroleum coke, pitch coke, carbon stator material. The strengths and weaknesses of this material, black, or graphite are mixed in based upon its performance in nuclear power plant pump seals specific quantities. over the last two decades and results of laboratory testing, Powder Formation were analyzed and documented, and collectively they provide The materials are crushed, milled a baseline of what constitutes an excellent stator material. and sieved until the grain structure is as required. Relative to this reference material, an excellent candidate material should have equal or higher elastic modulus, flexural Binder Addition strength, compressive strength and thermal conductivity, and Petroleum based pitch is added at an equal or lower thermal expansion coefficient. It would also elevated temperatures. have an equal or greater amount of graphite, to provide good wear resistance due to the high lubricity provided by graphite. Milling Four candidate stator materials were selected for comparative The mixture is milled into the testing against the reference material by applying these criteria desired grain size. to a list of commercially-available stator materials. The actual names of these materials have not been used in this paper Homogenizing because their test results, which are reported later, are The powder with binder is processed to ensure the mixture is protected intellectual property. However, code names have homogenous throughout. been assigned to each of the materials to allow the results of the test program to be reported here. These code names are summarized below: Compaction The powder is then pressed by one Stator-Ref: This is the reference stator material of the following techniques. against which the four candidate materials are compared. Isostatic Molding Die Molding This method presses the material in This method presses the material Material-S1 to Material-S4: These are the four all directions and is used for from two directions and is used for candidate stator materials. complex shapes in lower quantities. simple parts in large quanitities. The exact processes used to produce the carbon graphite stator materials that were tested are proprietary. Baking However, Figure 1 shows a typical set of processes for Heated in furnace at 1000°C to 1200°C. Pyrolysis converts binder these materials [2]. into carbon, which binds material.
Impregnation Optional Carbonization Synthetic resin enters the porous Slowly Heated in furnace; 800°C to structure created by pyrolysis of the 1300°C. Pyrolysis converts binder. impregnant into carbon.
Final Machining The components are machined to their final dimensions using conventional techniques.
Special Treatments Some components are top coated with specific polymers to decrease water absorption.
Figure 1: Production Process for Carbon-Graphite
2 Copyright © 2012 by ASME Rotor Materials After the sintering process, the component is final machined to Candidate rotor materials were selected for their mechanical its required dimensions [3]. and physical properties—strength, modulus of elasticity, Silicon Carbides thermal properties (conductivity and expansion coefficient) The silicon carbide family can be further divided into two and corrosion resistance. Chemical composition and forming sub-families—reaction-bonded silicon carbides and techniques were also investigated to better understand the pressure-less sintered silicon carbides. influence of these factors on physical properties and corrosion resistance. Reaction Bonded Silicon Carbides The properties of the candidate materials were compared to a The production process for reaction bonded silicon carbide specific grade of tungsten carbide, which is referred to here as starts with the creation of a single-phase silicon carbide the reference material. Similar to the reference stator material, powder (α-SiC). There are several known methods to create the strengths and weaknesses of this material provide a this powder. The simplest method is known as the Acheson baseline of what constitutes an excellent rotor material. process, whereby high-purity quartz sand, plus graphite or coal However, as discussed earlier, the susceptibility of the are electrically-heated in a resistance furnace between 1600°C reference rotor material—a specific grade of tungsten and 2400°C for about 36 hours. After cooling, the high-purity silicon carbide is separated and further processed into the carbide—to nickel leaching excluded it as a candidate for a required size fraction by crushing, milling, and sieving. The new rotor material. α-SiC is then milled to the sub-micron size and graphite flakes Ten candidate rotor materials from four material and polymer binders are then added to the silicon carbide families - tungsten carbide, silicon carbide, silicon nitride and powder. By heating this mixture under a vacuum, pyrolysis aluminum oxide—were selected for comparative testing converts the polymer binder into additional carbon. The against the reference material by applying these criteria to