DEVELOPMENT OF A DUAL PURPOSE VISUALIZATION TECHNIQUE FOR THE STUDY OF ROTATING SHAFT SEALS

Transcription

1 Proceedings of the 3rd ASME/JSME Joint Fluids Engineering Conference July 18-23, 1999, San Francisco, California FEDSM DEVELOPMENT OF A DUAL PURPOSE VISUALIZATION TECHNIQUE FOR THE STUDY OF ROTATING SHAFT SEALS Carlos H. Hidrovo/Massachusetts Institute of Technology ABSTRACT This paper discusses the development, improvement and implementation of a dual purpose visualization technique to study rotating shaft seals. The technique is based on the use of a transparent counter face in the form of a quartz shaft to provide a window of observation to the seal-shaft interface and the phenomena that takes place there. It allows for the study of the elastohydrodynamic (EHD) lubrication of rotating shaft seals by means of the visualization of the oil film at this interface with the use of the Laser Induced Fluorescence (LIF) technique. In addition, with the use of time lapse video recordings, investigations of the dirt ingestion failure mechanisms effects on the overall life and performance of rotating shaft seals can be performed. The technique has already proved successful, providing important clues in both the understanding of the lubrication and particle ingestion failure mechanics of rotating shaft seals. INTRODUCTION Although rotating shaft seals have been in use for decades, there is still little understanding on how they work and why exactly do they fail. Most of the improvements and current developments of what is consider state of the art rotating shaft seal technology have taken place on a trial and error basis. This is mainly due to the complex mechanics governing rotating shaft seals, which not only involve different disciplines as hydrodynamics, materials science and tribology, but is further complicated by the intricate coupling of these disciplines. As it is the case with the conceptualization of most engineering systems, there are two approaches to study rotating shaft seals, which need not be mutually exclusive, but actually complementary: the analytical (or theoretical) and the experimental. Due to the previously mentioned complex mechanics of rotating shaft seals, the analytical approach requires the use of very elaborate and complicated models and algorithms, which could be very time consuming, and might not produce the desired results. Furthermore, most of the advancements in the theory and analysis of the mechanics of rotating shaft seals to date have taken place as a result of an experimental study. It is only after experimentation, where sufficient knowledge has been gained and an understanding of Douglas P. Hart/Massachusetts Institute of Technology the basic principles has been achieved, that such advances have been attained. This was the case of the pioneering work done by Jagger (1957), who discovered that a successful seal would run on a thin film of lubricant. This lead to the investigation of the load support mechanism and the development of the hydrodynamic pressure generation by microasperities models, to explain it (Jagger, 1966; Johnston, 1978; Gabelli and Poll, 1990). So is the case with the sealing mechanism, where experimental studies have lead to the suggestion of a reverse pumping mechanism (Kammüller, 1986; Müller, 1987; Horve, 1987, 1991, 1992) and the development of complex elastohydrodynamic models, and the use of computer algorithms, to explain it (Salant and Flaherty, 1994, 1995). As much as experimental studies have contributed to pave the way for further understanding of the working mechanics of rotating shaft seals, it has still been difficult to get a full picture of some of the phenomena going on under the seal. This is mainly due to the geometrical constraints under which rotating shaft seals work, and which make it difficult to study them. In addition, most of the studies have concentrated on the lubrication and working mechanics of rotating shaft seals, but practically no work has been done on the failure modes of rotating shaft seals. Of particular interest is the failure mode due to dirt ingestion, since it is thought to account for as much as 80% of all failed seals. One of the purposes of this research was to develop an experimental technique that would allow additional studies of some of these phenomena by observation and examination of the area underneath the sealing lip (sealshaft interface), under normal operating conditions. A visualization technique, based on previous work by Poll et al., 1992, that would provide a window of observation at the sealshaft interface was developed, allowing for two types of studies: oil film thickness and dirt ingestion failure. EXPERIMENTAL APPROACH A new, but widely used, type of rotating shaft seal labeled unitized was selected as the test subject for this research (Fig. 1). The main reason for this decision is that not much work has been done on this type of seals and, as it is the case with most rotating shaft seal technology developments, there is no real 1 Copyright 1999 by ASME

