Influence of surface and material technologies on the loss of lubrication performance of gears

Abstract

Enabling gears to withstand loss of lubrication in gearboxes without secondary oil supply systems can reduce weight and space demand and thus fuel consumption. This study investigates the potential of surface and material technologies on the loss of lubrication performance of gears. Thereby, superfinished, coated, and nitrided gears are compared to ground gears. Systematic experiments under loss of lubrication are performed at a back-to-back gear test rig with circumferential speeds of up to 20 m/s and Hertzian pressures in the pitch point of up to 1723 N/mm². Torque loss, pinion bulk temperatures, and tooth flank surface are analyzed. The results show that surface and material technologies can greatly influence frictional behavior and damage initiation of gears operating under loss of lubrication. With the materials and conditions tested, superfinishing yields to accelerated rise of frictional losses and thus scuffing. Coatings lead to significantly enhanced service life under loss of lubrication by friction reduction and scuffing avoidance.

Introduction

Helicopter transmissions and geared turbofans must withstand loss of lubrication (LOL) to ensure safe operation and landing (European Union Aviation Safety Agency 2019; Federal Aviation Administration 2007). LOL can occur due to oil leakage and oil pressure loss (Gasparini et al. 2014), at negative-g maneuvers, or at windmilling of the blades when the engine and oil pump are shut down (Groll 2000). LOL causes off-design oil flow in the transmission and interrupts oil supply to tribological contacts. As friction increases and heat dissipation deteriorates, the risk of scuffing and thermal damage increases.

The oil has a significant influence on the LOL performance. Faruck et al. (Faruck et al. 2020) determined the time-to-failure under LOL of different oils on an SRV (Schwing, Reibung und Verschleiß) tribometer. The best-performing oil showed a mean time-to-failure of 40 min compared to less than 5 min for the worst-performing oil. The results correlated partially to the scuffing load capacity obtained from FZG scuffing tests acc. to method A/8.3/90 (ISO 2000). On a ball-on-disk tribometer, Riggs et al. (Riggs et al. 2017) showed an improvement of the time-to-failure by 28% by a MIL-PRF-32538 (Naval Air Warfare Center Aircraft Division 2015) specified oil with a viscosity of 9 mm²/s at 100 °C compared to a DOD-PRF-85734 (Naval Air Systems Command 2004; Naval Air Warfare Center Aircraft Division 2019) oil with 5 mm²/s at 100 °C. Murthy et al. (Murthy et al. 2020) showed on a ball-on-disk tribometer that oil additives based on nano-graphene platelets (NG) and a phosphonium-phosphate-based ionic liquid (IL) increase the time-to-failure by a factor of two.

Beneath the influence of the oil, the base material influences the LOL performance. Alberti und Lemanski (Alberti and Lemanski 1972) compared spiral bevel gears made of carburized high hot hardness tool steel and case-hardened steel AISI 9310. Within the harshest LOL tests, overheating and considerable metal flow of both steels occurred with estimated gear bulk temperatures of over 600 °C. However, no significant improvement was found for the high-hot hardness steel. Also, Berkebile et al. (Berkebile et al. 2018) found that the influence of different base materials on the LOL time-to-failure of spur gears is comparably small. A maximum of 16% improvement was achieved with AISI M50 bearing steel compared to AISI 9310. High-hot hardness steels like Pyrowear 53 performed less than AISI 9310. Kreider and Lee (Kreider and Lee 1987) showed LOL time-to-failure improvement of tapered roller bearings by bearing cup and cone ribs made of oil-impregnated sintered steels. Under LOL, thermal effects and different coefficients of thermal expansion of oil and sintered steel caused the oil to bleed out and lubricate the roller bearings. Ebner et al. (Ebner et al. 2017) investigated spur gears made of oil-impregnated sintered steel.