a mixture is then sintered at a temperature above 1500°C under list of commercially-available rotor materials. Code names a vacuum or inert atmosphere. The two main techniques for have been assigned to each of the rotor materials (as was done achieving this are the “pack” or “wick” technique, where the for the stator materials) to allow the results of the test program mixture is packed in silicon powder, or is suspended over to be reported here. These code names are summarized below: silicon, respectively. The liquid silicon1 penetrates the porous Rotor-Ref: This is the reference rotor material, mixture through capillary action. It reacts with the graphite to against which the ten candidate materials are form a two-phase silicon carbide (α-SiC and β-SiC), which compared. bonds the structure together. Initially, approximately only 0.5% of the graphite is dissolved, however the exothermic Material-R1 to Material-R10: These are the ten dissolution causes the local temperature to increase leading to candidate rotor materials. further dissolution. The carbon diffuses away to cooler sites, A description of these four material families and their and becomes supersaturated in the silicon leading to production processes, where this was available, is provided in precipitation of new SiC. The transformation of some β-SiC the following sections. to α-SiC occurs behind the reaction front. The final microstructure contains both α-SiC and β-SiC [4]. Tungsten Carbides Pressure-less Sintered Silicon Carbides A general understanding of the production process for tungsten carbide, shown in Figure 2, can help explain the The production process for sintered silicon carbide has several properties or characteristics that were used in the selection of a variations. Pressure-less sintering (also called direct sintering) rotor material. Generally, tungsten carbide powder is a of silicon carbide is a relatively new method that produces a derivative of amononium-paratungstate (APT) [3], a white very pure silicon carbide, with no free carbon or graphite crystalline salt which is formed from the minerals sheelite or particles. The process begins with single phase silicon carbide wolframite. The APT is then reduced into pure tungsten powder (α-SiC); the same raw material used in the production powder in a heated hydrogen environment. The tungsten of reaction bonded silicon carbide. The α-SiC powder is powder is mixed with soot (carbon) and is heated at high mixed with sintering aids such as boron and carbon resin. The temperatures in hydrogen to create tungsten carbide powder addition of the boron is believed to enhance material transport [3]. during the sintering process allowing the carbon resin to react with the free silicon and silicon oxide to form silicon carbide. The tungsten carbide powder is mixed with binders, such as The mixture is isostatically pressed and machined into its nickel, and is wet-milled. The milling allows for control of approximate geometry using conventional techniques. The the grain size and homogeneity of the solution. The liquid in mixture is then sintered by heating in an inert gas or vacuum the slurry is removed by techniques such as spray drying. atmosphere between 1950°C and 2100°C [4]. The sintering Once in the powder form, the material is ready for process results in a fine-grain, single-phase grain structure compaction. The powder is then either uniaxially die pressed (α-SiC) with no free silicon and very low porosity [5]. or cold isostatically pressed into a desired shape. Once the compact is ready, it could be pre-sintered before it is soft machined, to facilitate production handling. The component is then vacuum sintered, generally between 1370°C and 1540°C, which shrinks the component approximately 50% by volume. 1 Melting point of silicon is approximately 1400°C. 3 Copyright © 2012 by ASME Silicon Nitrides Tungsten Carbide Reaction Bonded Silicon Carbide Pressureless Sintered Silicon (WC) (RbSiC) Carbide (SSiC) One hot isostatically-pressed silicon nitride was selected for testing from the silicon nitride family. A typical silicon nitride has a crystal structure with predominantly covalent bonds, Raw Material Raw Material Raw Material Sheelite or Wolframite are which are very strong and arranged in a similar way to the High purity quartz sand, coke and High purity quartz sand, coke and sequenced into ammonium atomic structure of diamonds. However, it is impractical to graphite are mixed. graphite are mixed. paratungstate (APT). manufacture components with such a material. Consequently, the crystal structure is transformed into a poly-crystalline Reduction Reduction Reduction material with micron-sized silicon nitride grains, held together Mixture is reduced into silicon Mixture is reduced into silicon Hot reduction of APT in hydrogen carbide powder in a resistance carbide powder in a resistance with a secondary glass-like phase. To improve the density, produces pure tungsten powder. furnace (870°C to 1370°C.) furnace (870°C to 1370°C.) toughness and