Polymer rectangular seal design for the inline CVT

Transcription

1 Polymer rectangular seal design for the inline CVT TU/e Masters Internship Report 2nd July 2009 Report no: DCT Authors P.F. van Oorschot ( ) J. Pustjens ( ) Supervisors Prof. ir. A.A. Frank M. Sc. Ph.D. (UC Davis) Prof. Dr. Ir. M. Steinbuch (TU/e) Eindhoven University of Technology Department of Mechanical Engineering Division Dynamical Systems Design University of California, Davis Department of Mechanical Engineering Hybrid Electrical Vehicle Center

2 2

3 Introduction At the department of mechanical engineering at the UC Davis extensive research is done on the Plug-in Hybrid Electric Vehicle (PHEV) under direction by Professor Andrew A. Frank. General research is done on a new transmission, the inline CVT, to improve the benefits of a hybrid powertrain. The design is motivated by the need for an inline CVT to directly replace a manual transmission for trucks or high horsepower passenger cars with longitudinal engine orientation. The CVT is designed in such a way that it can physically replace the conventional Manual and Automatic Transmissions. The inline CVT uses a GCI Chain. This chain concept uses the basic pulley disks CVT concept with hydraulic pressure actuated pulleys. In these kind of automatic transmissions very high pressures up to 80 bars (1200psi) and rotational speed (8,000 rpm) occur. In the current design the rotating connections are sealed by use of a clearance seal. A long tube is inserted in the main shaft with a tight tolerance; this provides so much hydraulic resistance that leakage is low. The design philosophy is that these seals have very little drag and will last a lifetime since there is no wear. However, in practice these seals do not qualify since they tend to fail easily by wear and subsequently by having an unacceptable high leakage rate. This paper focuses on investigating a better sealing solution. After research the decision was made to compare the current clearance seal to polymeric rectangular seals. Rectangular seal rings are located between the shoulder of a groove in the rotating shaft and the surrounding housing. The seal design is quite similar to piston rings but instead of a reciprocating motion the seal rotates. In the past grey cast iron alloys were used as sealing ring material. Since the last decade however they are slowly replaced by high temperature resistance plastics such as polyetheretherketon (PEEK), polyamidimid (PAI), polyimid (PI), and thermoplastic polyimide (TPI). A test set up is build to test the different sealing methods. Eventually, a new rectangular seal design is proposed for the inline CVT which will perform better in terms of reliability and leakage rates. 3

4 Table of contents 1. The inline CVT Demand in automotive industry for the inline CVT Engineering design for the ICVT Sealing methods Operating conditions Types of seals Choice of seal Polymer performance in practice Sealing Design Rectangular Seal Design Test setup Experimental results Test conditions Testing the clearance seal Clearance seal leakage Testing the rectangular seal Pulley leakage Final theoretical design Choice of shape and gap geometry Design parameters of the seal Plug and bushing design O-ring design Conclusion References A O-ring design parameters

5 5

6 Chapter 1 The inline CVT 1.1 Demand in automotive industry for the inline CVT Worldwide a movement in the automotive industry can be recognized toward hybrid powertrains with electric motors. Besides that, there is a need to directly replace a manual transmission for a CVT for trucks and high horsepower passenger cars with rear wheel drive and longitudinal engine orientation. Based on those two developments, a transmission system is needed that provides the capability to be used in either conventional or hybrid electric vehicles (HEV) without large mechanical modification. This means that this technology must be equivalent to a geared transmission in among others: power, torque, efficiency, durability, size, and yet be completely automatic and finally at a lower cost. Because there is currently no CVT available that can be used as a direct replacement for transmissions in rear wheel driven SUVs or small trucks, a new concept for a CVT is developed by UC-Davis (UCD) and Gear Chain Industries (GCI). The new transmission is a modification of a conventional pulley-type CVT design. The inline CVT (ICVT) consists of two sets of driving pulleys equipped with GCI Chains in place of the Van Doorne Transmissie push belt. When put in series, these pulleys and chains allow for greater ratiospread in a compact layout. Furthermore a Servo- Hydraulic control system instead of the engine-driven pump is used. The efficiency is increased to 95%, near that of manual transmissions. The application of a servo-hydraulic control system is advantageous for hybrid powertrains in which all- electric modes can be used. The size of the CVT is considerably smaller than an automatic transmission of equal power comparable to a manual transmission of equal power. 6

7 Figure 1.1 Sectional view of the ICVT 1.2 Engineering design for the ICVT The inline CVT uses the conventional pulley disk CVT principle. Roughly, there are three remarkable determining features that make the inline different in its design. Firstly, the ICVT is based on two conventional pulley disc CVT s in series. Secondly, a GCI chain is used in place of a pushbelt. Thirdly, a servo hydraulic control system is used instead of an engine driven oil pump. The in- and output shaft of the ICVT are placed inline, to make transfer with a conventional transmission for a rear wheel driven car easy. The primary drive train, which can be a combustion engine or electric motor, connects to the input pulley (primary pulley) of the first CVT. The output pulley of the first CVT, is connected to the input pulley of the second CVT. Finally, the output pulley (secondary) of the second CVT connects to the secondary drive train and the wheels. The only pulleys that are actuated are the primary and secondary. The other two pulleys are idler pulleys which are not actively actuated and share the same idler shaft. The pressure in the pulleys control the ratio and clamping force, thereby determining the maximum transmittable torque. An advantage of two CVTs in series is that the two stages have individually a much smaller ratio span than just one CVT, because the ratios of both can be multiplied. This is profitable for efficiency, since both CVTs operate closer around ratio one. The length in 7