Superfinishing of surfaces reduces friction and can have a positive effect on the LOL performance. On a ball-on-disk tribometer, Riggs et al. showed an increased LOL time-to-failure behavior with superfinished samples (Riggs et al. 2016, 2017). The results confirm that also under LOL, scuffing initiation is more sensitive to increased sliding speeds at the beginning and end of gear contact than at the single gear contact, where sliding speed decreases but load increases. They also showed that changing the oil from Mobil Jet II, specified as MIL-PRF-23699 (Naval Air Warfare Center Aircraft Division 2019), to AeroShell 555, specified as DOD-PRF-85734 (Naval Air Systems Command 2004; Naval Air Warfare Center Aircraft Division 2019), improves the positive effect of superfinishing. Berkebile et al. (Berkebile et al. 2018) achieved the best LOL performance by combining superfinished surfaces and oils with ionic liquid additive, improving the LOL time-to-failure of gears by a factor of 9. Contrary findings were found by Kozachyn et al. (Kozachyn et al. 2018) as they show a worse LOL performance with superfinished gears. A potential explanation is that the surface topography of superfinished surfaces allows for less oil retention. The authors state that there is not sufficient data to prove this hypothesis.

Tribological coatings are generally known for their friction reduction and high wear resistance. Riggs et al. (Riggs et al. 2020) showed on a ball-on-disk tribometer that applying lubricious nanocomposite coatings increases the LOL time-to-scuffing significantly. The positive effect is explained by the lubricious effect and sacrificially wearing the coating under LOL. Murakawa et al. (Murakawa et al. 1999) found an increase in the LOL lifetime by almost three times with WC/C (tungsten carbide/carbon) coated gears compared to ground gears. Kozachyn et al. (Kozachyn et al. 2018) evaluated the LOL behavior of various coatings applied on gears. Applying the coating to only one gear of the gear pair increased the LOL time-to-failure from 2.5 to 31.8 min. The LOL behavior showed a remarkable dependency on the considered oil.

Handschuh et al. (Handschuh et al. 2011) conducted systematic LOL experiments with and without emergency lubrication. Without emergency lubrication, the gearbox failed within several minutes. Isaacson and Wagner (Isaacson and Wagner 2018) conducted in-situ measurements of frictional losses during LOL and characterized the consequences of LOL into no scuffing, scuffing without progression to catastrophic failure, scuffing with progression to catastrophic failure, and immediate catastrophic failure. Morhard et al. (Morhard et al. 2024) investigated the torque loss and bulk temperature of gears facing LOL. The results show that scuffing damage is accompanied by a steep increase in torque loss and bulk temperature and that the operating condition has a significant influence. No scuffing was found at p_C = 1343 N/mm² and v_t = 8.3 m/s, scuffing marks were found at p_C = 1723 N/mm² and v_t = 8.3 m/s, and severe scuffing was found at p_C = 1343 N/mm² and v_t = 20 m/s. Increased circumferential speed supported oil centrifugation and reduced the resistance against LOL. The remaining and circulating splash oil in the transmission showed a substantial increase in the LOL performance of gears.

Literature results show a great potential of material and surface technologies to improve the LOL behavior. Many studies refer to the tribometer level. Based on a previous study by the authors (Morhard et al. 2024), this study focuses on the LOL performance of gears considering superfinishing, coating, and nitriding. Thereby, the torque loss and bulk temperature are analyzed.

Methods and materials

This section describes the experimental setup, the test gear geometry, the material and surface variants, the considered oil, the investigated operating conditions, and the experimental procedure.

Experimental setup

The test rig considered in this study is an adapted FZG efficiency test rig with a center distance of 91.5 mm based on the back-to-back principle. Figure 1 shows the layout of the test rig. Figure 2 shows a photograph of the assembled test gears in the test gearbox. The test rig and experimental setup are described in detail by Morhard et al. (Morhard et al. 2024). Nevertheless, it is briefly summarized in the following to improve readability.

