- Email: [email protected]
Measurement of temperature at sliding polymer surface by grindable thermocouples
Tribology International 88 (2015) 100–106
Contents lists available at ScienceDirect
Tribology International journal homepage: www.elsevier.com/locate/triboint
Measurement of temperature at sliding polymer surface by grindable thermocouples Oleksii Nosko a,n, Takuo Nagamine b, A.L. Nosko c, A.M. Romashko c, Hiroki Mori b, Yuichi Sato b a
Department of Machine Design, KTH Royal Institute of Technology, Brinellvägen 83, Stockholm 100 44, Sweden Department of Mechanical Engineering, Saitama University, 255 Shimookubo, Sakura Ward, Saitama City 338 8570, Japan c Department of Lifting and Transport Systems, Bauman Moscow State Technical University, ul. Baumanskaya 2-ya, 5, Moscow 105005, Russia b
art ic l e i nf o
a b s t r a c t
Article history: Received 30 September 2014 Received in revised form 19 February 2015 Accepted 9 March 2015 Available online 18 March 2015
This paper is devoted to experimental study of capabilities and limitations of grindable thermocouples as applied to polymer materials sliding on metal. Chromel–alumel and chromel–copel grindable thermocouples have been developed and tested for wide ranges of contact pressure and sliding velocity. The background temperature of the sliding surface can be determined as the lower envelope of the signal from the grindable thermocouple. Steady and unsteady regimes of sliding have been investigated. For steady sliding, the accuracy of the temperature determination increases with measurement duration. In the case of unsteady sliding, accurate temperature determination requires multiple tests under the same conditions. The thickness of the thermocouple junction has been analyzed for correct comparison of experimental and calculated temperatures. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Sliding Temperature Thermocouple Signal processing
1. Introduction It is well-known that temperature affects frictional behavior and failure of sliding components. Especially noticeable is its influence on properties of polymer materials which are widely used in brakes, clutches and sliding bearings. For example, the friction coefficient and wear rate of polymer/metal pair can change by several times within the range of working temperatures [1]. Therefore, temperature should be taken into account when predicting performance characteristics of polymer/metal friction pairs. Generally, temperature is distributed in the sliding component non-uniformly. It reaches its maximum at the sliding surface and decreases with the distance. Polymers have low thermal conductivity, which results in a large temperature gradient at the sliding surface. For highly-loaded friction pairs, this gradient attains a magnitude of 0.1–1 1C/mm [2]. Consequently, for adequate description of thermal effects it is necessary to know the temperature of the surface layer not thicker than several tens of micrometers. Different techniques have been developed to measure temperature of sliding surfaces [3]. The most commonly used methods are the infrared radiation technique and thermocouple technique. The infrared radiation technique is based on that the radiative power depends on temperature. Infrared detector is focused on the contact
n
Corresponding author. E-mail address: [email protected] (O. Nosko).
http://dx.doi.org/10.1016/j.triboint.2015.03.015 0301-679X/& 2015 Elsevier Ltd. All rights reserved.
area through a transparent component [4] or at a surface close to the contact area [5]. The technique enables recording temperature with high sampling rate and spatial sensitivity. However, the requirement for at least one transparent sliding component is not practical in most cases. The thermocouple technique is based on the Seebeck effect, i.e., the direct conversion of temperature to electric voltage. The common approach is installing a thermocouple in the stationary component as close as possible to the sliding surface. The smaller is the thermocouple junction, the higher is the measurement accuracy. Various miniature thermocouples have been developed. The smallest of them would probably be a thin film thermocouple with a junction of the order of 1 μm in the direction perpendicular to the sliding surface [6,7]. Miniature thermocouples provide reliable measurements, but their service life is short due to the microscopic distance between the junction and wearable sliding surface. Another approach within the thermocouple technique is to use grindable thermocouples [8,9], also called contact thermocouples [3] or tape thermocouples [10,11]. Double-pole grindable thermocouple consists of two separate insulated wires embedded in the component with their ends exposed at the sliding surface. The frictional deformation and heating join the wires together into a junction. Thereby, the grindable thermocouple provides measurements under intensive wear conditions. In the case of a metallic component, the construction of the grindable thermocouple is simplified by eliminating one wire and using the component instead of it (single-pole grindable thermocouple [12]).
O. Nosko et al. / Tribology International 88 (2015) 100–106
Grindable thermocouples make a good alternative to infrared detectors and miniature thermocouples when applied to metallic friction pairs. They have been successfully used for measuring temperature of ground workpieces [9,12–16]. On the other hand, their application to non-metallic materials is considerably limited [17,18]. One of the main reasons for this is the dissimilarity in properties of non-metallic materials and metals. It has been testified [19] that the measurement error of the grindable thermocouple is dependent on the difference between the thermophysical properties of the sliding component and thermocouple wire; the more essential is the difference, the larger is the error. Moreover, the friction of the thermocouple wire on a counterbody differs qualitatively from the friction of non-metallic material on the counterbody. The involvement of the thermocouple in friction may cause distortion of the temperature field in the vicinity of the thermocouple junction. Thus, the problem of interpretation of the temperature signal from the grindable thermocouple should be resolved to expand the application of grindable thermocouples to non-metallic materials. This paper provides an experimental study of characteristics of grindable thermocouples as applied to polymer materials. A special attention is paid to the issues of interpretation of the temperature signal and estimation of the measurement accuracy. The applicability of grindable thermocouples to temperature measurements is investigated for steady and unsteady sliding conditions.
