Titanium MATERIALS You will find the figures mentioned in this article in the German issue of ATZ 10/2003 beginning on page 978. Flächendichtstellen für hohen dynamischen Innendruck in der Getriebetechnik
Sealing Surface Areas for High Dynamic Interior Pressure in Transmission Technology
Extensive experimental and theoretical investigations on sealings between housing parts were carried out at the Institut für Maschinenelemente of the University of Stuttgart. These housings, for example for hydraulic controls for automatic transmissions, are exposed to high dynamic interior pressure. The conclusions of this FVA research project were summarized in a design catalogue with recommendations for the engineer in designing a reliable seal. This catalogue allows the engineer to test the effect of particular parameters as early as in the concept phase.
1 Introduction
By Kuno Fronius, Martin Jäckle and Bernd Bertsche
ATZ worldwide 10/2003 Volume 105
Pulsating interior pressure in transmission components results in the fact that screws for the housing parts are additionally extended, thus relaxing the pressure on the loaded parts. This dynamic process has a particularly negative effect on the profile seal. The seal is additionally relaxed in zones of low area pressure, or might even be pushed out. Housing parts with low stiffness are increasingly deformed, thus producing gaps and splitting, which, in the worst case, may lead to leakage. For this reason, the association Forschungsvereinigung Antriebstechnik e. V. (FVA) has initiated a research project with the title “Gehäusegestaltung unter pulsierendem Innendruck” (Housing Design for Exposure to Pulsating Interior Pressure) under project No. 308/II. The project was supported financially by the Arbeitsgemeinschaft industrieller Forschungsvereinigungen e. V. (Committee for Industrial Research Associations) “Otto von Guericke “ (AiF) under the project No. 12450N. The AiF
enables research and development to be carried out for small and medium-sized companies. The present project was conducted at the Institut für Maschinenelemente (IMA, Institute for Machine Elements) of the University of Stuttgart. The IMA has been active for many years in the examination of highly dynamically loaded sealing areas [2–6], Figure 1. These sealing areas are basically loaded with a torque. In this field of research, the focus was on flange joints that have been designed in accordance with a lightweight construction concept. Due to the lightweight construction, increased relative movements at the sealing plane take place, thus provoking damage to the sealing material. Several test rigs were set up to perform the experimental tests on the transmission housings. Examples of the application of flat gaskets for transmissions that are loaded with a pulsating interior pressure include hydraulic controls, such as control plates for automotive automatic transmissions, or pumps for auxiliary transmission units.
25
RESEARCH
Sealings
Figure 2 shows the hydraulic control plate with the channel plate of an automatic transmission. For the systematic examination of sealed joints under high dynamic interior pressure, comprehensive experimental and theoretical investigations were also conducted. The results have been summarised in a catalogue of engineering and design recommendations that help the engineer to design the best sealed joint from a given sealing material with the desired flange design. The final report [1] contains detailed results of the project. 2 Experimental Tests with Pulsating Interior Pressure
The experimental tests are sub-divided into various test series. In the first part, the load limits of the materials used are examined. In this part, numerous sealing materials are examined in order to determine their discriminating features. In the following part, selected materials undergo standard tests on model flanges. In these tests, the material behaviour of the seal is examined under almost real conditions. A particular feature of these investigations is the homogeneous pressurisation of the seal, using the interior pressure test rig. The findings from these tests then form the first directives for testing real flanges. The geometry of these flanges no longer has a circular ring shape, as was the case with the model flanges, but follows the geometry of the channel plate of a control box. 2.1 Description of the Interior Pressure Test Rig
The interior pressure tests were conducted with the test cells as shown in Figure 3. The circle-shaped seal ring to be tested (di = 40 mm; da = 60 mm) is positioned between two model flanges and, simulating a real screwed condition, is exposed to an approximated real pressure by means of a screw. In doing so, the seating behaviour and the resulting relaxation of the seal under a given interior pressure are simulated under conditions that are very close to reality. A hydraulic unit loads the interior chamber with an adjustable pulsating interior pressure. 2.2 Load Capacity Tests