2 understanding of why these seals outperform their counterparts, namely the radial lip seals. In addition, the results of this project could be compared with those of past research done on radial lip seals, in order to obtain a clearer and more global picture of the working mechanics that must be common to all kinds of rotating shaft seals for successful operation. due to the difference in refractive indices between quartz and air, the light bounces back inside the shaft, in a similar fashion as in a fiber optic (Fig. 2). Once the light reaches the seal-shaft interface, if it encounters the surface of the seal contacting the shaft, the incident light is reflected by that seal surface. If the light encounters oil, it leaves the shaft and travels through the lubricant, since quartz and oil have very similar refractive indices, where it can be absorbed or reach the seal surface and be reflected by it. Figure 3 shows the area of observation and the images obtained with this experimental setup. Figure 1: Two types of rotating shaft seal designs. For the study of the oil film thickness, the Laser Induced Fluorescence (LIF) visualization technique was adopted, since it has been widely used in the past, with great success, for oil film thickness measurements of engine piston rings during operation. This scenario is closely connected to the one in this research, where the rotating shaft seals studied are those used in engine crankshafts, making it a perfect guideline on which to base this work. For the study of the dirt ingestion failure mode, it was decided to use a test dust with a known particle distribution, in this case the Arizona fine dust. Once a window of observation to the seal-shaft interface is provided, the dirt ingestion failure process can be recorded with the use of a camera and frame grabber, to take pictures of the phenomena going on under the seal. Since the dirt ingestion failure process takes several hours, the best way to study it is to speed it up. This can be accomplished by subjecting the seal to a very dusty environment and by making sure that the dirt goes into the sealshaft interface. This would in fact accelerate the real time failure process, by subjecting the seal to very extreme conditions. Another option is to use time lapse video, where the process is artificially accelerated by means of taking pictures or videos of the dirt ingestion phenomena at a given rate or speed and then playing these pictures or videos at much higher rates or speeds. In addition, the dirt ingestion failure mode study can be further aided with the use of fluorescent particles, which are mixed with the test dust. This allows better visualization and tracking of the dust as it makes it ways through the area of interest (in this case the seal-shaft interface). EXPERIMENTAL SETUP In order to gain optical access to the seal-shaft interface, a hollow quartz shaft was manufactured. This would indeed provide a window of observation to this area, with minimal perturbation of the normal operating conditions of the rotating shaft seal. Due to the good optical qualities of the quartz, the shaft can be rotating without hindering the observation of the seal-shaft interface area. For full optical access to the area, a mirror angled at 45 must be located in the hollow cavity of the quartz shaft. In order to provide illumination to the area of observation, a scheme based on the total internal reflection of light inside the shaft (Poll et al., 1992) was used. The light is injected through the frontal beveled face of the hollow shaft and Figure 2: Experimental setup schematic. Figure 3: Area of observation relative to seal lip. An 8-bit, 512x460 pixels, CCD camera with a maximum frame rate of 30 Hz and minimum shutter exposure of 1/16,000 sec., was used to obtain images of the seal-shaft interface. For LIF purposes, the oil was mixed with a fluorescent substance. Two different dyes were tried: Rhodamine 6G and Pyrromethene 567. Initial tests were carried out using Rhodamine 6G. However, Pyrromethene 567 proved to be a much better dye because of its higher fluorescent efficiency and oil solubility when compared to Rhodamine 6G. Likewise, two types of Lasers were used: a continuous Argon-Ion Laser and a pulsed Nd:YAG Laser. The continuous Laser was used to obtain average oil film thickness images. Since normal operating conditions for the rotating shaft 2 Copyright 1999 by ASME