101
2.2. Experimental set-up Fig. 3 shows a schematic of a friction machine developed. The sample is installed into a holder which is, in its turn, attached to a lever by use of a beam. The holder and beam are carefully leveled. The lever can rotate about a vertical axis. Friction disc is machined with high precision. Its friction surface is ground to eliminate waviness and distortion. It is mounted on a horizontally installed shaft together with an inertia disc. The shaft is driven by a rotor. By applying a force to the lever, the sample is pressed against the disc. The adjustable parameters of friction are the contact pressure p and linear velocity v of the disc at the average friction radius r¼35 mm. Two beams with different stiffnesses are used to support the sample. The stiff beam (cross section 20 4 mm2) allows conducting tests in the regimes of steady sliding and deceleration. On the other hand, friction-induced tangential oscillation of the sample is possible when using the pliable beam (cross section 16 2 mm2). The tangential displacement x of the sample is measured by a laser sensor with a resolution power of 0.2 μm. The laser spot, 70 mm in diameter, is focused on the polished upper face of the holder. The signals from the grindable thermocouple and laser sensor are processed by a data logger with a sampling rate fs of up to 100 kHz. The data logger provides two modes of signal processing: no filter; low-pass filter with a cut-off frequency of 500 Hz. 2.3. Friction materials
2. Experimental technique 2.1. Grindable thermocouple tape arrangement Two types of grindable thermocouples based on chromel– alumel and chromel–copel pairs are used in the experiments. Grindable thermocouple is manufactured from two wires of a small diameter. One end of each wire is flattened into a thin tape. The thickness h of the tapes is equal to 60 mm for chromel–alumel and 20 mm for chromel–copel. The tapes are trimmed, coated with thin insulation layer and sandwiched together, forming the tape arrangement, as shown in Fig. 1. The insulation material is polyimide for chromel–alumel and mica for chromel–copel. It is essential that the tapes are completely insulated from each other. Friction sample represents a cube with a side of 1 cm. It is split into two equal pieces. The grooves are made accurately in the pieces, as depicted in Fig. 2. The tape arrangement is placed into the grooves. The cyanoacrylate adhesive is applied to the pieces (dotted areas in Fig. 2) and then the pieces are glued together under pressure. As a result, the tape arrangement is securely clamped in the grooves between the two pieces and its tip lies at the sample surface.
Two polymer friction materials denoted as SFP04 and 145-40 are tested. SFP04 is used as brake pads in motor vehicles. It
Fig. 2. Installation of tape arrangement into friction sample.
Fig. 1. Chromel–alumel thermocouple tape arrangement.
102
O. Nosko et al. / Tribology International 88 (2015) 100–106
Fig. 5. Single microcontact between tapes of chromel–copel thermocouple.
fluctuations of the signal occur [14]. The following results are obtained under stable electrical contact.
3. Thermocouple calibration Fig. 3. Schematic of friction machine.
Grindable thermocouple is calibrated using a furnace with controllable temperature. It is placed in the furnace together with a well-calibrated conventional thermocouple made from the same pair of alloys. The conventional thermocouple indicates the reference temperature. The temperature in the furnace increases stepby-step and at each step the steady signal from the grindable thermocouple is recorded. For example, Fig. 7 shows the steady signals from chromel–alumel thermocouple.
4. Results and discussion 4.1. Background temperature of sliding surface Fig. 4. Junction of chromel–copel thermocouple.
comprises barite, zirconium oxide, mica, phenol resin, copper, etc. 145-40 is used as friction components in brakes and clutches of motor vehicles and lift-and-transport machines [20]. Its composition is phenol resin, asbestos, barium hydroxide, copper, alumina, etc. The friction disc is steel S275.
2.4. Formation of thermocouple junction When the friction disc slides on the sample, the tape arrangement is also involved in friction. Under the action of thermomechanical effects, the tapes are smeared over the insulation layer and come into contact with each other, creating a thermocouple junction. A typical micrograph of the thermocouple tip taken after testing the sample is presented in Fig. 4. One can see the deformed tapes and junction which includes 3 visible microcontacts. Fig. 5 shows an enlarged micrograph of a single microcontact. A typical evolution of the signal Ts generated by the grindable thermocouple is shown in Fig. 6. Ts includes numerous temperature peaks caused by the interactions of asperities of the thermocouple tip with the disc [16]. If the electrical contact between the tapes due to the microcontacts is unstable, this affects adversely the quality of Ts: the level of noise increases and random
SFP04 sample with chromel–alumel thermocouple is rubbed until the electrical contact between the tapes becomes stable. Then the sample is carefully disassembled. The thermocouple is shifted in the sample so that there is a gap of about 100 mm between its tip and the sliding surface, as illustrated in Fig. 8. The sample is assembled and tested under constant sliding velocity. In the course of the test, the friction becomes stationary, which implies the uniform contact between the sample and disc, steady temperature fields in the sample and disc, uniform decrease of the gap due to the wearing of the sample. The thermocouple uninvolved in friction operates as a conventional thermocouple. At a certain instant t ¼ t s the gap disappears completely and the thermocouple starts sliding on the disc. At t 4 t s the thermocouple is involved in friction. Fig. 9 illustrates a typical evolution of Ts. The interval t o t s corresponds to the stationary friction. On this interval, Ts has slight fluctuations which are most probably related to the interactions between the sample and disc in the vicinity of the thermocouple junction. At t ¼ t s , the junction lies directly at the sliding surface, but the temperature field in the thermocouple tip and sample has not been affected by the friction of the thermocouple on the disc yet. Consequently, at this instant Ts is equal to the rise Tb of the background temperature of the sample sliding surface. At t 4 t s , Ts behaves unpredictably and overestimates Tb in general case. However, it is apparent that the lower bound of Ts coincides with Tb.