The load capacity tests serve first of all to classify the materials. Secondly, they help to examine the minimum pressure that is necessary for a given sealing material in order to withstand a predetermined interior pressure. Comprehensive tests were conducted to determine the stress or tightness curves, during which the parameters of the
26
interior pressure and the loading of the seals were varied. The tests were conducted at room temperature. An increase in temperature brings about better test conditions, since the seal is exposed to an increased pressure due to the differing extensional factors of the pressed components. The process of recording the tightness curves can be described as follows. Each seal is pre-loaded with an assembly pressure of 1 to 10 MPa in stages of 1 MPa each. At each pressure stage, the interior pressure is increased with pulsation until the seal leaks and eventually fails. A new seal is used for each new pre-loading, which means that ten test runs are necessary for one tightness curve. In Figure 4, the best candidates of the sealing materials tested are plotted with their corresponding failure curves for a comparative evaluation. The application field of a seal must be precisely defined in order to allow the most suitable sealing material to be chosen. For loads of below 6 MPa, significant differences are noticeable between the candidate seals. In the upper range of loads, with 8 to 10 MPa, the seals have a practically identical behaviour. In the load range up to 2 MPa, metal seals with reinforced seams show the poorest performance. Such seals require a certain basic pressure before a good sealing capacity is achieved. Anaerobically hardening liquid seals have very good results, particularly at low stresses. Liquid seals with a high adhesive capacity are tight in all stress ranges up to the load capacity limit of 140 bar. However, disassembly is only possible to a limited extent, which is a disadvantage. After characterization of the various sealing materials on the basis of their tightness curves, some especially selected materials are now exposed to a specific load collective. Based on the real-life load collective, as shown in Figure 5, the influences of temperature, bar width, flange surface, flange material, hydraulic medium, initial pressure, interior pressure and pulsation frequency on the sealing material are now examined. The surface pressure of the resilient seal and the metal seals with the reinforced seam drops in the seating phase to approximately 9 MPa. The anaerobic liquid seal, which only fills the roughness depths of the surface, has a tightening thickness of a few micrometres, with the result that its seating is less pronounced. During the heatingup phase to 70 °C, the surface pressure of the sealing material increases due to the different heat extension factors of the steel screw and the aluminium flange. Here too, strong differences can be noticed. The small
tightness thickness of the liquid seal leads to an increase in surface pressure after the heating-up phase of the test rig to 17 MPa. The other two seals are further compressed. The surface pressure only reaches a value of 15 MPa. In the operational phase, the seal is relaxed, and, due to the interior pressure, each pressure phase further relaxes the interior pressure. At 120 bar, the surface pressure of the resilient gasket and the metal seal with a reinforced seam drops to 12 MPa. During the operational phase, the seals show practically identical behaviour. After relief of the interior pressure and cooling down to room temperature, the surface pressure once again decreases. It becomes clear that the value of 10 MPa that was preselected at the start of the tests could not be achieved again at the end of tests. 2.3 Tests on Real Flanges
The results obtained from the examinations of the model flanges served as a basis for further investigations. In doing so, parameters with little influence on tightness were left out. The test cells were also modified and were given a shape closer to reality. While the interior pressure test rig for the model flanges was loaded by a single screw, screwing at the new test rig is done by several screws, Figure 6. The screwing pattern and type were chosen in compliance with those used in hydraulic control units for automatic transmissions. This test rig allows the behaviour of the seals to be simulated as close to reality as possible. Hydraulic oil with a defined service pressure is supplied via the side of the first channel. The second channel remains without pressure. The leakage that flows over from the oil-supplying channel via the seal bar is then measured. The leakage that is produced can be directed into a measuring