3 seals studied required the shaft speed to be 1800 RPM (30 Hz), using the Argon-Ion Laser with the camera running in a continuous mode (no shutter), provided LIF images of the average oil film thickness over one revolution of the shaft. To obtain instantaneous images of the oil film thickness, the pulsed Nd:YAG Laser had to be used, since even with the minimum shutter exposure of the camera, a ten pixel displacement of the shaft would have been recorded. This is avoided with the Nd:YAG Laser which provides light for a very short period of time (7-9 ns pulse), recording a three thousands of a pixel displacement of the shaft, and effectively freezing the image (Fig. 4). Figure 4: Camera aperture time and laser pulse width comparison. Figure 5: Instantaneous LIF images. Top Right: Schematic of oil film thickness. Bottom Right: Grayscale versus pixel position plot. To make dirt ingestion recordings, images of the seal-shaft interface, under dusty conditions, were taken every 5 seconds, until the seal failed (visually detected profuse oil leakage). The pictures were then played at a rate of 15 frames per second, speeding up the failure process 75 time. The experimental setup, which allows the observation of a particular and unique location on the seal, was further improved with the use of a rotary shaft encoder to synchronize all the equipment with respect to the quartz shaft. This provided and additional restraining handle on the setup, which could be used to fix the seal-shaft interface location to be observed and studied. This would provide better quality time lapse videos, by getting rid of some of the "noise" produced by random surface irregularities along the circumference of the shaft during the acquisition of a movie. OIL FILM THICKNESS The LIF images of the lubricant at the seal-shaft interface provided a qualitative measure of the oil film thickness through this interface. It was found that instead of decreasing monotically from the air side to the oil side, as would have been expected from the shape of the oil lip when the seal is not installed, the oil film thickness reaches a minimum in the middle of the oil lip contact band. In addition, it was found that the grooves etched on the surface of the oil lip do not display that much oil (Fig. 5). These grooves are though to have a reverse pumping effect, which prevents the seal from leaking. It was believed that they accomplished this effect by carrying oil in themselves from the air side to the oil side, against the pressure difference. However, the LIF images show that they do not seem to carry that much oil at all, so this is not the mechanism by which they produce reverse pumping, if they do at all. Figure 6: LIF images of average oil film thickness under forward running (normal) conditions. Top: 900 RPM. Bottom: 1800 RPM. Figure 7: LIF image of average oil film thickness under reverse running conditions. 3 Copyright 1999 by ASME

4 Investigation of the oil film thickness versus shaft rotational speed revealed a film thinning with increasing speed behavior (Fig. 6). Of interest is the fact that this film thinning occurs mainly towards the air side, with the oil side film thickness remaining essentially constant. The film thinning could be attributed to a reverse pumping effect, although changes in the oil viscosity due to higher temperatures within the lubricant, induced by higher sliding surfaces speeds, could contribute to the film thinning behavior. LIF images under reverse running of the shaft revealed that the oil film at the sealshaft interface is disrupted and can no longer be sustained, suggesting leakage (Fig. 7). This agrees with observations and assessments of oil leakage during tests with reverse running of the shaft. DIRT INGESTION FAILURE One of the expectations from the dirt ingestion recordings was to have the ability to quantify the degree of failure incurred by the rotating shaft seal through time. It was thought that the seal surface would incur an aggressive wear, which could have been quantitatively measured. Previous work done by Ayala et al., 1995, had used the lapse time video technique on oscillating face seals, in order to asses failure and wear on these type of seals when subjected to an abrasive slurry environment. Successful measures of the wear band on the seal surface through time were achieved with this technique (Fig. 8). dirt layer sediment and dirt particles clustering on the surface of the rotating shaft seals. By carefully reviewing the dirt ingestion time lapse videos, it was found that the "pits" most likely correspond to irregular indentation and clustering in the dirt layer that forms on the rotating shaft seal surface. Figure 9: Dirt ingestion at failure. Note the presence of irregular "pits" on the upper part of the oil lip. Figure 8: Oscillating face seals in an abrasive slurry. Left: Start of test. Right: End of test. The wear band that has formed on the face of the seal is quite visible at the end of the test. The dirt ingestion time lapse videos showed that there was no particular "wear" pattern on the surface of the rotating shaft seal. An arrangement of irregularly placed amorphous "pits" started to form on the surface of the rotating shaft seal contact band area (Fig. 9). The number and size of these pits seem to increase with time, and more importantly, their progression seem to coincide with the onset of profuse oil leakage (i.e., seal failure). It was initially thought that these "pits" corresponded to irregular wear zones on the rotating shaft seal surface. A posteriori examination of the surface of failed rotating shaft seals, after clean up, with the use of a Scanning Electron Microscope (SEM) revealed that even though the surface of the seals incurred some wear, it was