1 Introduction In accordance with the guidelines summarized in [1], transport casks for radioactive material are subject to ambient temperatures between -40 °C and 38 °C under normal transport conditions. Due to the heat generation by the radioactive contents (often nuclear fuel assemblies or waste from reprocessing of nuclear fuel assemblies), transport casks may have surface temperatures of up to 90 °C. Many transport casks are cylindrically shaped and often equipped with impact limiters (shock absorbers) at each end (see Figure 1), to reduce the loads on the main parts of the containment such as cask body, lids and screws. Impact limiters are usually filled with layers of wood and, in accident conditions, absorb a ma- Aus Wissenschaft und Forschung 5 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 Friction coefficients for wood-wood and wood-steel high pressure contact under temperatures between -40 °C and 90 °C Martin Neumann, Tobias Gleim, Thomas Gradt* Eingereicht: 6.9.2022 Nach Begutachtung angenommen: 28.2.2023 Dieser Beitrag wurde im Rahmen der 63. Tribologie-Fachtagung 2022 der Gesellschaft für Tribologie (GfT) eingereicht. Holz wird vielfach in Stoßdämpfern von Transportbehältern für radioaktives Material eingesetzt. Dafür wird oftmals Fichtenholz verwendet, das in mehreren Lagen und senkrecht zueinander liegender Faserrichtung zwischen einer inneren und einer äußeren Stahlhülle eingebaut wird. Die Reibung zwischen den dabei entstehenden Grenzflächen ist von entscheidender Bedeutung für die Stoß- und Energieabsorption z. B. bei einem Aufprall des Behälters auf eine harte Unterlage. Um detaillierte Informationen für entsprechende numerische Simulationen zu bekommen, wurden in dieser Arbeit die Reibungszahlen von Holz- Holz- und Holz-Stahl-Paarungen bei für derartige Behälter relevanten Beanspruchungsbedingungen im Temperaturbereich zwischen -40 °C und 90 °C gemessen. Die Ergebnisse zeigen eine gleichmäßige, mit steigender Temperatur abfallende Reibung, wobei die Werte für Holz-Stahl-Paarungen zwischen 0,43 bei -40 °C und 0,22 bei 90 °C sowie für Holz-Holz- Kontakte zwischen 0,3 bei -40 °C und 0,24 bei 90 °C liegen. Schlüsselwörter Castor-Behälter, Stoßdämpfer, Energieabsorption, Reibung, Holz Wood is widely used in impact limiters of transport casks for radioactive material. Encapsulated by an outer and inner steel structure, spruce wood is often applied in layers of alternating direction. The friction at the interfaces between these layers is of crucial importance for the impact and energy absorption e.g., at an accidental impact of a cask against a hard component. In order to get detailed information for corresponding numerical calculations, in this study the friction coefficient for the combinations wood-wood and wood-steel was measured in the temperature range between -40 °C and 90 °C according to the relevant stress conditions for such casks. Results show decreasing friction with increasing temperature, ranging from 0.43 at -40 °C to 0.22 for 90 °C for wood-steel combinations and from 0.3 at -40 °C to 0.24 at 90 °C to for a wood-wood combination. Keywords transport casks for nuclear waste, impact limiter, energy absorption, friction, wood Kurzfassung Abstract * Dr.-Ing. Martin Neumann (federführender Autor) Orcid-ID: https: / / orcid.org/ 0000-0001-8806-345X Dr.-Ing.Tobias Gleim Orcid-ID: https: / / orcid.org/ 0000-0002-6373-9799 Dr. Thomas Gradt Orcid-ID: https: / / orcid.org/ 0000-0002-0542-4123 Bundesanstalt für Materialforschung und -prüfung (BAM) Unter den Eichen 87, 12203 Berlin 2 Materials and Methods 2.1. Experimental Setup The experiments to determine the coefficient of friction µ were carried out at the Federal Institute for Materials Research and Testing (BAM), partly at the Department of Physics at the FU Berlin. In order to test the frictional behavior of the sample pairs for the case described in the introduction under appropriate boundary conditions, the apparatus [6] shown in Figure 2 was used. The experimental setup with the sample pads is inserted into a cryostat, which is typically used for low temperature experiments, see Figure 2 left, but it is also suitable for the tests of this study. As described in the introduction, the cask with the wooden filled limiters is designed to withstand temperatures between -40 °C and approx. 90 °C. Therefore, tests are carried out at temperatures of -40 °C, 20 °C and 90 °C under adiabatic conditions. -40 °C is produced by cold nitrogen gas flowing through a heat exchanger attached to the plunger and the load frame of the tribometer, whereas 90 °C was performed by heating the pads and the machine parts surroundings electrically. No precautions were taken for 20 °C, as this corresponds to the room temperature in the laboratory. The test apparatus can be used for normal forces up to 150 kN, whereas in the present case the experiments are carried out at approx. 