Fig. 1

FZG efficiency test rig for LOL investigations (Morhard et al. 2024)

Fig. 2

Photograph of assembled test gears in test gearbox (Morhard et al. 2024)

For friction characterization of gears operating under LOL, a torque meter measured the torque loss T_L2 of the test and auxiliary gearbox. The test and auxiliary gear stages were into-mesh injection lubricated with a volume flow rate of 2 l/min. For LOL investigations, a valve bypasses the injection lubrication of the test gearbox to the oil sump of the oil rig. A short injection pipe and an adapted nozzle outlet diameter of 3.8 mm ensured that no remaining oil dripped on the gears during LOL. LOL was applied only to the test gearbox, and the bearings of the test gearbox were grease-lubricated and shielded. Therefore, changes in T_L2 were directly referable to changes in the gear mesh in the test gearbox. A torque meter derived the applied load in the closed loop between the test and auxiliary gearbox. Pt100 resistance thermometers measured the pinion and gear bulk temperature in the middle of one tooth 3 mm below the gear flank surface to characterize the thermal behavior.

Test gear geometry

The test gear geometry was the same as in the previous study of the authors (Morhard et al. 2024). The basic geometry refers to the FZG test gear of type C_mod. A tip relief reduces the tooth load at the beginning of the contact where high sliding speed occurs (Lechner 1966; Michaelis 1987). As Radil and Berkebile (Radil and Berkebile 2020) proposed, an adjusted tooth thickness tolerance to increase the backlash was applied to avoid jamming due to thermal expansion under LOL. The main gear parameters of the resulting test gear geometry of type C_mod,LOL are given in Table 1.

Table 1 Main parameters of FZG test gear of type C_mod,LOL (Morhard et al. 2024)

Material and surface variants

Six different material and surface variants were selected based on the state of the art. As a reference (REF), ground gears from the previous study of the authors (Morhard et al. 2024) were considered. The influence of superfinished gears was investigated by the variant SUF. The influence of different coated gears was investigated with the variants CO1, CO2, and CO3. Additionally, a nitrided gear variant NIT was considered. Figure 3 shows corresponding gear flank surface photographs of the material and surface variants.

Fig. 3

Gear flank surface photographs of the considered material and surface variants

Table 2 lists the material and surface properties of the considered gear variants. The gear variants REF, SUF, CO1, CO2, and CO3 were made of the typical aerospace steel AISI 9310 with a tempering temperature of 149 °C (Alberti and Lemanski 1972; Riggs et al. 2016, 2017). The gear variant REF was ground, and the gear variant SUF was additionally isotropic superfinished. In the chemically assisted superfinishing process, the ground surfaces were finished with non-abrasive stones with oxalic acids. The gear variants REF and SUF featured a surface hardness of 790 HV10 and a case hardness depth CHD_550HV1 of 1.2 mm. Based on the SUF gear variant, CO1, CO2, and CO3 were additionally microblasted and coated. For CO1, a tungsten-doped amorphous hydrogenated carbon coating (a-C:H:W (WC/C)) was applied via sputtering. The surface hardness was approx. 8 to 15 GPa, and the coating thickness was approx. 2 to 3 µm. For CO2, a multilayer coating consisting of chromium nitride (CrN) and a top layer was applied via plasma-assisted chemical vapor deposition. The surface hardness was approx. 15 to 25 GPa, and the coating thickness was approx. 3 to 4 µm. With CO3, a multilayer coating consisting of a tungsten-doped amorphous hydrogenated carbon coating (a-C:H:W) via physical vapor deposition and a top layer coating was investigated. The surface hardness was approx. 15 to 25 GPa and the coating thickness was approx. 2 to 3 µm. The general lower coating thickness in the tooth root flank area compared to the tooth tip flank area is related to the coating processes as the tooth root flank area is less accessible. The gear variant NIT was made of Nitralloy N with a nitriding temperature between 524 and 566 °C and ground as the variant REF. The white layer was removed. The surface hardness was 850 HV0.5, and the nitriding hardness depth NHD_460HV0.5 is 0.5 mm. For all variants, the surface roughness was measured by profile method with a measurement length of L_t = 4 mm and a cut-off wavelength of λ_c = 0.8 mm. The roughness measurements were carried out along the involute profile perpendicular to the grinding direction. The variant REF had an arithmetic mean roughness of Ra₁ = 0.33 µm for the pinion and Ra₂ = 0.25 µm for the wheel. The variant SUF had a decreased roughness of Ra₁ = 0.11 µm and Ra₂ = 0.09 µm. Due to microblasting, the surface roughness of the variants CO1, CO2, and CO3 was slightly increased compared to the variant SUF. The surface roughness of the variant NIT was slightly lower than that of the variant REF.