strength, the material is then liquid-phase sintered so that the alpha phase dissolves into the Carburization Powder Processing Powder Processing inter-granular glass, and the beta phase silicon carbide Tungsten and soot are mixed and Further processing includes Further processing includes precipitates out of the glass. Magnesia is used as a sintering heated at high temp in hydrogen to crushing, milling, sieving into crushing, milling, sieving into produce tungsten carbide powder. desired size fraction. desired size fraction. aid [6]. This process results in an extremely dense material, with virtually no pores. Binder Addition Binder Addition Sintering Additives Graphite flakes and a polymer Aluminum Oxides Binder and additives are mixed and Small amounts of boron and carbon binder are added to the silicon wet milled. (resin) are added to facilitate carbide powder. Two aluminum oxides with silicon carbide whisker pressureless sintering reinforcement were also selected for testing. The silicon carbide whiskers are discontinuous needle-shaped crystals, Milling Compaction Compaction Slurry is milled to achieve The silicon carbide and additives typically 0.1 to 5 microns in diameter and greater than The mixture is isostatically pressed homogeneity and desired grain are compacted into a “green” into pre-sintered or “green” material 5 microns in length [7]. One known technique for the size. compact. production of silicon carbide whiskers is a carbothermic reduction reaction of silica and organic carbon (typically rice Soft Machining Spray Drying Pyrolysis The pre-sintered material is formed hulls). This whisker composite is formed using pressure-less Slurry is dried into the ready to The binder is pyrolyzed to produce into the required geometry by press (RTP) powder form. additional carbon and open porosity sintering, as outlined in Figure 2. During pressure-less by heating it under vacuum. conventional techniques. sintering, the whiskers are known to inhibit densification since they interfere with particle rearrangement and shrinkage. Blending Test Sintering RTP powder is tested for Green material is exposed to inert Techniques such as high shear mixing and ultrasonic composition and pressing Infiltration (Wick Method) Infiltration (Pack Method) gas or vacuum atmosphere heated dispersion have been developed to ensure the whiskers are properties. The material is suspended above The material is surrounded by a to between 1950-2100°C dispersed homogenously after sintering [8]. molten silicon at a temperature powder mix with silicon. It is then Compaction above 1500°C under vacuum. heated > 1500°C under vacuum Final Machining Test Seals and Test Rig The RTP powder is either uniaxially The silicon carbide is then die pressed, or cold isostatically machined to the final form through Testing was performed at a reduced scale using the CAN6 pressed. conventional methods. Reaction with Graphite Exothermic Reaction cartridge seal design—a design which has been adapted for The molten silicon reacts with The reaction with graphite is use in BWR reactor water cleanup pumps. It relies on Presintering (optional) graphite to form α and β exothermic, leading to increased hydrostatic film lubrication and uses the same proven “green” slugs are produced by SiC, holding the body together. solubility of graphite in the silicon. hydraulically pressing powder. principles for preventing unwanted deflections due to pressure and temperature changes and for creating a self-relieving response to heat generation at the seal faces. Soft Machining Carbon Precipitation The compacts are soft machined The carbon diffuses to cooler locations, Powder Production Multiple seal rings were made using all candidate materials. into a proper shape. where it becomes supersaturated in the silicon and precipitates as new SiC. These rings have a diameter of roughly 75 mm and they are Powder Preparation pressurized from the outside. The stationary seal ring is Vacuum Sintering Sand Blasting isolated from the seal gland with springs and is free to adjust Preshape is vacuum sintered Compaction generally between 1370°C and As the material cools, silicon expands, its axial position to mate with the rotor, while the rotor is 1540°C followed by hot isostatic pushing impurities to the surface, Process Preparation keyed to the rotating shaft sleeve. Sample candidate rings are pressing (using inert gas) which are removed by sand blasting. shown in Figure 3 against a one-cm. grid. It can be seen that Processing of SiC Final Machining Final Machining the face of the rotor is wider than that of the stator, so the The sintered material is then final The silicon carbide is then machined to Final Machining stator defines the effective sealing interface. machined by different methods the final form through conventional (EDM, grinding). methods. Production Processes Two seals were tested simultaneously in opposing ends of the existing test rig which was purpose-built for the development of the CAN6 seal. This rig is shown in Figure 4.