8 longitudinal direction is slightly increased compared to a conventional system. Also will the ICVT be slightly larger in vertical direction. The used chain is developed by Gear Chain Industrial which features the needed high torque capacity combined with high efficiency. The chain consists of pins which are tapered to the pulley surface. To allow a smooth low friction transition from rectilinear motion to rotation, the pins have an involute surface. Figure 1.2 Schematic view of hydraulic system A conventional CVT uses a single engine driven pump to build up the required pressure. Then pressure is distributed by electronically controlled valves for both pulleys. The ICVT uses a servo hydraulic actuated system. Two bidirectional external gear pumps control the clamping pressure and ratio. Each pump is actuated by a brushless DC motor. The primary pump builds up pressure to clamp the chains while the second is used to change the driving ratio (fig. 1.2). An advantage of this system is that the transmission can be controlled independently by a microprocessor. In theory, energy losses are smaller than with a conventional engine driven pump because hydraulic pressure is only developed when needed. The oil pressure for both primary and secondary pulley is delivered via a central mounting block. On both sides of this block are two fixed narrow tubes which are placed in the pulley shafts with a very tight tolerance, which functions as a clearance seal. The design philosophy was that hydraulic resistance between tube and shaft will give low leakage and that these seals have very little drag and last a lifetime since there is no wear. Furthermore, the leakage between tube and shaft and at the end of the tube will operate as lubrication for the bearings. However, the clearance seal is not optimal at this moment. In practice leakage is unacceptable high and lifetime is not very high. Presumably this is amongst other things caused by small misalignment of tube and shaft which causes damage and fatigue on the tubes, which first will lead to high leakage and finally to failure of the tube. The clearance seal tube is a grinded tube with an oil bleeding hole in the middle and a plugged or welded end. The clearance seal assembly consists of five parts (Fig 1.3). The tube is concentric mounted with an o-ring to give it a little freedom in movement. A small pin obstructs the axial movement. 8

9 Figure 1.3 exploded view of clearance seal assembly Figure 1.4 Assembly overview of one pulley with the clearance seal tube 9

10 Chapter 2 Sealing methods Before a new seal can be designed and tested there has to be analyzed why the current clearance seal is leaking, what kind of shaft seals are available and which seal would be the most suitable for this particular application. To determine what will be a suitable seal, the current situation has to be analyzed, the problems herein and the operating conditions in which the seal has to perform. 2.1 Operating conditions Current situation: The pulley-discs in the inline CVT are actuated by hydraulic pistons. The pulley-discs and actuating pistons are connected to a rotating shaft. To operate the pistons a hydraulic fluid connection is needed. At this moment the hydraulic fluid is delivered into the rotating shaft using a clearance seal. A small tube which is fixed to the casing slides into the shaft. In the middle of the tube is a hole where the fluid can enter the piston cavity. The clearance between the tube and the shaft is small. Because of the hydraulic resistance fluid can only flow slowly out past the tube. The clearance seal is also used to lubricate the bearings on both ends of the shaft. the clearance seal tube is closed at the end, the leakage to this side can leave the shaft via a small hole to lubricate the ball bearing. See fig 1.4 for an overview of the construction. 10

11 Problems: The seal is not optimal at this moment. There are several problems which have to be solved : - The clearance seal is not reliable. The tube has broken off in the past. A reason for this can be contaminations in the oil which get between the clearance seal. The friction between the shaft and the tube will get too high and the tube breaks. Another reason why this could happen is the wobbling of the shaft because the bearings have some runout, causing fatigue in tube and break. - The leakage of the clearance seal is too high, it would be preferable to decrease the leakage of the seal. Some leakage is ok for the lubrication of the bearings. The leakage also increases over time by the wear of the tube. Studying old clearance seal tubes showed that they wear a considerable amount. Meaning there is not always a full lubricating layer of oil between the tube and the shaft, as was expected theoretically. More on the causes of failure and leakage can be found in chapter 5. Operating conditions: Before different sealing methods can be investigated, the operating conditions need to be known which the seal has to comply with. The most important parameters are the working pressure and the surface speed at the seal. Max rpm shaft: Max pressure: Temperature: Working fluid: 5000 rpm 80 bar / 1200 psi 90 Celsius / 200 Fahrenheit ATF Surface speed at outer diameter: Sealing 6mm tube: 1.57 m/s ft/s Sealing 8mm tube: 2.09 m/s ft/s Sealing 10mm tube: 2.62 m/s ft/s 11

12 2.2 Types of seals Different types of seals are investigated. Because of the relatively high pressure/speed combination a lot of seals are incapable for this specific situation. First all types of seals which are not capable will be listed. Seals which are better suited will be discussed below. Types of seals not suitable for this situation: Lip seals - Maximum pressure capability too low, sometimes up to 10 bar psi - Maximum surface speed up to 100 m/s ft/s Compression packing seal - maximum pressure capability of 25 bar psi - maximum surface speed up to 20 m/s ft/s O-rings / non-energized elastomer seals - Not well suited for dynamic sealing, maximum speed 0.1 m/s - Wears quickly in dynamic situations. Magnetic fluid seals - Maximum pressure per stage only 0.2 bar - 3 psi - Very expensive 12