Table 2 Material and surface properties of the considered gear variants

Oil

As in the previous study (Morhard et al. 2024), the turbine oil AeroShell 500 was used in the test gearbox. It fulfills the specification MIL-PRF-23699 (Naval Air Warfare Center Aircraft Division 2019) and is based on a synthetic ester oil with a kinematic viscosity ν(40 °C) = 23 mm²/s and ν(100 °C) = 5.2 mm²/s. The density is ρ(15 °C) = 1,005 kg/m³. The scuffing load stage determined in the FZG scuffing test A/8.3/90 (ISO 2000) is 7 (Morhard et al. 2024).

Operating conditions and experimental procedure

The considered operating conditions correlated with the previous study (Morhard et al. 2024). Table 3 shows the applied load stages (LS) with corresponding pinion and wheel torque T₁ and T₂ and Hertzian pressure p_C, the applied circumferential speed v_t with corresponding pinion and wheel speeds n₁ and n₂, the applied oil injection temperature ϑ_Oil and the resulting transmitted power P_t.

Table 3 Considered operating conditions with load, speed, oil injection temperature, and resulting transmitted power

Each experiment started with a 30-min run-in of a new gear pair under oil injection lubrication with an oil injection temperature of ϑ_Oil = 90 °C. For uncoated variants, LS9 and v_t = 0.5 m/s were applied; for coated variants, LS7 and v_t = 0.5 m/s were applied.

After run-in, the test gearbox was initially injection-lubricated at the operating condition of interest (cf. Table 3) until torque loss and system temperatures reached quasi-steady-state conditions. Then, the oil injection in the test gearbox was manually shut off to investigate LOL. After the LOL time interval of interest, injection lubrication was reactivated.

A moderate operating condition with LS7 and v_t = 8.3 m/s, resulting in a transmitted power of P_t = 42 kW, was investigated to analyze the general characteristics of torque loss and bulk temperature under LOL. The influence of increased speed on the LOL behavior was investigated at LS7 and v_t = 20 m/s, resulting in P_t = 100 kW. The influence of increased load under LOL was investigated at LS9 and v_t = 8.3 m/s, resulting in P_t = 68 kW.

For the moderate operating condition, the LOL experiments were repeated two times with the same gear pair so that all variants faced three times LOL. For each LOL cycle, the oil was turned off, LOL faced for 20 s, and then the oil turned on again. At the operating condition with increased speed, all variants faced LOL once. At the operating condition with increased load, the variants REF, SUF, and NIT faced LOL once. The coated variants CO1, CO2, and CO3, faced three times LOL. For each LOL cycle, the oil was turned off, LOL faced for 20 s, and then the oil turned on again.

Results and discussion

This section discusses the experimental results of the LOL investigations of the considered surface and material variants. Thereby, torque loss measurements, gear bulk temperature measurements, and tooth flank surfaces after LOL investigations are shown. Note, that the gear bulk temperature is understood as the temperature measured 3 mm below the gear flank surface (cf. Fig. 2). The general LOL behavior and more operating conditions for the variant REF are presented in the previous study (Morhard et al. 2024).

Torque loss and gear bulk temperature

This section shows the results of the LOL experiments for the moderate operating condition (LS7 and v_t = 8.3 m/s), the operating condition with increased speed (LS7 and v_t = 20 m/s), and the operating condition with increased load (LS9 and v_t = 8.3 m/s).

Moderate operating condition

Figure 4 shows the measured torque loss T_L2 and the pinion bulk temperature ϑ₁ for LS7 and a circumferential speed of v_t = 8.3 m/s resulting in P_t = 42 kW. For a time t < 0 s, the test gearbox was injection-lubricated with ϑ_Oil = 90 °C. After reaching a quasi-steady-state condition, LOL was applied at t = 0 s for 20 s. After 20 s with LOL, the oil supply was activated again. For representation in Fig. 4, the second LOL experiment per variant is depicted.

