Introduction Many applications require seals that retain lubricants or other fluids within machines. In applications with rotating shafts, radial lip seals are widely used [1], [2]. The radial lip seal, the surface of the shaft and the fluid, which has to be sealed, form a tribological system. Elastomeric radial lip seals are standardized in national and international standards [3], [4], [5]. They consist of a metal insert to which a sealing lip is attached, Figure 1. During assembly on the shaft, the sealing lip and a garter spring are elastically widened. The sealing lip is pressed against the shaft surface, so that a narrow contact area is formed. The contact pressure in the contact area significantly affects the function of the sealing system and is thus of huge interest. However, the spatial distribution of the contact pressure is not easy to measure. Therefore, the radial load F r is used as an alternative to quantity the contact in an integral manner. Mathematically, the radial load is defined according to DIN 3761-1 [6] as the integral of the load exerted by the sealing lip perpendicular to the shaft surface. Practically, the radial load is mostly measured using the so-called split-shaft method, Figure 2. This method is described in the German standard DIN 3761-9 [7] and involves two nearly half-cylindrical parts called “mandrel” representing the Aus Wissenschaft und Forschung 5 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 rachwissenschaft \ ltphilologie \ Sport munikationswissenche Sprachwissenment \ Altphilologie d Kommunikations- Historische Sprach- Management \ Altstik \ Bauwesen \ tschaft \ Tourismus gie \ Kulturwissenhichte \ Anglistik \ \ BWL \ Wirtschaft How to measure the radial load of radial lip seals Simon Feldmeth, Mario Stoll, Frank Bauer* Vorgetragen auf der Jahrestagung der Gesellschaft für Tribologie vom 28. bis 30. September 2020 Eingereicht: 15.9.2020 Nach Begutachtung angenommen: 25.6.2021 Die Radialkraft eines Radial-Wellendichtrings (RWDR) gibt an, wie stark dessen Dichtlippe an die Welle angedrückt wird. Die Radialkraft beeinflusst das Funktionsverhalten des RWDR maßgeblich. Die DIN 3761-9 beschreibt die Messung der Radialkraft nach dem Zwei-Backen-Verfahren, lässt dabei aber Interpretationsspielraum. Im Rahmen der Normüberarbeitung wurde am Institut für Maschinenelemente (IMA) der Universität Stuttgart in einer Parameterstudie untersucht, welchen Einfluss die Eigenschaften des Messgeräts, die Ausführung der Messbacken und die Gestaltung des Messvorgangs auf das Messergebnis haben. Basierend auf den Studienergebnissen werden Empfehlungen abgeleitet und in einem Best-Practice-Leitfaden zusammengefasst, der eine zielgerichtete und reproduzierbare Messung der Radialkraft ermöglichen soll. Schlüsselwörter Radial-Wellendichtring, Radialkraft, DIN 3761-9, Radialkraft-Messung, Radialkraft-Messgerät, Messverfahren, Empfehlung, Best-Practice-Leitfaden The radial load of a radial lip seal indicates how strongly the sealing lip is pressed on the shaft. The radial load significantly affects the function of the seal. The German standard DIN 3761-9 describes the measurement of the radial load according to the split-shaft method but leaves room for interpretation. During the revision of the standard, a parameter study was conducted at the University of Stuttgart. This study analyses the influence of the measurement device, the mandrels and the measuring procedure on the results. Based on the study results, recommendations are derived and summarized in a best-practice guideline, which should enable an appropriate and reproducible measurement of the radial load. Keywords Radial Lip Seal, Radial Load, DIN 3761-9, Radial Load Measurement, Radial Load Measuring Device, Measuring Method, Recommendation, Best Practice Guideline Kurzfassung Abstract * Dipl.-Ing. Simon Feldmeth Orcid-ID: https: / / orcid.org/ 0000-0003-0018-0710 Dr.-Ing. Mario Stoll Orcid-ID: https: / / orcid.org/ 0000-0001-7091-2396 PD Dr.-Ing. Frank Bauer Orcid-ID: https: / / orcid.org/ 0000-0001-7799-7628 Universität Stuttgart Institut für Maschinenelemente (IMA) Pfaffenwaldring 9, 70569 Stuttgart www.ima.uni-stuttgart.de TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 5 pairs of leaf springs support the movable mandrel and allow it to move frictionless in the measuring direction. An improved measuring device was developed and built at the University of Stuttgart, Institute of Machine Components (IMA): A position sensor and a linear actuator were added to realize a position control of the movable mandrel, Figure 3. This control system allows controlling the position of the movable mandrel and thus the effective diameter of the mandrels during the measurement (“Automatic Diameter Control”, ADC). This ensures that the mandrels widen the sealing lip to the correct cir- Aus Wissenschaft und Forschung 6 