13 Types of seals potentially suitable for this situation: Energized plastic/ptfe seals - Maximum pressure capability up to 500 bar psi - Speed limit of 0.5 to 2.0 m/s 390 ft/s Advantages - cheap - compact - low leakage Disadvantages - will wear, needs replacement after a certain amount of hours. - hard to find a suitable seal for this application, working conditions on the edge of capabilities of this seal. Suitable for high pressure and low speed or high speed and low pressure. In this situation with high pressure and moderate speed it is out of specifications. Possible product: Telleborg - Roto Glyd Ring Mechanical face seals - Very wide range of pressure and speed capabilities. Advantages - very low leakage - high reliability, low wear Disadvantages - relatively large - more expensive - more difficult installation possible product: John Crane - S_1648 Cast iron rectangular ring seal Capable of high speed and pressure combinations Advantages - Very simple and cheap - Reliable and little wear - Used in older automatic transmissions 13

14 Disadvantages - Relatively high leakage - Wears, fails quickly if lubrication is lost Polymer rectangular ring seal Viscoseals Same design as Cast iron seals, used in newer automatic transmissions Advantages over cast iron - Lower leakage - Better wear characteristics - More robust, can handle a short time of abuse. Design adjustable to speed and pressure Advantages - virtually no wear - good sealing properties - simple and relatively compact Disadvantages - high leakage at low rpm - tight tolerance in design - requires more space. 2.3 Choice of seal The best option for the shaft seal seems to be a rectangular polymer ring seal. This seal has many advantages over the other seals in this situation. First of all, this is the most economic option for sealing, second the implementation is relatively easy in comparison with other seals and takes up very little space. This makes it possible to implement and test the seal in the current CVT casing. If the seal shows good results, it can replace the clearance seal with just little adaptations. The disadvantage of the relatively high leakage compared to for example a mechanical face seal is no problem because a small amount of leakage is required for the lubrication of the bearing. 14

15 Working principles of a rectangular ring seal Rectangular ring seals are often used in rotating connections in for example automatic transmissions. The seal is very similar to a piston ring, except that instead of a translating motion, there is a rotating motion. The rectangular seal is positioned in a groove in the rotating shaft. Because of the fluid pressure, the seal gets pressed against the circumference of the housing and the shoulder of the groove. There the seal acts like a face seal. See figure 2.1 for a graphical presentation. The seal can move relatively to the housing as well as to the shaft, in practice the seal often shows little relative movement to the housing and most relative movement to the shaft. See reference 2. Figure 2.1 rectangular seal on shaft Choosing a suitable polymer for the rectangular seal Several different polymers are used in mechanical seals, some of the most common materials are: PTFE(Teflon) Pure PTFE is used in applications with relatively low P*V loads. The big advantages of Teflon are it s very low coefficient of friction and very good chemical resistance. But it s mechanical strength is not very high. Often PTFE is added to other polymers in a low concentration to increase the characteristics. Other common used additives to high performance polymers include graphite. Recently new high performance plastics have been developed, capable of withstanding higher temperatures and loads. High temperature (HT) plastics: Polyamidimid (PAI) This polymer can be injection molded, after this it has to be cured for a number of days. Different grades of polyamidimid are suitable for use as a seal ring. Torlon from Amoco is a series of polyamidimids suitable for this kind of application. 15

16 Polyetheretherketon (PEEK) PEEK is a semicrystalline thermoplastic. An example are Victrex PEEK polymers. Polyimid (PI) This is a thermoset polymer, which cannot be melt processed. They are manufactured using compression forming. A well known polyimid used for transmission seals is Vespel SP21 from dupont. SP21 has a graphite content of 15%. Another high performance polyimid suitable for this application is Meldin 7000 from saint-gobain. Polymer properties In table 2.1 a few important properties of different possibly suitable polymers are noted. Material Base material Filling Operating temperature Chemical resistance Tensile strength Thermal conductivity Coefficient of friction Teflon PTFE - 260' C ~ 500' F ,9 Mpa 0,96 [W/M*C'] 0,04 Meldin 7021 Polyimid 15% Graphite 300' C ~ 572' F Mpa 0,71 [W/M*C'] 0,12 Vespel SP21 Polyimid 15% Graphite 300' C ~ 572' F ++ 65,5 Mpa 0,87 [W/M*C'] 0,12 Victrex 30% 450FC30 PEEK PTFE,Graphite,Carbon 140' C ~ 275' F + 55 Mpa 0,87 [W/M*C'] 0,25 Torlon 4275 Polyamidimid 24% Graphite, PTFE 275' C ~ 525' F Mpa 0,60 [W/M*C'] 0,14 Table 2.1 Properties of polymers The properties of the materials can give some idea about the performance of the seal. Chemical resistance All the materials listed here have excellent chemical resistance, although in concentrated acids PEEK is a little less resistant. The compatibility with Automatic transmission oil is good for all materials. Tensile strength The tensile strength of PTFE is much lower than the other polymers, making it less suitable for higher pressures required in the transmission. The tensile strengths of PI and PEEK are about twice as good as pure PTFE. PAI has the highest tensile strength, which makes it probably the most suitable for high pressure sealing. Operating temperature All the materials are capable of running at the required operating temperature, the limit for PEEK is closer than for the other polymers. At high loads it could be possible that the local temperature near the seal reaches 140 C. Which would ruin the seal. Friction PTFE has the lowest coefficient of friction, but lacks tensile strength. The coefficient of friction is similar for PAI and PI at 0,12. PEEK has a somewhat higher coefficient of friction of 0,25. 16