Fig. 4

Torque loss T_L2 and pinion bulk temperature ϑ₁ under LOL at LS7 and v_t = 8.3 m/s (data for variant REF from Morhard et al. 2024)

Under injection lubrication (t < 0 s), the variants SUF, CO2, and CO3 featured a lower torque loss than the variants REF, CO1, and NIT. The lower torque loss correlates with results from Hinterstoißer et al. (Hinterstoißer et al. 2019) and is related to lower surface roughness and thermal insulation effect of coatings (Schwarz et al. (Schwarz et al. 2020)). The coated variant CO1 did not show a decreased torque loss under injection lubrication prior to LOL. The lower torque loss of the variants SUF, CO2, and CO3 resulted in lower pinion bulk temperatures ϑ₁ compared to REF. The torque loss and bulk temperature of the variant NIT were comparable to those of the REF variant.

When LOL was applied (t > 0 s), the torque loss T_L2 increased due to increasing gear friction. Additionally, the limited heat dissipation due to the absence of cooling oil increased bulk temperature ϑ₁. With reactivated injection lubrication, the torque loss dropped for the moderate operating condition to pre-LOL level within seconds. The measured bulk temperatures showed a short temperature increase at all LOL tests after reactivated injection lubrication, possibly due to thermal inertia and the heat capacity of the gears. The values of ϑ₁ reached pre-LOL temperatures approximately after five minutes.

The general torque loss behavior aligned with the previous study (Morhard et al. 2024). The surface and material variants showed only minor differences. The increase of torque loss and bulk temperature during LOL were similar. Noticeable is the increase of the loss torque towards the end of LOL for variant SUF and a smaller torque loss increase during LOL for variant CO3 compared to other variants.

Figure 5 shows the measured torque loss T_L2 for all three conducted LOL experiments per variant at moderate operating conditions. All three LOL experiments showed a very good repeatability for variants REF, SUF, CO2, and CO3, whereas variants CO1 and NIT showed a noticeable decrease in torque loss after the first LOL experiment. The trend of decreasing torque loss with the number of LOL experiments can be explained by further run-in or by a divergent amount of residual oil in the gearbox. The initial run-in of the variant NIT was found to have less effect than variant REF due to the hard nitride layer. This resulted in a higher torque loss compared to the variant REF in the first LOL experiment. Possibly due to further run-in of the hard nitride layer, the first LOL experiment decreased the torque loss of the variant NIT.

Fig. 5

Repeatability of measured torque loss T_L2 under LOL at LS7 and v_t = 8.3 m/s (with data for variant REF from Morhard et al. 2024)

Operating condition with increased speed

Increasing the circumferential speed to v_t = 20 m/s corresponds to an increase of P_t from 42 kW (LS7 and v_t = 8.3 m/s) to 100 kW (LS7 and v_t = 20 m/s). Figure 6 shows the corresponding torque loss T_L2 and the pinion bulk temperature ϑ₁. As the torque loss increased drastically during the experiments with the variants REF and SUF, LOL times were limited to 15 s to prevent damage to the measuring sensors at the test rig. The coated variants CO1, CO2, and CO3 enabled much lower torque loss and bulk temperature during LOL. Therefore, LOL times were increased to 30 s for CO1 and even 40 s for CO2 and CO3. As the nitriding steel Nitralloy N of the variant NIT allows for higher bulk temperatures than AISI 9310 of the other variants, the LOL time of NIT was also increased to 20 s.

Fig. 6

Torque loss T_L2 and pinion bulk temperature ϑ₁ (wheel bulk temperature ϑ₂ for SUF due to sensor damage) under LOL at LS7 and v_t = 20 m/s (the variant NIT reached the temperature sensor limit at 250 °C; data for variant REF from Morhard et al. 2024)

Under injection lubrication (t < 0 s), the measured torque losses were slightly lower compared to the moderate operating condition (cf. “Moderate operating condition”) as the higher circumferential speed reduces solid contacts as the film thickness increases. The differences between the variants were comparable to the moderate operating condition. The measured bulk temperatures were higher due to higher power losses with increased circumferential speed.