O. Nosko et al. / Tribology International 88 (2015) 100–106
103
Fig. 6. Evolution of Ts: 145-40, chromel–copel, fs ¼100 kHz, no filter, p ¼ 0.3 MPa, v¼ 6.6 m/s.
Fig. 7. Calibration of chromel–alumel thermocouple.
measurement accuracy. For this purpose, we use a technique [16] that allows determining the lower envelope of the signal. As an example, consider a short fragment of Ts depicted in Fig. 11. Find the local minimum points on Ts. A local minimum is defined by the condition that the relevant value of Ts is lower than those at the previous and following time steps. Determine the lower envelope on the basis of the local minimum points (see line “n¼1” in Fig. 11). The procedure is applied n times. At each iteration, the part of the signal related to the temperature peaks is deleted. Introduce into consideration the maximum relative deviation ε of Ts from Tb and the average period P between the nearest points of Ts. Fig. 12 shows the dependencies εðnÞ and P ðnÞ for the fragment of Fig. 11. In general case, under the transition from the iteration n to ðn þ 1Þ, ε decreases, while P increases. SFP04 samples with chromel–alumel thermocouples and 145-40 samples with chromel–copel thermocouples are tested under different friction conditions. In each test, the evolution of Ts is obtained and presented similarly to that in Fig. 9. The values of ts and Tb are determined. A fragment of Ts which corresponds to the interval t 4 t s and includes 107 points is analyzed. By applying the signal processing technique above, the relation between ε and P is found. The results can be seen in Fig. 13. They show that for P of up to 10 s we have ε 4 0:1, i.e., Tb is overestimated by above 10%. When P is of about 100 s, the overestimate of Tb does not exceed 5%. Thus, for constant or slowly changing friction conditions, Tb can be determined with sufficient accuracy as the lower envelope of Ts. 4.3. Measurement of temperature under unsteady sliding
Fig. 8. Gap between thermocouple tip and sliding surface.
The fragments of the experimental data with T s T b are thoroughly analyzed. When Ts approaches Tb, the occurrence frequency and magnitude of the temperature peaks decrease noticeably. In some cases the peaks are not observed at all, as shown in Fig. 10, which may imply the intermittent contact between the thermocouple and disc. 4.2. Measurement of temperature under steady sliding Appropriate application of grindable thermocouples requires a special processing of the temperature signal with estimation of the
The applicability of grindable thermocouples for unsteady sliding is investigated on an example of decelerative sliding typical for brakes. 145-40 sample with chromel–copel thermocouple installed is tested with no gap between the thermocouple tip and sliding surface. The test procedure includes two stages: (1) the disc is accelerated by the rotor to a linear velocity v0 at the radius r; (2) the rotor is switched off and the sample is pressed against the rotating disc; due to the friction the disc velocity decreases to zero during t 0 seconds. The test is repeated N times under the same conditions, i.e., the fixed values of p and v0 . Fig. 14a shows an experimental data for a single test. Ts changes in a wide range due to the temperature peaks. A continuous thick line depicts the lower envelope of Ts for n¼ 4. Consider a signal created by superimposition of N temperature signals. The superimposed signals and their lower envelopes for N¼5, N¼10 and N¼20 are presented in Fig. 14b, c and d respectively. With increase of N, the lower envelope of the superimposed signal shifts down and approaches a limit curve depicted with a dashed line. The tests show that at N420
104
O. Nosko et al. / Tribology International 88 (2015) 100–106
Fig. 9. Relation between Ts and Tb: SFP04, chromel–alumel, fs ¼ 1 kHz, no filter, p¼ 0.5 MPa, v¼ 6.8 m/s.
Fig. 10. Evolution of Ts: 145-40, chromel–copel, fs ¼1 kHz, 500 Hz filter, p ¼ 0.3 MPa, v¼ 6.6 m/s.
Fig. 12. Dependencies εðnÞ and P ðnÞ.