17 Considering the different parameters, the PAI and PI polymer types look the most promising to use for the seal. 2.4 Polymer performance in practice In the SAE paper Wear Performance of Ultra-Performance Engineering Polymers at High PVs by G.S. Underwood (ref 1). Different polymer seals are tested under high pressure and speed to determine which polymers are the most robust and have the best wear characteristics. The following polymers are investigated in this paper: - Torlon 4275 (PAI) - Torlon PXM (PAI) - Vespel SP-21(PI) - Victrex PEEK 450FC30 (PEEK) The results from the different experiments in this paper show that the PEEK polymers have the lowest performance overall. The PEEK seal starts to wear badly already at relatively low PV values compared to the other polymers. At low speed (0,25m/s / 50ft/min) but high pressure (up to 135 bar) the PAI polymers can handle the most. This could be expected from the parameters as PAI has the highest tensile strength. The PI polymer fails quicker at higher pressure, but can still withstand more than 100 bar, which is sufficient for the transmission seal. At higher speeds the wear characteristics of the PI polymer are slightly better than for the PAI polymers. At the highest speed and load tested (4m/s / 800ft/min) only the polymers Torlon PXM and Vespel SP-21 stand the test, hereby still showing good wear characteristics. Leakage The leakage rate of a rectangular seal depends mostly on the type of cut and not much on the type of polymer, as follows by a study of M. Gronitzki and G.W.G. Poll (ref 2). The amount of leakage can be chosen by the type of cut, butt-cut rings will show the highest amount of leakage but are also the cheapest to manufacture. Taking into consideration a small amount of leakage is required, probably a butt-cut ring will do the job. Choice of polymer Looking at the results the polymer Vespel SP21 seems most suited for this situation as the polymer has enough strength to withstand the pressures and shows the best wear characteristics at higher speeds. Also the polymer is widely used and can be manually machined to a ring. The polymer Torlon PXM01017 also shows good result, with a wear rate as low as SP-21 and a better performance at extreme pressures. 17

18 Chapter 3 Sealing Design As a first stage a preliminary seal design has been made that requires no structural changes to the transmission. This way the functioning of the seal and all potential side effects can be studied. In a second stage, the new sealing design can be fully integrated in the inline design. However, that is beyond the scope of this investigation and will not be dealt with in this paper. As mentioned in chapter 2 a polymeric rectangular ring seems to be the most viable sealing solution. 3.1 Rectangular Seal Design A single rectangular seal is chosen which will be mounted using the same thread as the clearance seal. Although polymer rectangular seals are found more and more in automatic transmissions last years, in industry there is still no way to order them in standardized sizes from a catalog. This lasts two options: produce seal rings in home, or order custom made seal rings from a seal ring manufacturer. The first option is cheaper, but requires high precision manufacturing. Furthermore, for different plastics injection molding is preferred above machining on a lathe. The second option is more expensive, but is recommended when seals are required on a large scale. During this project, there has been correspondence with LS Polymer Technology, one of companies that do research on polymeric face seals. Other known companies are St. Gobain (producing Rulon and Meldin) and DuPont (Vespel). More information about these contacts can be found in appendix 1. To be able to do actual measurements during the project period without the custom made seal rings, Teflon-based rings were found in a local transmission shop (AAMCO) in Davis. Wear characteristics will not be the same as from the T.P.I., P.A.I. or P.E.E.K. rings. However, these characteristics are extensively tested and described by other research papers (ref 1,2). The AAMCO seals are mainly suitable for leakage rate measurements. 18

19 Dimensions of the AAMCO seals had to be determined because they were not specified. ID OD Radial Thickness Axial thickness = mm = mm = 1.25 mm = 1.30 mm Figure 3.1 the AAMCO seal For the AAMCO seal a threaded steel plug has been designed which is mounted on the central mounting block (fig 3.2). On the end a groove is cut where the seal fits in, The groove is designed according to SAE paper J2310 (ref 4). The plug has a hexagon hole to easily mount it and axially a hole is drilled for the oil. In the rotating pulley shaft an aluminum bushing is mounted with a press fit Aluminum is chosen for its soft properties and it s bearing capabilities in combination with steel. In case the free play of the bearings causes the plug and bushing to interact. First a steel bushing was used, but this cold welded to the plug. In the current pulley shaft a small hole on the end provides lubrication to the ball bearing. In the new design this hole will not be used and plugged with a bolt, so the bearing will not be lubricated via this way. 19

20 Figure 3.2 The aluminum mounting block and above the plug and bushing 20

21 Chapter 4 Test setup For the experimental investigation of the clearance seal and the rectangular seal rings a test rig was designed and built at the HEV lab at UC Davis. The setup uses original parts from the inline CVT. This permits tests and measurements under conditions which closely reflect real operation. The setup allows measuring parameters like the leakage rate, rotational speed, pressure and volumetric efficiency. Leakage is determined with a balance and 2 oil sumps. The weight of the source sump is measured before and after a test. This way the amount which is pumped into the transmission is measured. This gives a more precise result instead of measuring the amount which leaks from the transmission. See figure 4.1 for a overview of the test rig. Figure 4.1 Schematic view of experimental setup The different parts of the setup will be briefly discussed. First of all a steel frame with two rails was designed and built on which the transmission, pump and motor are mounted. The transmission is fixed to the main frame. The motor driving the transmission is mounted on a moving support so different motors and couplings can be used if necessary. To account for the small misalignment between the main shaft and the motor a jaw coupling is used. 21