The material and surface variants showed different characteristics of torque loss and bulk temperature development during LOL. The variants REF and SUF showed a sudden and steep increase in torque loss after approx. 9 s of 15 s LOL time. The variant CO1 reached a plateau of torque loss after approx. 17 s of 30 s. The plateau was lower than the maxima of the variants REF and SUF. The variants CO2 and CO3 showed only a slight increase of torque loss, even after 40 s of LOL time, with the variant CO3 showing the lowest torque loss increase with approx. 0.5 Nm. The variant NIT featured a comparable behavior to the variants REF and SUF, even though the steep torque loss increase started approx. 2 s later.

The measured bulk temperatures followed the trend of the torque loss increase during LOL. Due to the absence of cooling oil during LOL, the bulk temperature increased continuously with LOL time. Due to a pinion sensor damage for the variant SUF, the wheel temperature ϑ₂ is depicted. Note that the wheel generally shows a slightly lower temperature than the pinion due to lower meshing frequency and higher thermal mass.

The variants REF, SUF, CO1, and NIT featured an increased post-LOL level of the measured torque loss and bulk temperature. According to Morhard et al. (Morhard et al. 2024), this correlates to surface damage and can be seen on the tooth flank surfaces in “Tooth flank surface analysis”.

Operating condition with increased load

The load increase from LS7 to LS9 at v_t = 8.3 m/s corresponds to an increase of P_t from 42 kW (LS7 and v_t = 8.3 m/s) to 68 kW (LS9 and v_t = 8.3 m/s). Figure 7 shows the corresponding torque loss T_L2 and the pinion bulk temperature ϑ₁. The LOL time was limited to 20 s. For representation, the first LOL experiments are depicted, as the variant CO1 faced coating damage during LOL.

Fig. 7

Torque loss T_L2 and pinion bulk temperature ϑ₁ (wheel bulk temperature ϑ₂ for CO2 due to sensor damage) under LOL at LS9 and v_t = 8.3 m/s (data for variant REF from Morhard et al. 2024)

Under injection lubrication (t < 0 s), the torque loss and bulk temperature were at a higher level compared to the moderate operating condition (cf. “Moderate operating condition”). This is due to increased load-dependent gear and bearing torque loss at higher loads. The differences between the variants were comparable to the moderate operating condition except for the variant CO1, which showed a lower torque loss than the variant CO2.

During LOL, the different material and surface variants showed comparable torque loss and bulk temperature characteristics for the operating condition with increased speed (cf. “Operating condition with increased speed”). Due to a pinion sensor damage of the variant CO2, the wheel temperature ϑ₂ is depicted (wheels generally show a slightly lower level than the pinion, cf. “Operating condition with increased speed”). Again, the torque loss of the variants REF, SUF, and NIT showed a sudden and steep increase: for the variant NIT after approx. 10 s and for REF and SUF after approx. 15 s LOL time. The variant SUF featured the steepest development of torque loss during LOL. The tooth flank damage was so severe (cf. “Tooth flank surface analysis”) that a second rise of torque loss occurred even after reactivated injection lubrication. This can be due to surface roughening during LOL time, initiating consequential damage after LOL. This behavior for variant SUF was reproducible. The coated variants CO1, CO2, and CO3 showed the lowest increase of torque loss during LOL, with the variant CO3 performing best. As found for the operating condition with increased speed, damage during LOL caused an increased post-LOL level of the measured torque loss for the variants REF, SUF, and NIT (cf. “Tooth flank surface analysis”).

Tooth flank surface analysis

Light microscope photographs were taken and analyzed to evaluate the tooth flank condition of the different variants after LOL experiments. Table 4 shows representative pinion tooth flank surfaces.