cup via a hose that is laterally connected with the non-pressurised channel. An Oring ensures that the test cell is isolated from the exterior surroundings. For the test evaluation, the method of statistical test planning is employed. The leakage rate per minute is defined as the target variable. The limit value is pre-defined with 60 ml/min. When this value is exceeded, the test is stopped. Statistical test planning offers several possibilities for test evaluation. First of all, the parameters are related to the sealing materials and are presented in a graph. In statistical test planning, these target graphs are called “Interaction Diagrams”, since they show the interaction of two influencing factors on one variable. An interaction diagram exists for each pressure stage. Thus, it is possible to make detailed statements on the cause of the leakage
ATZ worldwide 10/2003 Volume 105
Titanium
quantity. The leakage rate is presented at its normalised value on the Y-axis of the graphs. A high bar represents a high leakage. The temperature is a very important parameter. In practice, a sealed joint is fitted at room temperature. In operation however, such a flange connection may easily reach a temperature of 70 °C, while other extreme conditions include temperatures of below 0 °C. Both temperature extremes are simulated in the examinations of the optimised flange. Apart from sealing materials, flange materials are also examined. The diagrams shown in Figure 7 clearly exhibit that strong differences prevail in the tested temperature ranges in combination with the flange material. Figure 8 presents an overview of all factors of influence in a Pareto diagram (influence diagram). A high bar means that the modification in the pertinent parameter has a great influence on the amount of leakage in the system. Accordingly, parameters with small bars are less relevant for design optimisation. The influence of the parameters is also dependent on the prevailing pressure of the medium. The sealing material has a predominant influence at all four pressure stages. This means that significant differences exist between the materials tested. Further important parameters are the distance between the screws, the flange material and the temperature. 2.4 Oncoming Flow Examinations
A further aspect is the resistance of seals against oncoming flows. Particularly for channel plates of automatic transmissions, non-pressurised zones exist alongside channels in which the pressurising medium bypasses the seal under high pressure. Therefore, the sealing material must possess a defined resistance against the oncoming flow in order to prevent damage at the sealing surface after many hours of service. Apart from affecting the seal, such damage would also conclusively lead to a contamination of the hydraulic medium by the separation of seal particles. In order to determine such resistance of the sealing material against oncoming flows, a test rig was designed that simulates real oncoming flow conditions in a simple manner, in the same way as they occur, for instance, at the channel plate of an automatic transmission, Figure 9. Our experimental examinations included tests in which the oil bypasses the seal, flowing from different directions at different flow speeds and pressures. The oil first hits the
ATZ worldwide 10/2003 Volume 105
face side of the seal. Subsequently, it is directed to by-pass the seal in such a way that angles of attack of 0°, 45° and 90° are achieved. The testing device allows transit flows of up to 40 litres/min at pressures of up to 160 bar. The sealing materials are sub-divided into two groups of resilient seals and coated metal seals. Figure 10 presents examples of two seals from each group. The characteristic oil flow is shown on the left-hand side of each partial figure. In order to make descriptions and positions at the seal more easily understandable, the marked points were provided with letters (A to E). The partial figure contains the total seal (left) and a scaled-up sectional view (right). The seals are always examined after each load stage. In the resilient seal 2, serious damage is already noticeable after the first load stage. The seal is holed at deviation point C. This is caused by the turbulent flow with corresponding aggressive vortices provoked by the sudden change in direction of the oil flow. These holes become worse and worse, stage by stage. Further damage is produced at the oncoming track DE. This is also a deviation point and the oil may penetrate below the seal, with the result that the turbulent flow practically perforates the seal. Roughness phenomena become more and more significant at the oncoming track BC. The influence of the different angles of attack can clearly be seen at the resilient seal. Point B with a 45-degree angle is less damaged than point D with a 90-degree angle of attack. The coated metal plate shows no noticeable surface damage after 40 hours of service. 3 Design Catalogue