22 For the experiments on the clearance seal and the rectangular seal we have to see where the leakage originates so we can analyze the problem. Half of the casing was sawed off to get a good look at the rotating parts. In tangential direction there are two holes in the shaft. To test the leakage of the different components the system will be tested with the secondary pulley installed and removed. To test the system when the secondary pulley is removed, the supply hole in the shaft needs to be plugged. This is done with a plug which is press fit into the shaft, the plug has thread inside so it can be removed(fig 4.2). At the end of the clearance seal, a small hole in the shaft to supply oil to the ball bearing also has to be closed. This is done by a threaded plug. In the central mounting block thread was cut to mount the clearance seal assembly, the rectangular seal test assembly uses the same thread. Figure 4.2 the finished setup, plug, connection flange Electric motors To drive the transmission shaft and the pump 2 three-phase brushless motors are used (Emoteq BH03402-T01-H). These motors have a maximum continuous output power of 670 W which is elaborate. The motors were available from a former project and the same motors are also used in the actual inline CVT. A problem is that these motors tend to degrade. To drive the motors a power supply and brushless amplifiers are needed. For this project a power supply with a maximum of 150V/20A is used. In the initial setup amplifiers for 80V were used. It turned out that they were too weak as the maximum rpm is proportional to the supplied voltage. They were replaced by amplifiers that can go up to 400V so that the 150V of the power supply can be fully used. To drive the motors the commutation of the different phases has to be correct, this was determined with the schedule in figure 4.3. The amplifiers are mounted in a separate frame where both amplifiers can be independently activated and speed is controlled by potentiometers in a analog open-loop setup. Both motors are mounted with an encoder to determine the rotational speed. 22

23 Figure 4.3 Commutation sequence brushless motors Pumps The same pump is used as in the actual transmission, a external gear pump from Haldex Barnes(type nr G1204C3A300N00). This pump was chosen based on findings from previous research at the HEV center research and the initial requirements were positive displacement, bi-rotational and high efficiency. Gear pumps work to pump the fluid by the rotation of two symmetric gears, one of which is on a drive-shaft. The Haldex pump has a high-pressure shaft seal, displacement of 1.07 cm3/ rev and can accommodate up to 20 MPa (3000 psi) continuously. On problem of the pumps is that volumetric efficiency drops drastically when the oil heats up to 80 C and the viscosity gets lower. Also the pumps tend to grow internal leakage over time by wear of the gears. Figure 4.4 the motor-pump combination and amplifier setup 23

24 Chapter 5 Experimental results To determine why and how the clearance seal is leaking first different clearance seals will be tested. By closing of leakage paths step by step, the amount of leakage at different points is determined. After the leakage is known, a reason for the bad performance will be investigated. At last the rectangular seal will be tested to compare the performance with the clearance seal. Furthermore the relation between rpm and leakage rate is measured. 5.1 Test conditions Leakage rate is determined by measuring the mass of the source sump before and after each run. Every run consists of a test at a specified pressure and rpm for 400 seconds. All tests have been run at room temperature of 20 C. Every single test has been repeated for at least 6 runs and the result is averaged. Two different shafts and 3 different clearance tubes have been tested; these have been numbered and can be recognized on the following points. Shaft 1 Shaft 2 Tube 1 Tube 2 Tube 3 Has O-ring groove added Used, plugged at the end Used, shows more wear, plugged at end New, welded at end There are 3 openings in the hole for the clearance tube, which means 3 paths were leakage could occur. These paths are noted as Front Rear Pulley The point were the clearance seal enters the shaft The leakage from the other side of the clearance seal, end of shaft The leakage, leaking from any side of the secondary pulley 24

25 5.2 Testing the clearance seal First different leakage paths, proportions and relations will be tested. After that the source and causes of leakage will be discussed in the next paragraph. Most of the test have been performed at 13,78 bar(200psi), because the first pump used seemed to have a very high internal leakage and could not produce more then 20.7 bar / 300 psi. After changing the pump, higher pressures could be reached. Determining the different leakage paths and proportions Hereby using only one shaft and one seal, so these parameters are constant Shaft Tube Leakage paths RPM Pressure [psi/bar] Avg. flow [ml/min] Nr. 1 Nr. 1 Front / 42,92 Rear Pulley 13,78 Nr. 1 Nr. 1 Front / 40,02 Rear 13,78 Nr. 1 Nr. 1 Front / 2,67 13,78 From these results follow the absolute leakage and proportional leakage per path: Front side 2,67 [ml/min] 6,2 % Rear side 37,35 [ml/min] 87 % Pulley 2,90 [ml/min] 6,8% It is clear that the leakage is very high and the biggest portion is coming out the rear end of the clearance seal, through the small hole for bearing lubrication. 25

26 Determining the difference in leakage between different seal tubes Using the same shaft, leakage paths front and rear only Shaft Tube Leakage paths RPM Pressure [psi/bar] Avg. flow [ml/min] Nr. 1 Nr. 1 Front / 40,02 Rear 13,78 Nr. 1 Nr. 2 Front Rear Nr. 1 Nr. 3 Front Rear / 13, / 13,78 42,42 13,7 Tubes 1 and 2 seal approximately equal, with tube 2 showing a slightly higher leakage. Tube 3 shows a significant lower leakage. Determining the difference in leakage between different shafts Using the same tube, leakage paths front and rear only Shaft Tube Leakage paths RPM Pressure [psi/bar] Avg. flow [ml/min] Nr. 1 Nr. 3 Front / 13,7 Rear 13,78 Nr. 2 Nr. 3 Front / 16,23 Rear 13,78 Shaft 2 shows a higher leakage rate than Shaft 1 Determining the relation between pressure and leakage rate Shaft Tube Leakage paths RPM Pressure [psi/bar] Nr. 1 Nr. 3 Front / 13,78 Nr. 1 Nr. 3 Front / 27,56 Nr. 1 Nr. 3 Front / 41,34 Nr. 1 Nr. 3 Front / 55,12 Nr. 1 Nr. 3 Front / 68,9 Avg. flow [ml/min] 0,81 1,68 2,38 3,07 3,83 26