Fig. 11. Determination of lower envelope of temperature signal.
new temperature peaks appear on the superimposed signal, but the changes in the lower envelope are negligibly small. According to the results obtained in Section 4.1, the limit curve coincides with Tb. Consequently, in the case of unsteady sliding, the
Fig. 13. Relationship between ε and P, fs ¼ 1 kHz: 145-40, chromel–copel, 500 Hz filter, p ¼0.2 MPa, v¼4.4 m/s; ■ 145-40, chromel–copel, 500 Hz filter, p ¼0.3 MPa, v¼6.6 m/s; ▲ SFP04, chromel–alumel, no filter, p ¼0.2 MPa, v¼ 7.7 m/s; ◆SFP04, chromel–alumel, 500 Hz filter, p ¼0.4 MPa, v¼ 6.6 m/s; ▼ SFP04, chromel–alumel, no filter, p ¼ 0.6 MPa, v¼ 7.7 m/s.
O. Nosko et al. / Tribology International 88 (2015) 100–106
105
Fig. 14. Determination of Tb for decelerative sliding: 145-40, chromel–copel, fs ¼ 1 kHz, no filter, p ¼ 1 MPa, v0 ¼ 6.6 m/s, t0 ¼ 13.4 70.2 s.
determination of Tb requires identifying the lower envelope of the superimposed signal.
Fourier transform show that φ takes a value between 0.9π and 1.07π at p ¼0.1–0.3 MPa and v ¼1.1–6.6 m/s. Suppose that a body oscillates on a counterbody, moving with a velocity v as follows:
4.4. Junction thickness x ¼ A sin ð2πf t Þ For correct analysis of the temperature signal, it is necessary to know the thickness of the thermocouple junction in the direction perpendicular to the sliding surface. This thickness allows to estimate at what distance from the sliding surface the temperature is measured. In addition, it can be useful when calculating the thermocouple time constant [15]. The pair SFP04/steel S275 has a negative friction–velocity slope [21]. When the pliable beam is used to support the holder, frictioninduced tangential oscillation of the sample occurs [1]. The displacement x changes harmonically with the frequency f ¼126 Hz and amplitude A of the order of 100 mm. The signal Ts from chromel–alumel thermocouple, involved in friction, oscillates almost harmonically with the frequency f and amplitude of several degrees Celsius. An example of the harmonic behavior of x and Ts is presented in Fig. 15. It is seen that there is a phase shift φ between x and Ts. The calculations done by use of the discrete
ð1Þ
If the positive direction of x coincides with the direction of the counterbody motion and dx=dt o v at any instant t 4 0, then the sliding velocity vs is defined as follows: vs ¼ v dx=dt ¼ v 2πf A cos ð2πf t Þ The friction heat is generated with specific power q ¼ μpvs ¼ μpv 2πf Aμp cos ð2πf t Þ where μ is friction coefficient. The oscillatory term of q, i.e., 2πf Aμp cos ð2πf t Þ, leads to the periodic temperature rise T in the body, which decays in amplitude with distance z from the sliding surface [22] pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffi! pffiffiffiffiffi! αAμp 2πf k z πf 3π z πf T¼ exp pffiffiffi sin 2πf t pffiffiffi ð2Þ 4 K k k
106
O. Nosko et al. / Tribology International 88 (2015) 100–106
of the temperature determination increases with measurement duration. An accuracy of 5% is attained when processing the signal for an interval of the order of 100 s. 3. For unsteady sliding, multiple tests under the same conditions and combination of the temperature signals into a superimposed signal are required. With increase of the number of the tests, the lower envelope of the superimposed signal approaches Tb. The temperature signals from about 20 tests allow for accurate determination of Tb. 4. The thickness of the thermocouple junction can be calculated on the basis of the phase shift between the friction sample tangential displacement and temperature signal in oscillatory sliding.
Acknowledgements Fig. 15. Harmonic oscillations of x and Ts: SFP04, chromel–alumel, fs ¼2 kHz, no filter, p ¼0.1 MPa, v¼ 2.2 m/s, x and Ts are reckoned from their average values.