5 kN. Ambient atmospheric pressure was present in the laboratory as well as in the test apparatus. Figure 2 right shows a section through the tribometer unit, which, in the case of cryogenic tests, is completely immersed into the refrigerant. The reciprocating movement is generated by a linear drive on the top flange of the cryostat. It is transferred to the inner samples by means of a thin-walled stroke transmission. The inner samples are mounted to the reciprocating plunger. Contrary to the principal drawing of Figure 2, for the tests of this study the inner samples have the shape of rectangular plates as shown in Figure 3. The circular counter bodies are held in position by a fork-shaped structure on which strain gauges are applied for individual measurement of the frictional forces. The load is applied via a roller-wedge system. When the wedges are pulled towards each other, the rollers move in the direction of increasing thickness of the wedges. This moves a thrust plate in the direction of the counter bodies, and the load frame, which encloses the entire assembly, is moved in the opposite direction. Thus, both specimens are loaded symmetrically and the resulting horizontal forces on the loading and driving systems are minimized. During the tests, the total friction force is measured by a load cell load on the top flange, displacement, normal force, and individual frictional forces are measured by strain gauges attached to the fork arms and load frame. The signal of the load with an accuracy of 0.1 % is used for the calculation of the mean friction coefficient of both samples. Since both samples are not Aus Wissenschaft und Forschung 6 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 jor part of the kinetic energy due to severe impacts. In Germany, spruce wood is widely used as an energy absorber for impact limiters. Upon impact, inertia forces cause compression of wood and lateral evasion of wood layers [1], which often results in corresponding friction. Although numerous data of tribological properties of material combinations under different boundary conditions are available in the literature, only little information can be found for wood as one of the mating materials. Therefore, in the present study, the tribological properties between the wood layers and at the interface between the wood and the steel of the encapsulated shell are to be investigated (see Figure 1 right). In contrast to homogenous materials such as steel, wood, due to its organic composition, has particular influencing parameters such as orthotropy, wood fiber composition and wood moisture, which may have a major impact on the friction coefficient [2], [3], [4]. Shock absorbers are usually airsealed in a sheet metal encapsulated casing with planed spruce wood and a wood moisture content of approx. 15 %. Due to orthotropy, the individual wood layers are orthogonally piled to each other, to avoid a fiber-related preferred direction. While the previously mentioned parameters such as fiber direction and wood moisture are kept constant in average, the temperature, on the other hand, may cause large differences due to the range of application and the heat generation due to the radioactive contents. In order to perform adequate numerical simulations of an accident scenario, the contact between the wood layers and the steel must be modeled accordingly. While some tabulated values for room temperature are available in the relevant literature [5], the temperature effect has not been sufficiently investigated. Therefore, in the present paper, the influence of temperature is studied for cross-stacked wood with a wood moisture content of about 15 %. The experimental investigations will focus on the corresponding temperature boundary conditions and the material pairing wood - wood and wood - steel. Figure 1: Schematic diagram of a transport cask with impact limiters including view of the internal structure of an impact limiter with compartments containing layered wood completely decoupled and the deformation of the forkshaped structure is small, the accuracy of the individual force measurements is limited. Nevertheless, these data can provide an indication of different friction characteristics of the specimens. Furthermore, in friction tests the data scatter usually is much higher than the error of the force measurement. Thus, the accuracy of this method should be sufficient. 