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 shaft. One of the mandrels is fixed; the other one is movable in the measuring direction. A load cell measures the force F m , with which the two mandrels are pressed together in the measuring direction by the sealing lip. The radial load is obtained by multiplying the measured load with the factor of pi. Measuring Devices DIN 3761-9 [7] specifies a device for measuring the radial load based on the split-shaft method, Figure 4. Two air side fluid side fluid garter spring membrane shaft metal insert housing contact area radial lip seal Ød sealing lip shaft surface sealing edge radial load Figure 1: Cross section of a radial lip seal movable mandrel stationary mandrel position sensor gap width frictionless parallel guidance (leaf springs) interchangeable load cell actuator seal ring clamping shoes stationary base movable base Ød sliding distance h m * * h m = height of measuring position measured load Figure 3: Schematic diagram of the improved measuring device radial load F r movable mandrel stationary mandrel projected radial load measured load F m = F r / Ød gap measuring direction Figure 2: split-shaft measuring method TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 6 cumference. Additionally, the control system offers features such as the so-called “Mounting Assistant” which can be used to move the mandrels temporarily to each other in order to facilitate the mounting process. Moreover, an optimized clamping system for the mandrels was implemented which allows users to change and accurately align the mandrels without any tools. Existing mandrels manufactured for the standard device (Figure 4) can be used with a clamping adapter. There are two versions of the improved measuring device, Figure 5 and 6: A compact tabletop version allowing mobile measurements and a version with temperature chamber allowing measurements in a range of -30 … 150 °C. Additionally, the temperature-controlled device contains four interchangeable load cells with different measuring ranges. Measuring Procedure DIN 3761-9 [7] defines the measuring procedure only in a very rough manner. Since this standard does not describe the procedure in detail, a huge variety of different measuring procedures arose during the past 30 years. Some companies and research institutes defined their own internal measuring standards. At the Institute of Machine Components (IMA), several methods have been developed which are suited to answer a wide variety of specific question. These methods comprise the preconditioning of the seal and the measuring procedure: • Method A: For quality control (especially applicable on many seals), one single measurement per seal at room temperature with a defined measuring time of 10 s is sufficient. • Method B: Performing 5 measurements per seal according to Method A and averaging of the 2 nd to the 5 th measured values is statistically more accurate and corresponds more to the installation conditions. The seal has to be rotated by 90° between each single measurements. • Method C: Storing the seal for 24 hours on an unsplit mandrel of nominal diameter before performing a measurement according to Method B provides the radial load after assembly and before operation. Due to the viscoelastic stress relaxation, this radial load is generally lower than that measured without storing the seal on a mandrel. • Method D: In order to obtain the radial load during operation, both the storing and the measurement according to Method C have to be performed at the operating temperature. This method provides the “real” radial load of the seal, which is even lower due to an accelerated relaxation and the thermal expansion of the sealing lip. Analysed Parameters Currently, the standard DIN 3761 [4] is under revision. In order to update and improve the instructions and recommendations for measuring the radial load a comprehensive analysis of the measuring procedure is performed at the University of Stuttgart. In this analysis, Aus Wissenschaft und Forschung 7 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 Figure 4: Standard device Figure 5: Tabletop IMA device Figure 6: Temperature-controlled IMA device TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 7 0.7 N (1.7 %). The arithmetic mean was 38.7 N corresponding to a radial load related to the circumference of 0.15 N/ mm. In order to identify even small influences of the varied parameters a special test plan was developed to reduce the influence of disturbance variables. For this purpose, the parameter study was divided into several test series, in which one or two parameters were varied systematically. Test series with contradictory results were repeated. In each test series, 4 to 8 different parameter combinations were analysed. In order to eliminate sequence effects that might occur due to the viscoelastic material