Figure 2: Production Process Steps for Tungsten Carbide [3], Reaction Bonded Silicon Carbide, and Pressure-less Sintered Silicon Carbide [4]
4 Copyright © 2012 by ASME Test Conditions The test conditions are summarized in Table 1. Table 1: Conditions for Test A
Test Variable Value
Pressure (MPa) 7.5
Nominal Spring Force (N) 165
PV (MPams-1) 81 Figure 3: Photographs of typical test material rings— stator on left, and rotor on right Balance Ratio 0.68
Both the stator and rotor were lapped to one lightband The spring force applied to the stator was increased by adding concave, both front and back, to provide an initial cumulative shims to increase the severity of some tests. This additional flatness of the two seal faces of two lightbands concave—a spring preload had the undesirable side effect of reducing the condition expected to reduce leakage and increase seal face available seal travel; however the available travel was wear rates. confirmed to be acceptable for the controlled test conditions. Test A Results The rotors and stators that were subjected to the steady-state conditions of Test A did not wear enough to produce reliable quantitative differences in their wear rates—the average wear of the seal faces was close to that of the uncertainty associated with such measurements (tens of micro-inches). However, qualitative results regarding flatness changes, visible wear, and discolouration (caused by frictional heating) were assessed to grade each of the materials tested. A summary of the detailed test results is provided in Table 2. The following comparative assessments were made from this test series. Material-S2 performed poorly when compared to other stator materials. Its suitability as a stator material would be a departure from the stator reference material; therefore these results supported Figure 4: CAN6 Test Rig its removal from the list of potential stator candidates. Test A: Steady-State Wear Material-S3 performed well against the softer Objective materials; however, it failed when run against the hard material Material-R8. This failure did not The objective of Test A was to assess the performance of the eliminate Material-S3 as a stator material, since the full set of candidate seal face materials at typical normal Material-R8 material appeared to cause damage to all operating conditions for a nuclear pump seal—a moderate opposing faces; however the test result was an condition for a well-matched combination of rotor and stator indication of the poor wear resistance of Material-S3. materials. Material-S4 was not available for Test A but was Background included in later tests. During normal operation of a properly designed pump seal, a Material-R8 performed poorly and failed when run fluid film separates the rotating and stationary seal faces. The rotation of one seal face relative to the other applies high shear with Material-S3, thus it was eliminated from the list force to the seal faces due to the thin size of the fluid film and of potential rotor candidates. the viscosity of water. These shear forces result in wear to the faces; the removed wear particles can remain between the seal faces and either act as a lubricant (e.g., graphite) or an abrasive (e.g., silicon carbide). Test A simulated normal nuclear operating conditions with some conservative additions to accelerate wear.
5 Copyright © 2012 by ASME Table 2: Summary of Results of Test A Test A Conclusions
Test Wear Flatness Discolouration Combined Overall, the steady state wear test series screened the initial Material Change Rating material selection to allow future tests to focus on performance differences within the best material families. Rotor-Ref Poor Poor Good Poor Material-S1 and Stator-Ref performed equally well. Very Very The silicon carbide materials performed well, validating their Material-R1 Very Good Very Good Good Good selection as the main rotor material family. Most notably, Material-R1 showed minimal effect on its mating stators while Material-R2 Good Good Good Good maintaining itself in good condition. Material-R3 Good Good Good Good The candidate materials for further comparative testing were as follows: Material-R4 Poor Poor Good Poor Stators—Stator-Ref, Material-S1, Material-S3, and Material-R5 Poor Poor Good Poor Material-S4; and, Rotors—Material-R1, Material-R4, Material-R5, and Material-R6 Good Good Poor Good Rotor-Ref. Material-R7 Good Poor Poor Poor Test B: Low Pressure Operation Material-R8 Poor Poor Good Poor Objective