where k is thermal diffusivity coefficient (k ¼4.9 mm2/s for chromel tape smeared over the insulation layer); K is thermal conductivity coefficient; α is heat partition coefficient. From the Eqs.(1) and (2) we derive pffiffiffiffiffi 3π z πf φ ¼ þ pffiffiffi 4 k or
pffiffiffi k 3π z ¼ pffiffiffiffiffi φ 4 πf
ð3Þ
According to the Eq. (3), the phase shift φ¼0.9π–1.07π obtained on the basis of the experimental data corresponds to the distance z¼ 52–110 mm which can serve as an estimate for the junction thickness. Calculate the time constant τ of the chromel–alumel thermocouple. According to the heat conduction theory [22], if a surface of a body is instantaneously heated by T 0 degrees at t ¼ 0 and its temperature is then maintained constant, the temperature rise T in the body is governed by the following equation: z ð4Þ T ¼ T 0 erfc pffiffiffiffiffi 2 kt where erfcð U Þ is complementary error function. If we suppose that T at the distance z reaches 63% of T 0 at the instant t ¼ τ, then due to the Eq. (4) for z¼ 52–110 mm we have τ¼1.2–5.4 ms. This is in a good agreement with the estimate τ¼3.4 ms reported in [9] for chromel– constantan grindable thermocouple with hE150 μm and mica insulation. 5. Conclusions The present study demonstrates the capabilities and limitations of grindable thermocouples as applied to polymer materials sliding on metal. On the basis of thorough analysis of the temperature signal Ts from the grindable thermocouple, the following results have been obtained. 1. Generally, Ts overestimates the sliding surface background temperature Tb due to the friction of the thermocouple on the disc. The lower bound of Ts coincides with Tb. 2. In the case of steady sliding, Tb can be determined as the lower envelope of Ts. This envelope is identified by finding the local minimum points of the temperature signal and removing the temperature peaks associated with these points. The accuracy
The authors wish to thank Mr. Hisao Miura at Tokai Material Co., Ltd. for supplying friction materials and Dr. Hideo Shibutani at Saitama University for his assistance in obtaining thermocouple micrographs. References [1] Kragelskii IV. Friction and wear. London: Butterworth; 1965. [2] Newcomb TP. Transient temperatures in brake drums and linings. Proc Inst Mech Eng Automob Div 1958;12:227–44. [3] Kennedy FE. Thermal and thermomechanical effects in dry sliding. Wear 1984;100:453–76. [4] Bowden FP, Thomas PM. The surface temperature of sliding solids. Proc R Soc Lond Ser A 1954;223:29–40. [5] Floquet A, Play D. Contact temperatures in dry bearings: three dimensional theory and verification. J Lubr Technol 1981;103(2):243–52. [6] Tian X, Kennedy FE, Deacutis JJ, Henning AK. The development and use of thin film thermocouples for contact temperature measurement. Tribol Trans 1992;35(3):491–9. [7] Kennedy FE, Frusescu D, Li J. Thin film thermocouples arrays for sliding surface temperature measurement. Wear 1997;207:46–54. [8] Peklenik J. Ermittlung von geometrischen und physikalischen Kenngrössen für die Grundlagenforschung des Schleifens. [dissertation]. Aachen: TH Aachen; 1957 in German. [9] Nee AYC, Tay AO. On the measurement of surface grinding temperature. Int J Mach Tool Des Res 1981;21:279–91. [10] Nosko AL, Kimura Y, Mizuguchi S. Measurement of surface temperature of sliding materials under sparse lubrication. In: Proceedings of the Japan international tribology conference. Nagoya, 1993. [11] Nosko AL, Kimura Y. Surface temperature measurement with tape-type thermocouples. In: Proceedings of the 6th nordic symposium on tribology. Uppsala, 1994. [12] Rowe WB, Black SCE, Mills B, Qi HS, Morgan MN. Experimental investigation of heat transfer in grinding. CIRP Ann 1995;44(1):329–32. [13] Babic D, Murray DB, Torrance AA. Mist jet cooling of grinding processes. Int J Mach Tool Manuf 2005;45:1171–7. [14] Batako AD, Rowe WB, Morgan MN. Temperature measurement in high efficiency deep grinding. Int J Mach Tool Manuf 2005;45:1231–45. [15] Lefebvre A, Vieville P, Lipinski P, Lescalier C. Numerical analysis of grinding temperature measurement by the foil/workpiece thermocouple method. Int J Mach Tool Manuf 2006;46:1716–26. [16] Lefebvre A, Lanzetta F, Lipinski P, Torrance AA. Measurement of grinding temperatures using a foil/workpiece thermocouple. Int J Mach Tool Manuf 2012;58:1–10. [17] Romashko AM. Study of heating of lift-and-transport machines disk brakes. [dissertation]. Moscow: Bauman Moscow Higher Technical School; 1979 in Russian. [18] Nosko AL. Development of method to calculate heating of lift-and-transport machines disk brakes with account for thermal contact resistance. [dissertation]. Moscow: Bauman Moscow Higher Technical School; 1985 in Russian. [19] Lebedev VG, Kochetkov AM, Zhabokritskiy RA. Assessing ground-surface contact temperature using grindable thermocouples. Sov J Mach Tools 1973;9:29. [20] Chichinadze AV. Polymers in friction assemblies of machines and devices. New York: Allerton Press; 1984. [21] Nosko O. Effect of temperature on dynamic characteristics of a pad sliding on a disc. [dissertation]. Saitama: Saitama University; 2013. [22] Carslaw HS, Jaeger JC. Conduction of heat in solids. 2nd ed.. Oxford: Clarendon Press; 1959.