minimal, and it did not corresponded to the "pits" observed on the dirt ingestion videos. The "pits" were far greater in size and much more irregular than any of the observed wear features on the surface of the seal. Further examination of failed rotating shaft seals right off from the test rig, no clean up, under the SEM revealed a These observations laid the groundwork for the proof of failure due to dirt ingestion. This failure mechanism can be described as follows: Dirt particles enter the seal-shaft interface (1) and begin to accumulate on the rotating shaft seal s surface, forming a dirt layer that coats most of the oil lip s surface (2). As time progresses, the dirt layer grows thicker with the ingestion and adhesion of more dirt particles entering the sealshaft interface (3). As the dirt layer grows thicker, it also becomes more irregular (4) with dirt particles accumulating and clustering at certain points while at other points detaching due to the shearing motion of the rotating shaft. This leads to the formation of the "pits" on the surface of the dirt layer (5), the size and shape of the irregularities being proportional to the thickness of the dirt layer. The growing dirt layer starts to push the seal away from the shaft, increasing the distance between the seal lip and the shaft. However, the effective average of the gap remains unchanged because of the layer of dirt that has attached to the seal s surface. As the dirt layer grows, it becomes more irregular. As portions of the dirt layer are detached through the shearing motion of the shaft, the "pits" can amalgamate, producing channels of sufficient size for the oil inside the chamber to leak (Fig. 10). 4 Copyright 1999 by ASME

5 for oil to fill and flow through them (this is probably also related to dirt layer thickness). Finally, there must be enough interconnections between the dirt layer "pits" to allow the oil from the sump to flood the seal-shaft interface, culminating in massive failure (profuse leakage). Figure 10: Dirt ingestion failure mechanism sequence. As the dirt layer grows, the gap at the seal-shaft interface increases. The dirt layer grows until the onset of aggressive "pitting" causes the system to leak. Although the quartz shafts used on the dirt ingestion tests did show very clear and distinct wear tracks, it is believed that the main reason rotating shaft seals fail due to dirt ingestion is because of this dirt layer formation on their surfaces, and the subsequent seal-shaft interface gap increase. For further proof of this theory, a rotating shaft seal "failed" (profuse oil leakage) during a dirt ingestion test was clean up and the dirt layer removed from its surface. It was then run, on top of the wear tracks of the quartz shaft, for six hours with forward rotation and six hours with reverse rotation. No detectable amount of oil leakage was register during the entire test. A dirt test was also performed using LIF in order to asses the oil film thickness at the seal-shaft interface during the dirt ingestion failure process. For this test only the oil was mixed with the fluorescent dye (Pyrromethene 567), so fluorescence (and therefore light intensity in the pictures) would be proportional to oil film thickness and independent, to a first approximation, of particle concentration under the seal-shaft interface. During actual testing it was found that due to the mixing of the dust with the oil, it was possible to partially track some of the dirt that had clustered and trapped oil. Figure 11 shows a sequence of snapshots taken during the test, where it can be seen how the dust starts to form dirt clusters that creep under the seal-shaft interface. Of special interest is the fact that while the progression and accumulation of clusters under the seal-shaft interface takes place, the oil film thickness remains essentially unchanged, up to the point where an almost sudden inrush of oil from the sump fills the seal-shaft interface and failure occurs (profuse leakage). This seems to support the theory that failure occurs due to the formation of a dirt layer on the seal surface, which increases the actual seal-shaft gap and subsequently develops channels through which oil can flow and flood the seal-shaft interface, culminating in massive failure (profuse oil leakage). The sudden nature of the seal failure (inflow of oil into the seal-shaft interface with subsequent leakage) is due to the fact that certain factors have to come together in order to produce leakage. The dirt layer, which presumably builds up gradually, needs to reach a certain thickness, so that the actual seal-shaft gap (distance between the surface of the seal and the surface of the shaft, disregarding the presence of a dirt layer) is large enough to overcome the possible surface tension effects that keep the oil film intact and within the seal-shaft interface. In addition "pits" in the dirt layer must be big enough to allow Figure 11: Sequence of snapshots taken during a dirt test where LIF was used to monitor the oil film thickness at the seal-shaft interface. Light