Table 4 Light microscope photographs of representative pinion tooth flank surfaces after LOL experiments (photographs of variant REF from (Morhard et al. 2024))

The left column in Table 4 refers to the pinion tooth flank surfaces after LOL at the moderate operating condition (cf. “Moderate operating condition”). All variants faced LOL thrice for 20 s, and the torque loss reached pre-LOL level after each LOL test. The light microscope photographs revealed only slight surface changes for the variants REF and NIT. The variant SUF featured slight visual changes as the surface was less reflective. The variants CO1, CO2, and CO3 showed a conspicuous appearance in the tooth root flank area. The variants CO1 and CO3 featured a limited horizontal wear streak, and CO2 featured a change in surface appearance over the complete tooth root flank area. It is assumed that the lower coating thickness towards the tooth root (cf. Table 2) led to initial coating wear in the tooth root flank area. The variant CO2 also showed an ellipsoidal coating surface change in the tooth tip flank area. Even though such changes and wear marks were visible, the torque loss level after LOL decreased to pre-LOL level. According to profile measurements, no significant profile deviations occurred in that area.

The middle column in Table 4 refers to the pinion tooth flank surfaces after LOL at the operating condition with increased speed (cf. “Operating condition with increased speed”). All variants faced LOL once for different LOL times. The post-LOL level of the measured torque loss for the variants REF, SUF, CO1, and NIT already indicated surface damages. The tooth flank surfaces confirmed this as they showed scuffing or coating damage. The severity of the damages correlates with the increased rate of torque loss (cf. Fig. 6). NIT gears showed scuffing only on some teeth, even though LOL time was 20 s and bulk temperatures exceeded 250 °C. Structural cuts of the pinion of the variant CO1 after 30 s LOL confirmed that the coating was not present anymore in the damaged areas. Profile measurements and structural cuts of the variant CO2 revealed that the coating was worn by approx. 3 μm in the tooth root flank area. The mating wheel showed a similar appearance so that a coating-metal contact remained. The variant CO2 featured again an ellipsoidal coating surface change in the tooth tip flank area. The variant CO3 only featured wear marks corresponding to those after 20 s LOL at the moderate operating condition.

The right column in Table 4 refers to the pinion tooth flank surfaces after LOL at the operating condition with increased load (cf. “Operating condition with increased load”). The variants REF, SUF, and NIT faced one time LOL, and the coated variants CO1, CO2, and CO3 faced three times LOL. Increasing the load led to similar trends as increasing the speed. The variants REF and NIT showed slight scuffing compared to the variant SUF with severe scuffing. The variant CO1 showed coating damage. The surface change in the tooth tip flank area of the variant CO2 was more expanded compared to the moderate operating condition and the operating condition with increased speed. The variant CO3 showed the least surface changes, even though coating wear marks in the tooth root flank area occurred.

Summary

Figure 8 shows the measured torque loss and pinion bulk temperature for all variants.

Fig. 8

Summary of measured torque loss T_L2 and bulk temperature ϑ₁ for all variants for a the moderate LOL operating condition, b the LOL operating condition with increased speed, and c the LOL operating condition with increased load (data for variant REF from Morhard et al. 2024)

For each variant, the quasi-stationary values under injection lubrication before LOL “pre-LOL”, the increase at the end of LOL time “Δt-LOL”, the maximum increase after LOL “Δmax-LOL”, and the quasi-stationary value under reactivated injection lubrication after LOL “post-LOL”. Table 5 summarizes the resulting tooth flank surface changes due to the LOL experiments.

Table 5 Summary of tooth flank surface changes after the LOL experiments for all variants for the moderate LOL operating condition (a), the LOL operating condition with increased speed (b), and the LOL operating condition with increased load (c)

The variant SUF showed a decreased loss of lubrication performance compared to the variant REF, which is only partly confirmed in the literature (cf. “Introduction”) and not fully understood. Kozachyn et al. (Kozachyn et al. 2018) refer to the reduced oil retention ability of superfinished surfaces due to a reduced amount of roughness valleys. The presented results support this hypothesis by measuring the steep increase in friction and the occurrence of profound scuffing. Also, the capacity of a superfinished surface to trap debris during LOL could be different compared to a ground surface. For example, Volchok et al. (Volchok et al. 2002) found that laser texturing can improve the fretting fatigue resistance as wear debris accumulates in laser pockets.