27 The result are plotted in fig 5.1 showing a clear linear relation 4 Leakage rate proportional to pressure Leakage [ml/min] Figuur Pressure [Psi] Leakage rate as a function of pressure Determining the relation between rotational speed and leakage Shaft Tube Leakage paths RPM Pressure [psi/bar] Nr. 1 Nr. 3 Front / 41,34 Nr. 1 Nr. 3 Front / 41,34 Nr. 1 Nr. 3 Front / 41,34 Nr. 1 Nr. 3 Front / 41,34 Avg. flow [ml/min] 2,38 2,24 2,46 2,34 The difference of flow rate at different rpm is so small it can be assumed that rpm has no significant influence on the leakage rate. For the rectangular seal design is also assumed that rotational speed has no influence on leakage rate. 27

28 5.3 Clearance seal leakage Clearance tube diameter Figure 5.2 clearance seal and shaft assembly, showing the measurement points. To determine why the leakage of the clearance seal is so high, especially at the rear, and differs a lot between the old and new seals. The clearance tubes diameters are measured at different points. Only the part of the clearance tube which actually seals is measured. The measurements are done at 5 different points as indicated in figure 5.2. The points are at equal distance of 0,84 inch/21,3 mm. In the following table the results of the measurements is shown. All the points are measured 5 times and the average value is shown. Clearance tube Units Point Tube 1 in Used, plugged mm Tube 2 in Used, plugged, worn mm Tube 3 in New, welded mm Table 5.1, Diameter measurements at different points on the clearance tubes Tube diameter measurements Diameter [mm] 6.3 Tube 1 Tube 2 Tube Measurement point [] Figure 5.2 Graphical representation of diameter measurement 28

29 As can be seen, the diameters of the different seals differ much. The new tube 3 has a much larger diameter than the older seals, with a difference of between 0.02 and 0.06 mm. This is a very large difference on a diameter of only 6 mm and explains why the leakage with the new seal is much lower. We can also see that the older seal tubes have different diameters at the different points. On tube 2 this shows most clearly, the tube has worn irregularly along the length. The difference in diameter along the length is very high at tube 2, with a difference of mm. While the tube 1 and 3 have only about mm difference in diameter along the length. These results explain the difference in leakage between the new welded tube and the older tubes. But they do not explain the difference in leakage between the front and the rear of the shaft. As the clearance seal path to both the front and rear end are equal length and the new tube has a uniform diameter. A explanation to this would be that the diameter of the hole in the shaft is not uniform. Shaft hole diameter Unfortunately no tools where available to measure the inside diameter of the shaft precisely at different points. To be able to determine if the shaft has a uniform hole diameter a tool has been made. A small cylinder is turned to the diameter of the hole, so it can slide in the hole. A rod is attached to this cylinder so it can be positioned throughout the hole. Now the hole in the shaft is filled with water and by sliding the cylinder along the hole we can measure the relative diameter by the force which is needed to move the cylinder and squeeze the water through the fit between the hole and the cylinder. During the experiment it became clear that the hole in the shaft does not have a uniform diameter, the diameter of the hole increases along the shaft. This fact explains the high leakage towards the back of the shaft. A explanation for the tapered hole would be wandering of the drill, assuming the hole is drilled. If the hole would be finished with a single cutter, it should be possible to make a uniform hole throughout the shaft with a tight tolerance. 29

30 5.4 Testing the rectangular seal To test the rectangular seal, the rear hole of the shaft has to be closed because there is no flow restriction to the rear of the shaft anymore. The hole supplying the pulley is also closed to focus only on the performance of the rectangular seal. At first the plug as well as the bushing was made out of steel, eventually it showed that the plug and bushing were cold welding occasionally because of the runout of the bearing. This means the seal surface was damaged during these test. Leakage of the rectangular seal with steel-steel bushing at different pressure Shaft Seal Leakage paths RPM Pressure [psi/bar] Nr. 1 Steel-steel Front / rectangular 13,78 Nr. 1 Steel-steel Front / rectangular 68,9 Avg. flow [ml/min] 0,61 1,54 Comparison of leakage rate between rectangular and clearance seal at 13,78 bar / 200 psi Rectangular - tube 1 22,5% the amount of leakage of tube 1 Rectangular tube 3 75% the amount of leakage of tube 3 Even though the seal surface has been damaged, the seal performs much better then the clearance seal. Next there has been built a new steel plug and aluminum bushing for the seal Leakage of the rectangular seal with steel-aluminum bushing at different pressure Shaft Seal Leakage paths RPM Pressure [psi/bar] Nr. 1 Nr. 3 Front / 13,78 Nr. 1 Nr. 3 Front / 27,56 Nr. 1 Nr. 3 Front / 41,34 Nr. 1 Nr. 3 Front / 55,12 Nr. 1 Nr. 3 Front / 68,9 Avg. flow [ml/min] 0,17 0,12 0,22 0,19 0,28 30