intensity is proportional to oil film thickness. The sudden inrush of oil through the interconnected "pits" can be seen in the last two snapshots. Finally, a dirt ingestion test using fluorescent particles was performed. The particles were made of a polymeric resin that when cured encapsulated the desired fluorescent dye (in this case Rhodamine 6G). The polymeric nature and spherical shape of the particles made them non-abrasive, helping assure that they would only act as tracers for the dirt, with no mayor particular role in the dirt ingestion failure. To further ensure that the particles would not interfere or have a particular effect on the dirt ingestion failure mode, and to provide even distribution of the particles with the dust, their size distribution was kept accordingly with that of the test dust. A mixture 25% fluorescent particles, 75% dust by volume was prepared. Figure 12 shows a sequence of snapshots taken during the test. Assuming an even mixing and distribution of fluorescent particles with the dust, and since the oil was not mixed with a fluorescent dye, light intensity in the pictures corresponds to dirt concentration. The progression and clustering of dust particles in the seal-shaft interface can be seen in this sequence. 5 Copyright 1999 by ASME

6 Figure 12: Sequence of snapshots taken during a dirt test where fluorescent particles were used to track the dust penetration and clustering at the seal-shaft interface. Light intensity is proportional to dirt concentration. CONCLUSIONS The experimental technique described in this paper proved to be an exceptional tool in the study of rotating shaft seals. In addition, it has many advantages over similar techniques used in the past. Normal operating conditions of rotating shaft seals are closely replicated by having the quartz shaft rotating instead of the seal, as is the case with most rotating shaft seals visualization techniques. The use of the pulsed Nd:YAG laser allows for instantaneous, still pictures of the phenomena going on at the seal-shaft interface. Finally, the use of time lapse video allows for better understanding of the long term phenomena that might play an important role in the working mechanics and overall life of rotating shaft seals, and that might not be easily disclose by short term examination. Some preliminary observations of the oil film at the sealshaft interface of "unitized" rotating shaft seals, using LIF, have been made, providing some qualitative understanding of its behavior. More significantly the use of time lapse video has provided important information regarding the dirt ingestion failure mechanisms that affect the long term behavior and overall life of rotating shaft seals. REFERENCES Ayala, H., Hart, D. P., Boyce, M., 1995, current work for the MIT-CATERPILLAR Seal Science Program. Gabelli, A. and Poll, G., 1990, "Formation of Lubricant Film in Rotary Sealing Contacts - Part I: Lubricant Film Modeling", Joint ASME/STLE Tribology Conference, Toronto, Canada, October Horve, L., 1987, "A Macroscopic View of the Sealing Phenomenon for Radial Lip Oil Seals", Proceedings, 11th International Conference on Fluid Sealing, B. S. Nau, ed., BHRA, pp Horve, L., 1991, "The Correlation of Rotary Shaft Radial Lip Seal Service Reliability and Pumping Ability to Wear Track Roughness and Microasperity Formation", SAE International Congress and Exposition, Detroit, paper No Horve, L., 1992, "Understanding the Sealing Mechanism of the Radial Lip Seal for Rotating Shafts", Proceedings, 13th International Conference on Fluid Sealing, B. S. Nau, ed., BHRA, pp. 5-20, Kluwer Academic Publishers. Jagger, E. T., 1957, "Study of the Lubrication of Synthetic Rubber Rotary Shaft Seals", Proceedings, Conference on Lubrication and Wear, Inst. of Mech. Eng., pp Jagger, E. T., 1966, "Further Studies of the Lubrication of Synthetic Rubber Rotary Shaft Seals", Proceedings, Institution of Mechanical Engineers, Vol. 181, Part 1, No. 9, pp Johnston, P. E., 1978, "Using the Fraction Torque of Rotary Shaft Seals to Estimate the Film Parameters and the Elastomer Surface Characteristics", Proceedings, 8th International Conference on Fluid Sealing, B. S. Nau, ed., BHRA. Kammüller, M., 1986, "Zur Abdichtwirkung von Radial- Wellendichtringen", Dr.-Ing. Thesis, Universität Stuttgart, Stuttgart. Müller, H. K., 1987, "Concepts of Sealing Mechanism of Rubber Lip Type Rotary Shaft Seals", Proceedings, 11th International Conference on Fluid Sealing, B. S. Nau, ed., BHRA, pp Poll, G., Gabelli, A., Binnington P. G., Qu, J., 1992, "Dynamic Mapping of Rotary Lip Seal Lubricant Films by Fluorescent Image Processing", Proceedings, 13th International Conference on Fluid Sealing, B. S. Nau, ed., BHRA. Salant, R. F. and Flaherty, A. L., 1994, "Elatohydrodynamic Analysis of Reverse Pumping in Rotary Lip Seals with Microundulations", ASME JOURNAL OF TRIBOLOGY, Vol. 116, pp Salant, R. F. and Flaherty, A. L., 1995, "Elatohydrodynamic Analysis of Reverse Pumping in Rotary Lip Seals with Microasperities", ASME JOURNAL OF TRIBOLOGY, Vol. 117, pp Copyright 1999 by ASME