Material-R9 Poor Poor Good Poor Test B simulated the effect of low pressure operation on the selected materials. This represented an abnormal condition Very that can lead to accelerated wear due to the absence of a Material-R10 Good Very Good Very Good Good lubricating water film between the rotating and stationary seal faces. These conditions provided a means of characterizing Stator-Ref Good Good Good Good the selected materials’ resistance to hard contact rubbing and wear. Material-S1 Good Good Good Good Background Material-S2 Poor Poor Poor Poor During normal operation of a properly designed pump seal, there is a film of fluid maintained between the rotating and Material-S3 Poor Poor Poor Poor stationary seal faces. The predominant source of seal wear is Material-R6 and Material-R7 were both purported to believed to occur when this film breaks down due to be corrosion-resistant tungsten carbide materials, unfavourable deflection of the seal faces, allowing direct however to be considered as a rotor candidate for contact at the rotating interface, otherwise known as hard further testing, they had to perform better than the rubbing. This deflection can be caused by temperature or silicon carbide materials. As this was not the case, pressure transients, as well as the general shape of the seal. both were eliminated from the list of potential rotor The fluid film can be lost when hydraulic forces on the seal, candidates. In addition, colour changes to the due to the geometry and the spring force, result in hard material indicated that neither material was as rubbing. This occurs when the total closing force approaches corrosion-resistant as initially believed. or overcomes the opening force. The term used to describe these forces is the balance ratio. Generally, mechanical seals Material-R2 performed worse than an equivalent are designed for operation at a balance ratio between 0.6 and material, Material-R1. In the interest of reducing the 0.8. Hard rubbing can occur when the balance ratio number of rotor material candidates, Material-R2 was approaches 1.0 or greater. During some reactor transients, eliminated. such as plant start-up, the seal is at low pressure while the Material-R3 performed worse than an equivalent pump is in operation—a condition that results in high balance material, Material-R4. In the interest of reducing the ratios that can exceed 1.0. number of rotor material candidates, Material-R3 was Test B simulated hard rubbing conditions between the seal eliminated. face pairs. When operating at low pressure, the spring force of As candidate rotor materials, Material-R9 and the CAN6 seal design is comparable to its opening force, thus Material-R10 were used to provide a sample of the the high balance ratio is achieved. performance of the aluminium oxide family as a Test Conditions pump seal material. Due to their less desirable material properties and the fact that they did not A balance ratio of 1.4 was achieved by operating the test rig at perform significantly better than the materials in the a pressure of 345 kPa and by increasing the spring force silicon carbide family, both materials were eliminated through the addition of a spacer on the CAN6 seal assembly. as potential rotor candidates. The test conditions are summarized in Table 3.
6 Copyright © 2012 by ASME The test included two stages. The first stage ran for 138 hours Material-R1 outperformed the other rotor candidates, with no source of external heating. The second stage ran for echoing the results of Test A. 120 hours with additional heat input to the test rig. Material-S1 and Material-S4 both outperformed the Table 3: Test B Conditions reference material. Material-S4 developed measurable wear against Material-R4; however it Test Variable Value outperformed Material-S1 when run against Material-R1. Pressure (MPa) 345 Material-S3 was not sufficiently tested to evaluate it Nominal Spring as a candidate. However, it did show very little wear 250 Force (N) when run against Material-R4. Material-R5 continued to perform worse than the -1 70 PV (MPams ) other silicon carbide materials and consequently was eliminated. Balance Ratio 1.4 Test B Conclusions
The low pressure operation test series identified material Test Seals combinations that performed poorly, such as hard coatings or The test seals included two single-stage CAN6 seals installed reduced free graphite. in the CAN6 test rig. The seal faces for Test B included the The candidate materials for further comparative testing were following materials: as follows: Stators—Stator-Ref, Material-S1 and Material-S4 Stators—Stator-Ref, Material-S1, Material-S3, and Material-S4; and Rotors—Rotor-Ref, Material-R1, Material-R4, and, Material-R5. Rotors—Rotor-Ref, Material-R1, and, Material-R4. In most cases, the stators and rotors were refurbished, since Test C: Erosion Resistance many had been used in Test A. All seal faces were lapped to 1.0 ±0.5 lightbands concave. Objective Test B Results Test C simulated an erosion groove across a lapped surface to compare the erosion resistance of the candidate seal face The rotors and stators that were subjected to the low pressure materials. The test conditions described below were intended conditions of Test B experienced a low wear rate similar to to accelerate the erosion rate of the seal face materials being that of Test A. Therefore, as