2.2. Materials The spruce wood analyzed and tested in this study is a commercial wood beam with grade S13 according to [7]. Care needs to be taken to ensure that the samples are homogeneous and free of marks. As a finishing treatment, the sawed rough wood samples are planed to a thickness of 3.5 mm. The final dimensions after treatment can be taken from Figure 3. The commercial wood beams have a density of 380-450 kg/ m 3 and a measured wood moisture content of 6.4-7.8 %. For the wood-steel experiments, the sample pads of the counter bodies are made of wood and the rectangular plates in the plunger are made of classical S235 steel with a surface roughness Ra = 0.35 µm. 2.3. Test Parameters and Realization For all tests, certain parameters were kept constant, while the temperature was varied in three steps (see Table 1). As initial normal force, a value close to the compressive limit of the wood perpendicular to the fiber direction (~ 6 MPa) was chosen, in order to simulate the substantial normal forces during impact limiter compression. Before starting each measurement, all force signals were set to zero. Subsequently, the load was applied, and the movement was started. Although the normal force decreased due to the compression of the wood, it was decided not to readjust it during the test. In the employed test apparatus, readjustment is only possible manually, which would be a potential source of error. Therefore, the focus was put on similar initial conditions for all tests and decreasing load was tolerated. 6 combinations of wood-wood and wood-steel and temperatures of -40 °C, 20 °C and 90 °C were tested. Three cyclic replicates were typically performed for each sam- Aus Wissenschaft und Forschung 7 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 Figure 3: Dimensions of circular and rectangular samples with for the counter bodies Figure 2: BAM cryo-tribometer for high loads Specification Initial Normal force Sliding speed Displacement Temperature Cycles per test Table 1: Test parameters Value 5 kN 0.1 mm/ s 1.8 mm -40°C, 20°C, 90°C 3 10 times with new samples for a statistical evaluation. Since the steel does not undergo any change, in the wood-steel tests, only the wood samples were changed. The test apparatus shows a sight force dependence of the cycle length. In order to compare the curves of the 10 repeats, the individual cycle durations were related to one reference cycle. In consequence, the time scale would be slightly different for each curve. Therefore, it was decided to use the cycle number for the scale of the x-axis. The duration of one cycle is about 50 s. Figure 4 shows the development of the normal and friction forces for all 10 tests of the wood-wood combinations at 90 °C and the resulting mean value (blue line). In diagram a) it can be clearly seen that the normal force decreases over the three cycles due to the compression of the material. In consequence, also the friction force decreases (b), but its development is very similar for the 10 individual tests. Only the first half cycles show larger differences in single cases, but in general Figure 4 indi- Aus Wissenschaft und Forschung 8 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 ple. For randomly selected test examples, up to six cycles were performed, although these did not provide any additional behavior or information. As mentioned above, the wood layers are always piled orthogonally to each other. Therefore, all samples of wood-wood combinations were mounted with orthogonal wood fiber direction. In the case of the wood-steel combination, only sliding transverse to the wood fiber direction was studied. 3 Results and Discussion In the following, the results from all tests are presented by means of different figures. Due to the reciprocating motion, the quotient of friction force and load has negative values during backward motion. To avoid negative values for the friction coefficient, F R / F N is used instead of µ for the axis labels in these cases. Each of the 6 combinations of material and temperature were repeated a) b) Figure 4: Development of load (a) and friction force (b) for the first three cycles of self-mated wood couples at 90 °C a) b) Figure 5: Friction loops for randomly selected tests of wood-wood (a) and wood-steel couples (b) at different temperatures cates a good reproducibility of the frictional behavior of this material combination. The development of the friction over the displacement is shown in Figure 5 for all three temperatures. Despite the decreasing normal force, in most cases the ratio of friction to normal force is constant. Diagram a) shows the behavior of the wood-wood couples and it reveals that the influence of the temperature for this combination is much lower than for the wood-steel couples (b). With a value of approx. 0.2, the lowest friction coefficient is observed at 90 °C in both cases. For the wood-steel couples the maximum coefficient of friction occurs at -40 °C with a value close to 0.6. For the self-mated wood couples, friction shows only a very small increase at room temperature and at -40 °C it doesn’t rise above 0.4. Remarkably, the sliding state is very stable, and no static friction peak is observed neither for the wood-wood nor for the wood-steel couples. Only small stick-slip behavior is observed at 20 °C and 90 °C for both combinations. At 20 °C and -40 °C the friction loops of the woodsteel couples show differences between the individual cycles, which is a hint for a load dependence of the friction coefficient. The reasons for this and the small fluctuations cannot be identified by these tests and were