behaviour during one test series, the 20 seals were divided into several groups of equal size. For each group, another sequence of the parameter combinations was used. At the end, the radial load was averaged over all groups, i.e. all 20 seals. Using Method B, each radial load value shown in the following is based on 20 x 4 = 80 individual measurements. Aus Wissenschaft und Forschung 8 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 12 parameters influencing the radial load are varied systematically. They can be sorted in 3 groups regarding the measuring device, the mandrels and the measuring procedure. Table 1 lists all varied parameters. The default settings are underlined. Test Material and Test Procedure The analysis was performed using 20 identical radial lip seals without protection lip according to DIN 3760 [3], Table 2. Many repeated measurements (using the procedure described in the following paragraph) show that the measurement results fluctuate even if all parameters - including the test person - remain the same. Repeating the same measurement (with the default settings listed in Table 1) for 19 times in a period of 21 months, there was a difference between minimum and maximum of all measurements of 1.9 N (4.9 %) and a standard deviation of Parameters of Group 1: Measuring Device Measuring device “Standard” device*, Tabletop IMA device, Temperature-controlled IMA device Load cell (maximum radial load) LC2 (150 N), LC3 (600 N), LC4 (3000 N) Automatic Diameter Control (ADC) On, Off Mandrel offset (relative to the nominal diameter) 0 µm, -100 µm, -200 µm (only negative offsets to avoid overstretching of sealing lip) Parameters of Group 2: Measuring Mandrels Mandrel surface (mean roughness depth measured in axial direction) Steel turned (Rz 3.3), aluminium turned (Rz 5.1), aluminium anodized (Rz 7.7) Sliding distance of sealing edge on mandrel cylinder** 11 mm, 21 mm, 31 mm Height of measuring position h m ** 7 mm, 27 mm Gap width between mandrels** 0.2 mm,1.0 mm, 2.0 mm, 4.0 mm Parameters of Group 3: Measuring Procedure Mounting direction Air side ahead, fluid side ahead Mounting motion Straight, with 30° rotation during mounting Lubricant Dry (no lubricant), FVA reference oil No. 3 Temperature of seal rings before measurement 8 °C, 21.5 °C, 35 °C Table 1: Varied parameters (default settings are underlined) Dimension 80x100x10 [mm] Manufacturer Freudenberg Sealing Technologies, Weinheim, Germany Profile BAUM5X7 Material 75 FKM 585 (fluoroelastomer) Year of Production 2014 (Best before 2024) Garter spring All measurements were performed with original garter spring Table 2: Radial lip seals * manufactured by Hottinger Baldwin Messtechnik in 1986 according to DIN 3761-9: 1984 ** see Figure 3 for definition TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 8 Results Parameters of Group 1: Measuring Device A) Measuring Device Figure 7 compares the radial load measured with the three different measuring devices described above. For a proper comparison, the Automatic Diameter Control was deactivated at the two IMA devices. A mandrel pair with an anodized aluminium surface and a clamping interface for the standard interface was used. A clamping adapter was used to mount this mandrel pair at the two IMA devices. The three measuring devices show a good agreement width a deviation of less than 4 % (related to the temperature-controlled IMA device). B) Automatic Diameter Control and Mandrel Offset Figure 8 shows the influence of the Automatic Diameter Control (ADC) and the mandrel offset. Activating the Automatic Diameter Control (with zero mandrel offset), the radial load is 8.9 % higher since the sealing lip is now widened to the nominal circumference. Additionally, the actuator can be used to set a defined offset of the mandrels. A negative offset of the mandrels corresponds to reducing the distance (gap) between the two mandrels. Moving the mandrels together reduces the effective circumference of the mandrels and decreases the radial load with a rate of 3.7 N per 100 µm offset in radial direction (for the case with ADC). C) Load Cell (Maximum Load/ Stiffness) Figure 9 shows the influence of the load cells which can be easily exchanged using the temperature-controlled IMA device. Activating the Automatic Diameter Control (ADC), there is no significant difference (lower than 1.8 %) between the different load cells. Without ADC, the radial load increases with an increasing measuring range. The higher the measuring range and accordingly the stiffness of the load cells, the less the mandrels move towards each other and the smaller the reduction of the effective circumference. The difference between the stiffest load cell LC4 (130µm/ 3000N radial load) and the softer load cell (LC2, 180 µm/ 150N radial load) is 2.9 %. Parameters of Group 2: Measuring Mandrels D) Surface of the Mandrels Figure 10 