Contents lists available at ScienceDirect
Tribology International journal homepage: www.elsevier.com/locate/triboint
Measurement of temperature at sliding polymer surface by grindable thermocouples Oleksii Nosko a,n, Takuo Nagamine b, A.L. Nosko c, A.M. Romashko c, Hiroki Mori b, Yuichi Sato b a
Department of Machine Design, KTH Royal Institute of Technology, Brinellvägen 83, Stockholm 100 44, Sweden Department of Mechanical Engineering, Saitama University, 255 Shimookubo, Sakura Ward, Saitama City 338 8570, Japan c Department of Lifting and Transport Systems, Bauman Moscow State Technical University, ul. Baumanskaya 2-ya, 5, Moscow 105005, Russia b
art ic l e i nf o
a b s t r a c t
Article history: Received 30 September 2014 Received in revised form 19 February 2015 Accepted 9 March 2015 Available online 18 March 2015
This paper is devoted to experimental study of capabilities and limitations of grindable thermocouples as applied to polymer materials sliding on metal. Chromel–alumel and chromel–copel grindable thermocouples have been developed and tested for wide ranges of contact pressure and sliding velocity. The background temperature of the sliding surface can be determined as the lower envelope of the signal from the grindable thermocouple. Steady and unsteady regimes of sliding have been investigated. For steady sliding, the accuracy of the temperature determination increases with measurement duration. In the case of unsteady sliding, accurate temperature determination requires multiple tests under the same conditions. The thickness of the thermocouple junction has been analyzed for correct comparison of experimental and calculated temperatures. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Sliding Temperature Thermocouple Signal processing
1. Introduction It is well-known that temperature affects frictional behavior and failure of sliding components. Especially noticeable is its influence on properties of polymer materials which are widely used in brakes, clutches and sliding bearings. For example, the friction coefficient and wear rate of polymer/metal pair can change by several times within the range of working temperatures [1]. Therefore, temperature should be taken into account when predicting performance characteristics of polymer/metal friction pairs. Generally, temperature is distributed in the sliding component non-uniformly. It reaches its maximum at the sliding surface and decreases with the distance. Polymers have low thermal conductivity, which results in a large temperature gradient at the sliding surface. For highly-loaded friction pairs, this gradient attains a magnitude of 0.1–1 1C/mm [2]. Consequently, for adequate description of thermal effects it is necessary to know the temperature of the surface layer not thicker than several tens of micrometers. Different techniques have been developed to measure temperature of sliding surfaces [3]. The most commonly used methods are the infrared radiation technique and thermocouple technique. The infrared radiation technique is based on that the radiative power depends on temperature. Infrared detector is focused on the contact
n
Corresponding author. E-mail address: [email protected] (O. Nosko).
http://dx.doi.org/10.1016/j.triboint.2015.03.015 0301-679X/& 2015 Elsevier Ltd. All rights reserved.
area through a transparent component [4] or at a surface close to the contact area [5]. The technique enables recording temperature with high sampling rate and spatial sensitivity. However, the requirement for at least one transparent sliding component is not practical in most cases. The thermocouple technique is based on the Seebeck effect, i.e., the direct conversion of temperature to electric voltage. The common approach is installing a thermocouple in the stationary component as close as possible to the sliding surface. The smaller is the thermocouple junction, the higher is the measurement accuracy. Various miniature thermocouples have been developed. The smallest of them would probably be a thin film thermocouple with a junction of the order of 1 μm in the direction perpendicular to the sliding surface [6,7]. Miniature thermocouples provide reliable measurements, but their service life is short due to the microscopic distance between the junction and wearable sliding surface. Another approach within the thermocouple technique is to use grindable thermocouples [8,9], also called contact thermocouples [3] or tape thermocouples [10,11]. Double-pole grindable thermocouple consists of two separate insulated wires embedded in the component with their ends exposed at the sliding surface. The frictional deformation and heating join the wires together into a junction. Thereby, the grindable thermocouple provides measurements under intensive wear conditions. In the case of a metallic component, the construction of the grindable thermocouple is simplified by eliminating one wire and using the component instead of it (single-pole grindable thermocouple [12]).
O. Nosko et al. / Tribology International 88 (2015) 100–106
Grindable thermocouples make a good alternative to infrared detectors and miniature thermocouples when applied to metallic friction pairs. They have been successfully used for measuring temperature of ground workpieces [9,12–16]. On the other hand, their application to non-metallic materials is considerably limited [17,18]. One of the main reasons for this is the dissimilarity in properties of non-metallic materials and metals. It has been testified [19] that the measurement error of the grindable thermocouple is dependent on the difference between the thermophysical properties of the sliding component and thermocouple wire; the more essential is the difference, the larger is the error. Moreover, the friction of the thermocouple wire on a counterbody differs qualitatively from the friction of non-metallic material on the counterbody. The involvement of the thermocouple in friction may cause distortion of the temperature field in the vicinity of the thermocouple junction. Thus, the problem of interpretation of the temperature signal from the grindable thermocouple should be resolved to expand the application of grindable thermocouples to non-metallic materials. This paper provides an experimental study of characteristics of grindable thermocouples as applied to polymer materials. A special attention is paid to the issues of interpretation of the temperature signal and estimation of the measurement accuracy. The applicability of grindable thermocouples to temperature measurements is investigated for steady and unsteady sliding conditions.