The variant CO1 had a low coating thickness and surface hardness, correlating to the lowest LOL performance within the coated variants and the observed coating damage. The variants CO2 and CO3 showed an improved LOL performance. The variant CO3 performed best within all operating conditions.

The variant NIT faced a strong torque loss and bulk temperature increase during LOL, which resulted in scuffing. However, the investigated nitriding steel improved the LOL behavior compared to the variants REF and SUF, as the scuffing damage was less pronounced and longer LOL times were feasible.

The coated variants CO1, CO2, and CO3 showed a reduced torque loss increase during LOL and longer LOL times compared to the other variants. Even though coating wear occurred in the tooth root flank area, the LOL performance seemed more or less unaffected. Coating-metal contact avoided scuffing as coating wear always occurred for both pinion and wheel in the tooth root flank area. A possible approach to avoid coating wear could be to coat only the tooth tip flank area. Investigations of Kalin and Vižintin (Kalin and Vižintin 2005) and Grossl (Grossl 2007) have shown that DLC-coated gears combined with steel gears reduce the wear of the coating with the drawback that the wear of the steel gears increases.

Conclusion

The presented study analyzed the influence of gear surface and material technologies on the loss of lubrication performance. Ground, superfinished, coated, and nitriding steel gears were compared for a moderate operating condition, an operating condition with increased speed of v_t = 20 m/s, and an operating condition with increased load of p_C = 1723 N/mm². The authors draw the following conclusions:

• ➣ The occurrence of scuffing limited the LOL performance of ground gears. Increased speed and load led to severe and slight scuffing.

• ➣ Superfinished gears showed severe scuffing at increased speed and load. The worse LOL performance compared to ground gears can be due to a reduced surface ability to retain oil and trap debris.

• ➣ DLC Coatings significantly improved the LOL behavior under the conditions tested. The torque loss and bulk temperature increase are limited, and fast damage progression is avoided. Coating wear was mainly found in the tooth root flank area, which preserved coating-metal contact and avoided scuffing.

• ➣ The best-performing coated variant (multilayer coating) withstood LOL without surface damage related to LOL for all conditions evaluated in this study.

• ➣ The torque loss behavior of nitrided gears was similar to ground gears, but the LOL performance improved, as LOL time could be slightly increased and less severe scuffing damage occurred. The reason for this might be the higher temperature stability.

This study shows the potential of DLC coatings to improve the LOL performance of gears. A previous study by the authors (Morhard et al. 2024) emphasized the significance of the availability of even small amounts of oil. Hence, considering coated gears and designing the gearbox housing to ensure the availability of remaining oil quantities at gears can significantly improve LOL performance and lead to simplified or omitted secondary oil supply systems. But this requires reliable durability of coatings on gears.

Nomenclature

b Face width mm

Ca Tip relief μm

C Pitch point -

CO1 Coated gear variant 1 -

CO2 Coated gear variant 2 -

CO3 Coated gear variant 3 -

d_a1 Tip diameter of pinion mm

d_a2 Tip diameter of wheel mm

H_IT Indentation hardness GPa

LOL Loss of lubrication -

LS Load stage -

L_t Measurement length mm

mod Modification -

NIT Nitrided gear variant -

n_1/2 Speed pinion/wheel shaft rpm

p_C Hertzian pressure at the pitch point N/mm²

P_t Transmitted power kW

Ra Arithmetic mean roughness µm

REF Reference gear variant -

SRV Schwing, Reibung und Verschleiß -

SUF Superfinished gear variant -

T_1/2 Torque pinion/wheel shaft Nm

T_L2 Measured torque loss Nm

v_t Circumferential speed m/s

WC/C Tungsten carbide/carbon -

x_1/2 Profile shift coefficient of pinion/wheel -

z_1/2 Number of pinion/wheel teeth

Greek symbols

α_n Pressure angle °

β Helix angle °

ε_α Transverse contact ratio -

ϑ Temperature °C

ϑ_Oil Oil injection temperature °C

ϑ_1/2 Bulk temperature pinion/wheel °C

ν Kinematic viscosity mm²/s

ρ Density kg/m³

λ_c Cut-off wavelength mm