not in the scope of the investigations. Figure 6 shows the ratio of friction to normal force of wood-wood couples at -40 °C (a) and 90 °C (b) for all 10 repetitions and the respective mean values over three cycles. Like Figure 4, also Figure 6 demonstrates that it is not possible to reproduce the friction coefficient exactly, but for an inhomogeneous material such as wood the reproducibility is sufficient. Statistical analyses of these test series are shown in Figures 7 to 9. The values represent the six friction coefficients in the stable sliding state during the three friction cycles, three at the forward and three at the backward motion. Figure 7 shows the scatter of the friction coefficient at 20 °C for wood-wood (a) and for wood-steel (b). Each of the six data points represents the mean value Aus Wissenschaft und Forschung 9 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 a) b) Figure 6: F R / F N -ratio for self-mated wood couples a) T = -40 °C; b) T = 90 °C a) b) Figure 7: Statistical evaluation of the friction coefficient; T = 20 °C; error bars: standard deviation for each data point; blue line and error bar: overall mean value and standard deviation; a) wood-wood; b) wood-steel cycle has a higher value, whereas all other ones are very similar . For wood-steel friction, a noticeable feature is seen for -40 °C by the increase in friction from half cycle to half cycle (Figure 9 a). The lowest friction occurs during the first half cycle and successively increases with the number of cycles, indicating an increasing friction coefficient with decreasing contact pressure. At 90 °C this combination shows a similar behavior as at 20 °C (Figure 7 b). The coefficient of friction fluctuates from half cycle to half cycle around the mean value and has a significantly lower standard deviation compared to the other temperatures. In summary, Table 2 shows the friction coefficients for the six combinations studied. As already explained for the figures, a dependence with temperature is observed Aus Wissenschaft und Forschung 10 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 and the standard deviation of 10 corresponding repeats. For the wood-wood interface it can be noted that the first half cycle has a slightly higher friction coefficient, whereas the other cycles show very similar values. The blue line and error bar represent the mean value over all six data points and the overall standard deviation. Compared to the wood-wood friction at 20 °C, a slightly higher friction coefficient is obtained for the wood-steel couples (see Figure 7 b). The friction coefficient is about 0.08 larger and the total standard deviation is comparable. Figure 8 shows the friction coefficient for wood-wood friction at temperatures of -40 °C (a) and 90 °C (b). As already indicated in Figure 5, the friction coefficient clearly shows a decrease with an increasing temperature. For -40 °C, a very constant friction coefficient is observed across all cycles. In comparison, at 90 °C the first half a) b) Figure 8: Statistical evaluation of the friction coefficient for wood-wood couples; error bars: standard deviation for each half cycle; blue line and error bar: overall mean value and standard deviation; a) -40 °C; b) 90 °C a) b) Figure 9: Statistical evaluation of the friction coefficient for wood-steel couples; error bars: standard deviation for each half cycle; blue line an error bar: overall mean value and standard deviation; a) -40 °C; b) 90 °C for both friction pairs. At the same time, the difference between the friction pairs can be presented. The woodsteel combination has a higher friction at -40 °C and 20 °C, whereas for 90 °C, an almost similar friction can be determined. The friction coefficients from Table 2 are depicted in Figure 10 for the respective mean values (solid lines) and the range of variations (dashed lines). For the investigated temperature range, a weakly nonlinear curve is obtained for both friction pairs. Comparing with the literature, it can be observed that for room temperature, i.e. approx. 20 °C, a similar or lower friction coefficient exists for the wood-wood and wood-steel combination, c.f. [8, 9]. In general, it can be noted that after each test, some of the wood samples have different surface structures. As examples, wood-wood couples after a test are shown in Figure 11. The couple of Figures 11 a) and b) shows no anomalies after the three cycles. Both specimens are smooth and only slightly compressed in the contact area. In contrast, the friction pair of c) and d) shows sawtoothlike grooves. In c) the grooves are clearly visible, whereas in d) they are only barely noticeable. A correlation between deviations in the graphs and the deformations (grooves or other plastic alterations) in the specimens could not be established. Also, no clear correlation could be found between the graphs where the first half cycle had a slightly higher increase and possible features in the samples, see for example Figure 4 b). 