The findings from the experimental and mathematical examinations are summarized in a catalogue for design and shaping recommendations [1]. 3.1 Shaping Recommendations using the Example of a Pump Lid
On the basis of comprehensive FE calculations, the pump lid stiffness of a (bushing) gear pump was optimised. An evaluation of the single variants of the pump lid according to stiffness and pressure distribution and dependent on fin shaping and arrangement is presented in Figure 11. The more black points a design has, the better is their sealing behaviour. 3.2 Shaping Recommendations using the Example of a Flange with Metal Seals
The shaping of the flange geometry for
MATERIALS
metal seals with reinforced seams is relatively uncritical. But the following recommendations should be observed: ■ The flange width should be at least twice as wide as the seam of the seal. Thanks to the line shape of the seam, very high pressures can be achieved locally, even with low screwing forces. ■ The roughness Rz of the flange surface should not exceed 50 % of the NBR rubber coating, so that the roughness valleys can be sufficiently filled. ■ The unevenness of the flange surface does not pose problems for metal seals with reinforced seams, since the seam is able to compensate sufficiently. ■ The metal seals with reinforced seams can bridge gaps of between 0.2 and 0.5 mm. The spring-back ratio sometimes exceeds 50 %. ■ The carrier material of metal seals with reinforced seams may be construction steel, spring steel or stainless steels, although spring and stainless steels show better properties than construction steels (gap bridging, yield limit). ■ It should be ensured that the limit area pressure of the flange material (about 200 MPa for aluminium) is not exceeded, as otherwise the flange material will be plastically deformed. 3.3 Shaping Recommendations for a Flange for Resilient Seals
If resilient seals are used under pulsating interior pressure, the following recommendations should be taken into account: ■ Resilient seals fail in an explosion-like manner. The material strength should be high enough to avoid expulsion by interior pressure. ■ The seal thickness is the surface of attack for the interior pressure. In this case, thinner seals perform better. However, they should not be too thin, as otherwise their adaptability and spring-back resilience behaviour will be negatively affected. ■ When the seals are punched, care should be taken that the area of cut is not damaged or cracked. The resistance behaviour of a damaged seal is strongly impaired by the notch effect. ■ Concentric grinding marks or rotation marks at round flanges may have a counterproductive effect when “blowing-out” the seal. ■ The utmost attention must be paid to the design and shaping of the flange width for resilient seals. The seal width should be selected to be as wide as necessary in order to ensure sufficient surface pressure. On the other hand, seals that are too narrow tend to tear out rapidly when exposed to interior pressure.
27
RESEARCH
Sealings
3.4 Shaping Recommendations for a Flange for FIP Liquid Seals
4 Conclusion
For FIP (formed in place) liquid seals, there are noticeable differences between the various liquid seal systems and technologies. The hardening mechanism of the product used must be observed. The transition from sealing to adhesion is progressive. This is why only product-specific recommendations for flange shaping can be given: ■ The film thickness of the liquid seal can have a range of between a few micrometres up to 0.3 mm. The liquid sealing medium is normally built-up in liquid form and pressed out from the sealing surface when assembled, for example when the screws are tightened. If the film thickness is low, the medium will fill only the roughness valleys of the flanges. Should a higher thickness be desired, the design of the flanges becomes more complex, since range spacers or grooves must be positioned. The surface roughness of the flanges (up to RZ = 25 μm) plays a subordinate role in interior pressure sealing. ■ When liquid seals are put into operation, the surfaces of the flanges must be free of dust and grease before the application is started. ■ For reliable hardening, anaerobic products such as liquid seal media require a flange width of between 5 and 7 mm in the zone between the screws. Below the screws, however, the flange width may be lower, due to the higher flange pressure. ■ Liquid seals have a lower spring-back resilience capability due to their lower film thickness. Therefore, the flange should have a design that is resistant to bending. Thanks to the adhesive properties, substantially higher pressures may be sealed compared to metal seals with reinforced seams or resilient seals.