31 Because the leakage rate is near zero, no significant measurements can be taken because the resolution of the scale is only ml/min per run. If we take the average of all measurements, we get a leakage rate of 0,196 ml/min This is a very small leakage compared to the clearance seals Rectangular alu tube 1 7,3% of the leakage of tube 1 Rectangular alu tube 3 24,2% of the leakage of tube 3 These are promising results and show that a rectangular seal has the capability to seal sufficient in this application. 5.5 Pulley leakage The leakage path for the pulley is the clearance between the shaft and the moving secondary pulley, because no seal is used here. During the test it can be clearly seen oil is leaking at this point. To stop this leakage an O-ring has been designed, implemented and tested. The design of this O-ring seal is discussed in chapter 6. After implementation no leakage occurred at this point anymore, the leakage at 40 bar /600 psi was about ml/min average using the rectangular seal, this is within significance limits with the previous results when the shaft was plugged. Also no visible leakage occurred at the shaft and pulley connection. 31

32 Chapter 6 Final theoretical design After the successful tests with the AAMCO rectangular face seal, a final design has been developed. This design is based on theoretical study and it is highly recommended to discuss the design with ring manufacturer before production. For the seal design a few design objectives have been formulated: The seal design must be plain and cost effective. A little leakage is no problem and even desirable, since it provides lubrication to the bearings. So complex designs which eliminate leakage are not necessary. The selected ring material is Vespel SP21 from the DuPont Company. Vespel was determined in chapter 2 as most suitable polymer for this application. The outer diameter of the seal should be kept low for two reasons. One is to limit the axial force on the shaft induced by the oil pressure. Another reason is to reduce surface speeds on the mating surfaces of the ring and shaft. 6.1 Choice of shape and gap geometry The simplest design is to choose a rectangular cross section. More complex and more expensive designs exist with chamfers on edges to reduce the contact area with the groove wall. Sometimes oil grooves are designed to carry lubricant to the face of the ring to improve wear characteristics. However, for this application the simple rectangular cross section complies. Also the gap geometry has to be chosen. There are three different options: a butt joint, a scarf joint and a stepped joint (figure 6.1). The butt joint is the simplest solution: a radial cut. The ring is direct formable and cost effective. The gap will close over temperature and can take compression without permanent deformation. The second option is the scarf joint. Usually a cut of 20 degrees is made. These rings cannot be produced in one step; the cut has to be machined after the full ring has been pressed. The leak rate for scarf joint seal rings is somewhat higher than that for compressed butt joint rings. However, scarf joint rings are not as sensitive to the minimum pressure of the transmission. The third option is the stepped joint. Just like the butt joint they can be direct formed in one step depending on geometry. In principle it behaves like a butt joint with a slightly reduced clearance. For the inline CVT the butt joint is chosen, since it is the simplest, robust and cost effective solution, and a little leakage desirable. One has to take into account, that chancing gap design also affects the shape of the groove. The groove in this final design is designed for a butt joint. 32

33 Figure 6.1 Different gap geometries 6.2 Design parameters of the seal Vespel has a quite large elastic deformation area which makes it possible to design the butt joint closed from 20 C up to operation temperature 80 C. The thermal expansion will be taken by the elastic deformation rate of the polymer without causing plastic deformation. Very low leak rates can be achieved this way over the whole operating conditions. Under thermal compression the ring can become warped, but under the influence of the oil pressure the ring will keep its shape and will be pressed against the surface. For this seal design the geometric parameters as shown in table 6.1 apply. 33

34 Parameter Abbreviation Explanation Bore B Inner diameter of the housing within which the ring is constrained to move Nominal ND Outer diameter seal ring (Equals the bore) diameter Radial thickness T Difference between inner and outer diameter of the ring (wall thickness) Axial thickness W Thickness of seal ring in axial direction, ring height Gap G Distance between open ends of seal ring when constrained at gauge diameter. Nomimal Gap NG Distance between open ends of the seal ring when constrained at the nominal diameter. Groove GRD Diameter of the groove that faces the inner surface of the seal ring root Groove GWD Axial width of the groove width Shaft diameter SD Diameter of shaft (the shaft which has the groove) Table 6.1 parameters for seal design One design objective is to keep the seal outer diameter low. One of the reasons is because the oil is fed in axial direction and sealing is only on one side, an axial reaction force will occur on the pulley shaft. This force has to be taken in consideration in the design and will increase square with diameter according to the following relation: F F p p = = p A p π r 2 o So it is preferable to keep the outer diameter of the seal as small as possible. With an outer seal diameter of for example 7 mm and an oil pressure of 80 bars, the pressure force is already 1.2 kn. The force will be guided through the shaft and act on the deep groove 34

35 ball bearing. This bearing (deep groove type 6310) is able to withstand axial force up to 19 kn. Besides that, a small seal diameter means lower surface speeds on the mating surfaces and consequently lower temperature rise. This means lower wear of the seal and less risk on oil carbonization. The minimum of the groove root diameter (GRD) is constrained by the oil feeding hole through the shaft. For this hole diameter 4 mm is chosen as minimum to prevent hydraulic resistance, so the GRD can get no lower than approximately 5 mm. The other seal parameters are deducted from the minimal groove diameter. The run out between the shaft and casing, approximately mm, has to be taken into account especially for small diameter seals. The used equations and design guidelines are used from the Vespel Design Handbook (ref 5). Now the most basic design parameters can be determined. The GRD was already determined on 5 mm: GRD = 5mm The nominal gap is closed when the seal is at room temperature, so: NG = 0mm A common size for axial thickness W is 1 mm. The best ratio between radial and axial thickness T is 0.9. W = 1mm T = 0.9mm At maximum temperature a 0.1mm lateral clearance for the ring in the groove is allowed. GWD = W GWD = 1.1mm Bore and root should have at least 0.05 mm radial clearance if this bearing is used. Now the bore and shaft diameter can be determined: B = GRD + 2(0.1+ T ) B = 5 + 2( ) = 7mm SD = B 2*0.025 SD = = 6.95mm The nominal diameter ND is chosen equally to the bore: ND = 7mm 35