4 Summary For numerical simulations of impact limiters with layered wood, friction coefficients are needed to account for a contact formulation. These impact analyses must be investigated for different temperatures as boundary cases according to corresponding regulations. In a first study, the friction coefficients for self-mated, cross-layered wood and wood against steel were investigated for the application-relevant temperatures -40 °C, 20 °C, and 90 °C. For these studies, an initial compressive force close to the compressive limit of the wood was assumed, which would need to be additionally verified in future analyses. The results reveal a higher temperature sensitivity of the wood-steel combination compared to self-mated wood couples. Based on the results and the respective standard deviations, a weakly nonlinear behavior is observed. The results at room temperature are consistent with the literature, for other temperatures no reference data are available. In addition to the mean value, the curves also show a possible scatter range through the specification of the standard deviation, which must not be disregarded for limiting case considerations. The tests for these six combinations represent only one sliding speed and more detailed limiting case considerations need to include a wider range of sliding speeds. The effect has al- Aus Wissenschaft und Forschung 11 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 Combination / Temperature -40°C 20°C 90°C Wood-wood 0,30 ± 0,03 0,28 ± 0,03 0,24 ±0,02 Wood-steel 0,43 ± 0,05 0,36 ± 0,03 0,22 ±0,01 Table 2: Mean values for the friction coefficient including standard deviation Figure 10: Friction coefficients for wood-wood and wood-steel for the mean (solid line) and standard deviation (dashed lines) for different temperatures Figure 11: Wood samples with different effects: a) and b) friction couple with a smooth and inconspicuous surface; c) and d) friction couple with grooves occurring in c) 6 References [1] M. Neumann, Investigation of the Behavior of Shock-Absorbing Structural Parts of Transport Casks Holding Radioactive Substances in Terms of Design Testing and Risk Analysis, BAM-Dissertation Series - Volume 45, 2009. [2] W. M. McKenzie & H. Karpovich, The Frictional Behaviour of Wood, 2: 139-152, Wood Science and Technology, 1968. [3] W. Yin, Z. Liu, P. Tian, D. Tao, Y. Meng, Z. Han, Y. Tian, Tribological properties of wood as a cellular fiber-reinforced composite, 5: 67-73, Biotribology, [4] M. Seki, H. Sugimoto, T. Miki, K. Kanayama, Y. Furuta, Wood friction characteristics during exposure to high pressure: influence of wood/ metal tool surface finishing conditions, 59: 10-16, Wood Science and Technology, DOI 10.1007/ s10086-012-1295-1, 2013. [5] V. L. Popov, Kontaktmechanik und Reibung - Von der Nanotribologie bis zur Erdbebendynamik, 3. Auflage, Berlin: Springer Berlin Heidelberg, DOI 10.1007/ 978-3- 662-45975-1, 2015. [6] T. Gradt, K. Assmus, H. Boerner und T. Schneider, „Apparatus for Friction Tests of Support Elements in Fusion Devices“, Journal of Physics, Bd. Conference Series 100, p. 062032, 2008. [7] DIN 4074-1: 2012-06, Strength grading of wood - Part 1: Coniferous sawn timber. [8] K. Möhler & G. Maier, The Coefficient of Friction of Spruce Timber in View of the Efficiency of Timber-Connections Using Frictional Resistance, 27(8), 303-307, Karlsruhe: Holz als Roh- und Werkstoff , 08.1969. [9] N. Guan, B. Thunell & K. Lyth, On the Friction Between Steel and Some Common Swedish Wood Species, 41: 55- 60, Holz als Roh- und Werkstoff, 1983. Aus Wissenschaft und Forschung 12 Tribologie + Schmierungstechnik · 70. Jahrgang · 2/ 2023 DOI 10.24053/ TuS-2023-0007 ready been shown for other wood combinations [2, 8, 9] and must be further investigated in this context. Another important aspect for impact limiters is a possible influence of the contact pressure, which was investigated here only for the initial value which was shown to be decreasing during the tests. Again, the literature [5, 8, 9] shows that there is an effect on the friction coefficient and thus it must be considered for a limiting case analysis. Since the samples have residual moisture in all cases, it must be assumed that the samples contain or are covered by frozen water during the tests at -40 °C. Although no frozen surface water could be detected after the performed tests, it may be present at the microscopic level, which can lead to deviating friction coefficients depending on the residual moisture and could therefore be an additional point to investigate for a better understanding. 5 Acknowledgments We would like to thank Matthias Heidrich and Olaf Berndes for conducting the test series. We would also like to thank Detlef Rätsch for the consistent quality in the production of the wood samples. And last but not least, we would like to thank Frank Wille as head of division for the safety of transport containers for the support and freedom for the preparation of this paper.