101
2.2. Experimental set-up Fig. 3 shows a schematic of a friction machine developed. The sample is installed into a holder which is, in its turn, attached to a lever by use of a beam. The holder and beam are carefully leveled. The lever can rotate about a vertical axis. Friction disc is machined with high precision. Its friction surface is ground to eliminate waviness and distortion. It is mounted on a horizontally installed shaft together with an inertia disc. The shaft is driven by a rotor. By applying a force to the lever, the sample is pressed against the disc. The adjustable parameters of friction are the contact pressure p and linear velocity v of the disc at the average friction radius r¼35 mm. Two beams with different stiffnesses are used to support the sample. The stiff beam (cross section 20 4 mm2) allows conducting tests in the regimes of steady sliding and deceleration. On the other hand, friction-induced tangential oscillation of the sample is possible when using the pliable beam (cross section 16 2 mm2). The tangential displacement x of the sample is measured by a laser sensor with a resolution power of 0.2 μm. The laser spot, 70 mm in diameter, is focused on the polished upper face of the holder. The signals from the grindable thermocouple and laser sensor are processed by a data logger with a sampling rate fs of up to 100 kHz. The data logger provides two modes of signal processing: no filter; low-pass filter with a cut-off frequency of 500 Hz. 2.3. Friction materials
2. Experimental technique 2.1. Grindable thermocouple tape arrangement Two types of grindable thermocouples based on chromel– alumel and chromel–copel pairs are used in the experiments. Grindable thermocouple is manufactured from two wires of a small diameter. One end of each wire is flattened into a thin tape. The thickness h of the tapes is equal to 60 mm for chromel–alumel and 20 mm for chromel–copel. The tapes are trimmed, coated with thin insulation layer and sandwiched together, forming the tape arrangement, as shown in Fig. 1. The insulation material is polyimide for chromel–alumel and mica for chromel–copel. It is essential that the tapes are completely insulated from each other. Friction sample represents a cube with a side of 1 cm. It is split into two equal pieces. The grooves are made accurately in the pieces, as depicted in Fig. 2. The tape arrangement is placed into the grooves. The cyanoacrylate adhesive is applied to the pieces (dotted areas in Fig. 2) and then the pieces are glued together under pressure. As a result, the tape arrangement is securely clamped in the grooves between the two pieces and its tip lies at the sample surface.
Two polymer friction materials denoted as SFP04 and 145-40 are tested. SFP04 is used as brake pads in motor vehicles. It
Fig. 2. Installation of tape arrangement into friction sample.
Fig. 1. Chromel–alumel thermocouple tape arrangement.
102
O. Nosko et al. / Tribology International 88 (2015) 100–106
Fig. 5. Single microcontact between tapes of chromel–copel thermocouple.
fluctuations of the signal occur [14]. The following results are obtained under stable electrical contact.
3. Thermocouple calibration Fig. 3. Schematic of friction machine.
Grindable thermocouple is calibrated using a furnace with controllable temperature. It is placed in the furnace together with a well-calibrated conventional thermocouple made from the same pair of alloys. The conventional thermocouple indicates the reference temperature. The temperature in the furnace increases stepby-step and at each step the steady signal from the grindable thermocouple is recorded. For example, Fig. 7 shows the steady signals from chromel–alumel thermocouple.
4. Results and discussion 4.1. Background temperature of sliding surface Fig. 4. Junction of chromel–copel thermocouple.
comprises barite, zirconium oxide, mica, phenol resin, copper, etc. 145-40 is used as friction components in brakes and clutches of motor vehicles and lift-and-transport machines [20]. Its composition is phenol resin, asbestos, barium hydroxide, copper, alumina, etc. The friction disc is steel S275.
2.4. Formation of thermocouple junction When the friction disc slides on the sample, the tape arrangement is also involved in friction. Under the action of thermomechanical effects, the tapes are smeared over the insulation layer and come into contact with each other, creating a thermocouple junction. A typical micrograph of the thermocouple tip taken after testing the sample is presented in Fig. 4. One can see the deformed tapes and junction which includes 3 visible microcontacts. Fig. 5 shows an enlarged micrograph of a single microcontact. A typical evolution of the signal Ts generated by the grindable thermocouple is shown in Fig. 6. Ts includes numerous temperature peaks caused by the interactions of asperities of the thermocouple tip with the disc [16]. If the electrical contact between the tapes due to the microcontacts is unstable, this affects adversely the quality of Ts: the level of noise increases and random
SFP04 sample with chromel–alumel thermocouple is rubbed until the electrical contact between the tapes becomes stable. Then the sample is carefully disassembled. The thermocouple is shifted in the sample so that there is a gap of about 100 mm between its tip and the sliding surface, as illustrated in Fig. 8. The sample is assembled and tested under constant sliding velocity. In the course of the test, the friction becomes stationary, which implies the uniform contact between the sample and disc, steady temperature fields in the sample and disc, uniform decrease of the gap due to the wearing of the sample. The thermocouple uninvolved in friction operates as a conventional thermocouple. At a certain instant t ¼ t s the gap disappears completely and the thermocouple starts sliding on the disc. At t 4 t s the thermocouple is involved in friction. Fig. 9 illustrates a typical evolution of Ts. The interval t o t s corresponds to the stationary friction. On this interval, Ts has slight fluctuations which are most probably related to the interactions between the sample and disc in the vicinity of the thermocouple junction. At t ¼ t s , the junction lies directly at the sliding surface, but the temperature field in the thermocouple tip and sample has not been affected by the friction of the thermocouple on the disc yet. Consequently, at this instant Ts is equal to the rise Tb of the background temperature of the sample sliding surface. At t 4 t s , Ts behaves unpredictably and overestimates Tb in general case. However, it is apparent that the lower bound of Ts coincides with Tb.