shows the radial load measured with different mandrel surfaces. The radial load is highest with anodizes aluminium (Rz 7.7) and lowest with turned steel (Rz 3.3). With turned aluminium mandrels (Rz 5.1), the radial load is in between. Independent of the lubrication, the radial load increases with increasing roughness. The difference related to turned aluminium is -1.6 % (turned steel) and Aus Wissenschaft und Forschung 9 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 32.5 34.3 33.5 20 25 30 35 40 45 standard device tabletop IMA device temp. controlled IMA device radial load [N] Figure 7: Influence of measuring devices 31.7 32.4 34.7 33.4 34.0 37.8 20 25 30 35 40 45 -200 -150 -100 -50 0 radial load [N] mandrel offset [μm] without ADC with ADC Figure 8: Influence of Automatic Diameter Control (ADC) and mandrel offset 34.6 35.5 35.7 37.2 36.7 37.4 20 25 30 35 40 45 LC2 LC3 LC4 radial load [N] without ADC with ADC Figure 9: Influence of load cells (LC) TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 9 +0.1 % (anodized aluminium) or with lubricant -0.6 % (turned steel) and +2.2 % (anodized aluminium). E) Sliding Distance on Mandrels and Height of Measuring Position Figure 11 shows the radial load measured with different sliding distances. Spacer rings were mounted on the mandrels to reduce the distance that the sealing edge slides over the mandrels cylinder before reaching the measuring position. Increasing the sliding distance from 11 to 31 mm decreases the radial load by 4.5 %. A further measurement with 11 mm sliding distance was performed, whereby the axial height of the measuring position was increased by 20 mm. This measurement provides the same result (difference below 0.1 N) and thus shows that only the sliding distance influences the radial load. F) Gap Width of Mandrels Figure 12 shows the radial load measured with mandrels that have the same diameter but different gap widths. The result is the same both with and without Automatic Diameter Control (ADC): The mandrels with a gap of 1.0 mm show the lowest radial load. Decreasing the radial load to 0.2 mm increases the radial load. Increasing the gap width to 2.0 and 4.0 mm increases the radial load as well. With ADC, the difference between 4.0 and 1.0 mm is 4.4 %. The difference between 2.0 or 0.2 mm and 1.0 mm is only about 1.3 %. Parameters of Group 3: Measuring Procedure G) Mounting Direction and Mounting Motion Figure 13 shows the radial load obtained with different mounting directions and mounting motions. Rotating the seal by approximately 30° while sliding on the mandrels the radial load reduces by 1.7 %. Mounting the seal with Aus Wissenschaft und Forschung 10 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 37.6 37.8 38.6 36.2 36.8 36.8 20 25 30 35 40 45 turned turned anodized steel aluminium radial load [N] with oil dry Figure 10: Influence of mandrel surface 38.4 38.1 36.7 38.4 20 25 30 35 40 45 0 10 20 30 40 radial load [N] sliding distance [mm] 7 mm 27 mm height h m : Figure 11: Influence of sliding distance and height of measuring position 37.0 33.4 37.7 20 25 30 35 40 45 air side ahead fluid side ahead radial load [N] 30° rota on straight Figure 13: Influence of mounting direction and mounting motion 33.8 32.4 34.0 34.8 35.3 34.8 35.2 36.3 20 25 30 35 40 45 0 1 2 3 4 radial load [N] gap width [mm] without ADC with ADC Figure 12: Influence of gap width TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 10 the fluid side ahead (and rotating the seal to avoid a buckling of the sealing lip) reduces the radial load by 9.7 %. H) Lubrication Figure 14 shows the radial load measured with dry and oil wetted mandrels. Lubricating the mandrels increases the radial load by 2.7 % (with ADC) or 2.1 % (without ADC). I) Temperature The seal rings were tempered for several hours before the measurement at 8 °C (refrigerator), 21.5 °C (room temperature) and 35 °C (heating oven). The measuring device and the mandrels were always at room temperature. The seals were taken out of the refrigerator or tempering oven separately just before each measurement. Due to the short measuring time (10 s) and the low thermal conductivity of the elastomer, there was no significant temperature adjustment. Seals tempered at 8 °C show a radial load that is 9.1 % higher. Seals tempered at 35 °C show a radial load that is 1.9 % lower, Figure 15. Summary and Conclusion A comprehensive parameter study was performed to analyse the effect of 12 parameters on the radial load measurement. The following conclusions can be drawn: • The results obtained with the three different measuring devices agree very well, if the Automatic Diameter Control (ADC) is deactivated. • The radial load presses the two mandrels together resulting in a lower distance of the mandrels and thus a lower effective circumference of the sealing lip. This results in a lower radial load