36 6.3 Plug and bushing design The groove for the seal ring can be integrated in the design of the central mounting block of the ICVT, but can also be modular. A threaded plug with a groove can be mounted on the mounting block (see picture). It is important to use Teflon tape or another sealing method on the thread to prevent leakage. The groove surface needs to be smooth and hard. The surface finish of the groove is of high importance, because of the abrasive action between mating materials. A smooth surface finish of 0,4 µm Ra is recommended and the mating material should be as hard as possible. However, if cast iron is used the finish of the mating surface is of less influence. Experiments have been done which demonstrated that cast iron mating surfaces with a finish of 2 µm Ra wear as well as those ground to 0.8 µm Ra (ref 5). For the outer bushing aluminum is chosen which is fixed in the pulley shaft with a press fit. On the side where the rectangular seal slides in, there must be a chamfer of 20 on the edge of the bushing. The aluminum acts as a bearing for the steel plug. In earlier experiments a steel bushing instead of aluminum was used. This caused problems when the plug and the bushing hit each other due to small misalignments. Figure 6.2 Plug design 6.4 O-ring design Early in the measurement process it was discovered that a large deal of the leakage occurs between shaft and sliding pulley disk. In the original design this leakage was accepted since it fulfilled two functions: bleeding the air out of the cylinder and lubricate the chain. However, the chain is also lubricated by a low pressure system at the underlying pulleys. Therefore is chosen to place an o-ring between pulley shaft and sliding disk. A problem that occurs is that air in the cylinder has no way to bleed out. A smart solution is to place an orifice in the cylinder in the outer wall on the side of the ball bearing. In this way an air bleeding hole is created and the ball bearing will be lubricated. For the O-ring a inner diameter of 50mm and a cross section of 3mm was chosen. The calculation necessary to determine the groove width and depth have been done online at 36

37 using their automatic calculator. All the data can be found in appendix 2. For the O-ring an FKM polymer (VITON) is chosen because this polymer has very good chemical resistance and excellent temperature resistance. 37

38 Chapter 7 Conclusion The clearance seal from the ICVT does not qualify requirements like lifespan and leakage rate. The theoretical design seems fine, but in practice leakage rate is too high and the clearance seal sometimes break off. Free play in the bearings causes fatigue in the seal what can lead to failure. Also was discovered that the hole in the pulley shaft was not as accurate as it should be. Deeper inside the hole the diameter increases, this probably happened during drilling of the hole and specifications were not totally clear to the manufacturer. When the different parts would be made as accurate as the designer intended, results probably would be better. However, a more attractive alternative solution for the clearance seal has been found in the polymer rectangular seal. In rotating connections in automatic transmissions the use of rectangular seals is common. The last two decades however, high temperature plastics like polyimide and polyamidimide are used to make rectangular seals for this kind of application. They have some interesting advantages like low leakrate, simplicity, robustness and cost efficiency. Leakrates of clearance and polymer rectangular seals were measured and compared. The leakage rate of the rectangular seal turned out to be 20 percent of the clearance seal. Lifetime has not been tested in this project, but is extensive tested in other papers. The results made obvious that polymer rectangular seals provide a viable solution for the sealing problem. A final design has been made proposing a butt joint seal made from Vespel SP21. However, this seal is not tested yet and it is highly recommended to discuss the design with the ring manufacturer and test it before taking into actual production. Also was discovered that a great amount of leakage (approximately 7% of total when using a clearance seal) occurs between sliding pulley disc and the pulley shaft. A little leakage is not bad, because air can bleed out. This amount of leakage however, is a waste for efficiency of the CVT. Therefore a design has been made for a Viton o-ring which is placed in a groove on the pulley shaft. Experiments turned out that leakage reduces almost to zero. To make it possible to bleed the air out an orifice can be placed which also lubricates the outer deep groove ball bearing. This has not been tested yet and needs further attention. When these changes would be applied in the ICVT it would be an improvement of overall efficiency since the servo hydraulic system consumes less power. 38

39 References 1. SAE PAPER Wear Performance of Ultra-Performance Engineering Polymers at High PVs, Geoffrey S. Underwood 2. Optimization of the tribological performance of rectangular seals in automotive transmissions, M. Gronitzki and G.W.G. Poll, Institute of Machine Elements, Engineering Design and Tribology, Hannover University, Germany 3. Horve, Leslie A., Shaft seals for dynamic applications / New York : M. Dekker, c Rectangular cross section polymeric sealing rings SAE J2310 Jan Vespel S Line Design Handbook, DuPont Vespel 6. Fluid Mechanics, fifth edition 2005, J.F. Douglas, Gasiorek, Swaffield and Jack, Pearson, Essex, England 7. O-rings, Trelleborg Sealing Solutions 8. The design of an inline GCI chain CVT for large vehicles,a.w. Brown (A.W. Brown Co.), J. van Rooij (GCI), and A.A. Frank (UCDavis) 9. NTN America,

40 A O-ring design parameters 40

41 41