explained for the Test A results, tested, thereby providing a discriminating test. the Test B results reported here are only qualitative as well. Flatness changes, visible wear, and discolouration were Background assessed to grade each of the materials tested. A summary of Both the stator and rotor can undergo erosive wear at hot the detailed test results is provided in Table 4. standby conditions where the seal faces are leaking without Table 4: Summary of Results of Test B rotation. As the leakage is localized, erosive wear can occur and cause higher leakage rates, eventually becoming the Test Flatness Combined predominant source of leakage through the seal and ultimately Wear Discolouration Material Change Rating reducing the life of the seal. Seal face erosion has been identified as a causal factor for many seal failures in the Rotor-Ref Poor Poor Good Poor industry. Very Very The erosion resistance tests simulated an extreme condition Material-R1 Very Good Very Good Good Good designed to accelerate the erosion of the candidate materials to more definitively show differences in the performance of the Material-R4 Poor Poor Good Poor candidate seal face materials. This was achieved by machining notches through the front face of the seal materials. Material-R5 Poor Poor Good Poor The geometry of these notches was chosen to attempt to Stator-Ref Poor Poor Good Poor simulate an abrasive mode of erosive wear. As hard particles within the water enter the notch, they are accelerated against Material-S1 Good Good Good Good the walls. The impact of the particles can cause small cracks and release additional abrasive material, causing further Material-S4 Good Poor Good Good erosion downstream. Test Conditions The following comparative assessments were made from this The test conditions were optimized to maximize the shear test series. forces acting on the walls of the groove. The pressure and
7 Copyright © 2012 by ASME temperature of the inlet water were held at 8.3 MPa and 40°C, Table 5: Erosion Test Inspection Summary respectively, for 400 hours. There was no shaft rotation. The seal faces were lapped to meet the following Condition of Condition at requirements: smooth Test Material groove cut into counter-face seal face Stator front face: 3 - 5 LB Concave opposite groove Stator back face: 6 - 8 LB Concave - Build-up of Rotor seal face: 7 - 9 LB Concave - Deep grooves material Two grooves were machined radially across each of the two - Noticeable erosive - Browning of Rotor-Ref wear mating seal faces and the seal was assembled such that the surfaces grooves were positioned at 90° to each other. The grooves - 50-60 μm depth pit - Widening of were machined to have an estimated notch shear stress of at ID of groove greater than 22 kPa. Figure 5 and the following dimensions C1a exit illustrate the approximate notch geometry; however some - Slight carbon minor deviations from these requirements were allowed, - Slight water build-up provided the desired shear stress was achieved. Stator-Ref staining - Widening of - Flat trace R = 0.38 ± 0.03 mm exit H = 0.33 ± 0.03 mm Maximum radial angular deviation across the seal face = ±2° - Significant - Carbon staining carbon build-up along lapped face Surface finish ≥0.8 μm Material-R1 - Little widening - 1.3 μm build-up of of exit material C1b
- Slight water - Little widening Material-S1 staining of exit - Flat trace
- Little carbon Figure 5: Simplified Notch Geometry - No carbon staining build-up Test C Results Material-R4 or water marks - Little widening - Flat trace The visual inspections of the parts are summarized in Table 5. of exit C2a Test C Conclusions Test C simulated a worst case erosion rate for a notch on the - Slight water - Little widening staining seal face materials. All test conditions were deemed Material-S1 acceptable for comparison. of exit - 0.5 μm build-up of material For the rotor materials, Rotor-Ref showed the most significant erosion. A visual inspection of the erosion surface signalled a - Carbon staining non-corrosive wear mechanism. Compared to tungsten along lapped face carbide, Material-R1 and Material-R4 showed high resistance - Widening of (traces show that it Material-R1 to erosive wear, Material-R4 more so than Material-R1. exit is not erosion) Other than a slight widening of the exits, no stator material - 1.3 μm build-up of showed a measurable amount of wear in the cut grooves or material C2b along the lapped faces. It was observed that Material-S1 had the least amount of groove widening, while Material-S4 and - Slight water - Widening of staining Stator-Ref demonstrated a greater degree of widening. exit Material-S4 - Slight carbon The results of the erosion testing did not eliminate any - Slight carbon build-up candidate stator or rotor materials, as all materials performed build-up as well as the control materials—Stator-Ref and Rotor-Ref. - Flat trace