Pulsating interior pressure in transmission components leads to additional extension of the screwing parts, thus producing relaxation of the parts involved. This dynamic process basically has a negative effect on the seal. As a result, in zones with a low surface pressure, the seal will be relaxed and might even be expelled. Housing parts with a low stiffness are increasingly deformed, producing gaps and splitting. In the worst case, this may lead to leakage. In order to carry out a systematic examination of this problem, several test cells for interior pressure tests were installed at the Institut für Maschinenelemente of the University of Stuttgart. At the same time, calculations based on the Finite Element Method (FEM) were carried out. The findings from the experimental and theoretical examinations are summarized in a design catalogue for design and shaping recommendations, in order to ensure the safe operation of sealed joints under interior pressure. The design catalogue was supplemented by so-called nomograms. The designer can use these diagrams to graphically determine the required parameters of the desired sealed joint in a simple way on the basis of only a few initial values. Future examinations will focus more thoroughly on tests of liquid seals. Work on this FVA research project has shown that the sealing and failure behaviour of liquid seals shows noticeable differences depending on whether one is using an anaerobically hardened seal, a film-forming seal or a silicone-based seal medium. In 2003, a project is being conducted within the scope of the FVA research with the topic “Limits of Application of Liquid Seals” at the Institut für Maschinenelemente.
[5]
[6]
[7]
[8]
[9]
Hettich, V.: Identifikation und Modellierung des Materialverhaltens von dynamisch beanspruchte Flächendichtungen. Dissertation, Institut für Maschinenelemente, Bericht Nr. 62, Universität Stuttgart, 1996 Lechner, G.; Naunheimer, H.: Automotive Transmissions: Fundamentals, Selection Design and Application. ISBN 3-540-65903-X, Berlin: Springer-Verlag, 1999 Gladen, R.; Dußler, K.: Eine einfache Methode zur Ermittlung der Einbauflächenpressung bei Flachdichtungen. In: Dichtungstechnik, Nr. 1, 1998, Vulkan-Verlag, Essen Kreuzer, R.; Romanos, G.: Zuverlässigkeit von Flächendichtungen auf Basis von Flüssigdichtmitteln unter dynamischer Beanspruchung. Vortrag in VDI-Bericht Nr. 1579, VDI-Verlag, Düsseldorf, 2000 Friedrich, H.; Wunderlich, P.; Brügel, E.: Prüfverfahren für das Festigkeitsverhalten von Weichstoff-Flachdichtmaterialien. In: Dichtungstechnik, Nr. 1, Mai 2001, Vulkan-Verlag, Essen
3.5 Nomograms
Tests have shown that the failure pressure of a sealed joint depends on many parameters. One way of graphically exhibiting these parameters in an understandable and clear form is to use so-called nomograms or straight-line charts, Figure 12. From one given parameter, the designer can graphically determine the magnitude of the other parameters. Moreover, it is possible to evaluate an existing sealed joint and predict the failure pressure on the basis of the nomograms. The graphs and data of the nomograms were gained in the above-mentioned tests with real flanges. They may be used as orientation values for other flanges. Tests must still be preformed, however, to confirm the safe operation of each individual sealed joint.
28
References [1]
[2]
[3]
[4]
Fronius, K.; Jäckle, M.; Bertsche, B.: Gehäusegestaltung im Abdichtbereich unter pulsierendem Innendruck. Abschlussbericht 308/II, FVA-Forschungsheft Nr. 689, Forschungsvereinigung Antriebstechnik e. V., Frankfurt/Main, 2003 Kubalczyk, R.: Gehäusegestaltung von Fahrzeuggetrieben im Abdichtbereich. Dissertation, Institut für Maschinenelemente, Bericht Nr. 89, ISBN 3-921920-89-2,Universität Stuttgart, 2000 Klöpfer, M.: Dynamisch beanspruchte Dichtverbindungen von Getriebegehäusen. Dissertation, Institut für Maschinenelemente, Bericht Nr. 72, ISBN 3-921920-72-8,Universität Stuttgart, 1996 Krieg, W.-E.: Untersuchungen an Gehäuseabdichtungen von hochbelasteten Getrieben. Dissertation, Institut für Maschinenelemente, Bericht Nr. 49, ISBN 3-921920-49-3,Universität Stuttgart, 1993
ATZ worldwide 10/2003 Volume 105