due to the lower widening of the sealing lip. This effect is greater with softer load cells as they allow the mandrels to move more to each other. The stiffness of the load cell must not be too high. Due to their working principle, load cells with strain gauges must provide a certain amount of elasticity. • Therefore, the Automatic Diameter Control (ADC) is a useful feature. The ADC can compensate for the finite stiffness of the measuring chain (including the load cell) and thus allowing precise measurements of the “real” radial load at nominal effective circumference of the sealing lip. • It is important to use defined mandrels with the same surface roughness, the same height (resulting in a comparable sliding distance) and the same gap width:  The measured radial load increases with rougher surfaces.  The measured radial load decrease with longer sliding distances.  The measured radial load is lowest for a gap width of 1.0 mm.  The height of the measuring position has no influence of the radial load. • During mounting the seal on the mandrels, frictional forces act on the sealing lip. Their axial components stretch/ compress and tilt the sealing lip and may alter the radial load. Lubrication, mounting direction and mounting motion affect the magnitude and orientation of the frictional forces acting on the sealing lip and thus may explain the observed influences:  Lubricating the mandrels with oil increases the radial load.  Rotating the seal while sliding on the mandrels reduces the radial load.  Mounting the seal with the fluid side ahead reduces the radial load drastically. • The measured radial load decreases with increasing temperature and vice versa. Aus Wissenschaft und Forschung 11 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 35.0 37.1 34.3 36.1 20 25 30 35 40 45 without ADC with ADC radial load [N] with oil dry Figure 14: Influence of lubrication 39.6 36.2 35.6 20 25 30 35 40 45 0 10 20 30 40 radial load [N] temperature [°C] Figure 15: Influence of temperature TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 11 measured on classical devices or on devices with deactivated ADC. The difference is in the order of 10 %, depending on the stiffness of the seal. • Ensure, that the mandrels meet the requirements defined in DIN 3761-9 such as surface roughness, gap width and sliding distance. Considering these recommendations an appropriate and reproducible measurement of the radial load can be achieved. References [1] Horve, Leslie: Shaft Seals for Dynamic Applications. New York : Dekker, 1996. [2] Müller, Heinz K.; Nau, Bernard S.: Fluid Sealing Technology - Principles and Applications. New York : Dekker, 1998. [3] DIN 3760. Rotary shaft lip type seals. September 1996. (in German) [4] DIN 3761. Rotary shaft lip type seals for automobiles. 15 Parts. November 1983 to January 1984. (in German, withdrawn in March 2017) [5] ISO 6194. Rotary shaft lip-type seals incorporating elastomeric sealing elements. 5 Parts. September 2007 to November 2009. [6] DIN 3761. Part 1. Rotary shaft lip type seals for automobiles. Terms, formula symbols, tolerances. January 1984. (in German, withdrawn in March 2017) [7] DIN 3761. Part 9. Rotary shaft lip type seals for automobiles. Test. Radial force measuring instrument digital. January 1984. (in German, withdrawn in March 2017) Aus Wissenschaft und Forschung 12 Tribologie + Schmierungstechnik · 68. Jahrgang · 3-4/ 2021 DOI 10.24053/ TuS-2021-0014 Best-Practice Guideline Based on the conclusions above, the following recommendations are proposed: • Choose a method that is suited for answering your specific question initiating the measurement:  Method A (1 single measurement at room temperature with 10 seconds measuring time) for a quick quality control - especially for a high quantity of seals.  Method B (repeating Method A five times per seal with rotation the seal by 90° between each measurement and averaging the second to the fifth measured value) for obtaining more accurate results that correspond more to the installation conditions.  Method C (with 24 h storage on unsplit mandrel before measuring) for obtaining the radial load after assembly and before operation.  Method D (Method C with storing and measuring at operating temperature) for obtaining the “real” radial load during operation. • Before measuring, store the seals at a defined temperature for a sufficiently long time (several hours for common seal sizes). • Perform all radial load measurements consequently without lubricant. • Always use the same mounting direction and mounting motion. For standard profiles, it is suitable to mount the seal with the airside ahead without rotation. • Use a device with an Automatic Diameter Control (ADC) to obtain the “real” radial load. Be aware that this “real” radial load is higher than the radial load TuS_3_4_2021.qxp_TuS_Muster_2021 03.09.21 13: 27 Seite 12