8 Copyright © 2012 by ASME Test D: Dry Running Table 6 provides a concise summary of the relative performance of the materials tested at simulated dry-running Objective conditions. The objective of the dry-running test was to assess the Table 6: Test D Material Performance Summary robustness of various combinations of seal face materials when tested as rotor-stator pairs at simulated dry-running Visible Comparative Degree of conditions—an extreme condition that occurs during abnormal Material Cracks Wear* operation and is known to severely damage mechanical seals. Testing at such conditions provides a measure of self-lubricity [Yes or No] [High, Medium or Low] and resistance to thermal stress-induced cracking of the seal Material-R1 No Low face material pairs. Test Conditions Material-R3 Yes High
During normal operation of a properly designed pump seal, Material-R4 Yes Medium there is a thin film of fluid maintained between the rotating and stationary seal faces. However, improper venting of a seal Material-R9 Yes High can leave air in the pump cavity which displaces this water film as it passes through the seal faces. Hard dry rubbing of Stator-Ref No High the seal faces occurs when the air, which replaces the thin water film, is unable to resist the forces pushing the seal faces Material-S1 No Low together. During dry rubbing, the shear forces on seals are significantly Material-S3 Yes High higher than normal due to the lack of lubrication, which is Material-S4 No Low provided by water at normal conditions. This causes increased wear on the materials and a high rate of frictional heating. *Medium wear is defined as an average seal face wear in the The high rate of wear combined with a lack of leakage can range of 100 to 200 µin or an average wear track depth in the also cause a build-up of seal face material between the two range of of 6 to 12 µin. High and low wear are therefore seal faces. The lubricity of the interface is determined by the defined as average wear values above and below medium lubricity of any dislodged material and the affinity for the wear, respectively. particles to be dislodged from the seal faces. All three of the other stator materials (Stator-Ref, Material-S1, The carbides (and other ceramics such as oxides and nitrides) and Material-S4) did not crack but Stator-Ref did show the typically used in rotors and stators of mechanical pump seals highest degree of wear. are susceptible to thermal cracking due to the high temperature gradients in service and their brittle nature. Therefore a dry Dry-Running Test Conclusions rubbing condition exaggerates the propensity for a seal face to The dry-running test was successful in differentiating the crack due to the increased temperature gradient arising from performance of the rotor-stator material pairs that were tested the lack of water cooling. Thermal cracking may manifest in that some material pairs performed very well while others itself in composite seal rings as thermal stress-induced cracks, did not. The results of this test also showed that heat checking in carbide rotors, or as blistering in carbon three stator materials—Stator-Ref, Material-S1, and graphite. Although the dry-running test simulates both an Material-S4—exhibited adequate self-lubricity during a extreme and abnormal operating condition, it provides an five-minute test at simulated dry-running conditions whereas important and useful assessment of a stator’s effective Material-S3 did not. Therefore, this test effectively eliminated lubricity under such conditions and the resistance of specific Material-S3 as a candidate stator material for further testing. rotor-stator pairs to thermal damage. The test results showed that Material-R1 was the best Test D Results performing of the four rotor materials tested and that its In general, the individual test series performed within this performance was very good. The thermal cracking resistance project were intended to identify differences in specific of Material-R4 was inconsistent and it showed moderate wear. performance parameters between various seal face material pairings with the combined objective of successively selecting a material combination for further performance testing. With this goal in mind, materials that developed cracks were strongly considered for de-selection from further testing. Additionally, materials or material pairs (rotor-stator pairs) that showed high wear were graded similarly to those that developed cracks. Figure 6 and Figure 7 show typical high and low wear patterns, respectively, for rotors after Test D.
9 Copyright © 2012 by ASME References [1] Redmond, P.E., “Failure Analysis of Four Graphite Pump Seal Faces”, Charles R. Morin Memorial Symposium on Failure Analysis and Prevention, 2009-Oct 25 to 29, Pittsburgh, Pennsylvania, USA. [2] Schunk Kohlenstofftechnik, “Manufacturing Process and Material Properties of Carbon and Graphite Materials”,
10 Copyright © 2012 by ASME