O. Nosko et al. / Tribology International 88 (2015) 100–106
103
Fig. 6. Evolution of Ts: 145-40, chromel–copel, fs ¼100 kHz, no filter, p ¼ 0.3 MPa, v¼ 6.6 m/s.
Fig. 7. Calibration of chromel–alumel thermocouple.
measurement accuracy. For this purpose, we use a technique [16] that allows determining the lower envelope of the signal. As an example, consider a short fragment of Ts depicted in Fig. 11. Find the local minimum points on Ts. A local minimum is defined by the condition that the relevant value of Ts is lower than those at the previous and following time steps. Determine the lower envelope on the basis of the local minimum points (see line “n¼1” in Fig. 11). The procedure is applied n times. At each iteration, the part of the signal related to the temperature peaks is deleted. Introduce into consideration the maximum relative deviation ε of Ts from Tb and the average period P between the nearest points of Ts. Fig. 12 shows the dependencies εðnÞ and P ðnÞ for the fragment of Fig. 11. In general case, under the transition from the iteration n to ðn þ 1Þ, ε decreases, while P increases. SFP04 samples with chromel–alumel thermocouples and 145-40 samples with chromel–copel thermocouples are tested under different friction conditions. In each test, the evolution of Ts is obtained and presented similarly to that in Fig. 9. The values of ts and Tb are determined. A fragment of Ts which corresponds to the interval t 4 t s and includes 107 points is analyzed. By applying the signal processing technique above, the relation between ε and P is found. The results can be seen in Fig. 13. They show that for P of up to 10 s we have ε 4 0:1, i.e., Tb is overestimated by above 10%. When P is of about 100 s, the overestimate of Tb does not exceed 5%. Thus, for constant or slowly changing friction conditions, Tb can be determined with sufficient accuracy as the lower envelope of Ts. 4.3. Measurement of temperature under unsteady sliding
Fig. 8. Gap between thermocouple tip and sliding surface.
The fragments of the experimental data with T s T b are thoroughly analyzed. When Ts approaches Tb, the occurrence frequency and magnitude of the temperature peaks decrease noticeably. In some cases the peaks are not observed at all, as shown in Fig. 10, which may imply the intermittent contact between the thermocouple and disc. 4.2. Measurement of temperature under steady sliding Appropriate application of grindable thermocouples requires a special processing of the temperature signal with estimation of the
The applicability of grindable thermocouples for unsteady sliding is investigated on an example of decelerative sliding typical for brakes. 145-40 sample with chromel–copel thermocouple installed is tested with no gap between the thermocouple tip and sliding surface. The test procedure includes two stages: (1) the disc is accelerated by the rotor to a linear velocity v0 at the radius r; (2) the rotor is switched off and the sample is pressed against the rotating disc; due to the friction the disc velocity decreases to zero during t 0 seconds. The test is repeated N times under the same conditions, i.e., the fixed values of p and v0 . Fig. 14a shows an experimental data for a single test. Ts changes in a wide range due to the temperature peaks. A continuous thick line depicts the lower envelope of Ts for n¼ 4. Consider a signal created by superimposition of N temperature signals. The superimposed signals and their lower envelopes for N¼5, N¼10 and N¼20 are presented in Fig. 14b, c and d respectively. With increase of N, the lower envelope of the superimposed signal shifts down and approaches a limit curve depicted with a dashed line. The tests show that at N420
104
O. Nosko et al. / Tribology International 88 (2015) 100–106
Fig. 9. Relation between Ts and Tb: SFP04, chromel–alumel, fs ¼ 1 kHz, no filter, p¼ 0.5 MPa, v¼ 6.8 m/s.
Fig. 10. Evolution of Ts: 145-40, chromel–copel, fs ¼1 kHz, 500 Hz filter, p ¼ 0.3 MPa, v¼ 6.6 m/s.
Fig. 12. Dependencies εðnÞ and P ðnÞ.
Fig. 11. Determination of lower envelope of temperature signal.
new temperature peaks appear on the superimposed signal, but the changes in the lower envelope are negligibly small. According to the results obtained in Section 4.1, the limit curve coincides with Tb. Consequently, in the case of unsteady sliding, the
Fig. 13. Relationship between ε and P, fs ¼ 1 kHz: 145-40, chromel–copel, 500 Hz filter, p ¼0.2 MPa, v¼4.4 m/s; ■ 145-40, chromel–copel, 500 Hz filter, p ¼0.3 MPa, v¼6.6 m/s; ▲ SFP04, chromel–alumel, no filter, p ¼0.2 MPa, v¼ 7.7 m/s; ◆SFP04, chromel–alumel, 500 Hz filter, p ¼0.4 MPa, v¼ 6.6 m/s; ▼ SFP04, chromel–alumel, no filter, p ¼ 0.6 MPa, v¼ 7.7 m/s.
O. Nosko et al. / Tribology International 88 (2015) 100–106
105
Fig. 14. Determination of Tb for decelerative sliding: 145-40, chromel–copel, fs ¼ 1 kHz, no filter, p ¼ 1 MPa, v0 ¼ 6.6 m/s, t0 ¼ 13.4 70.2 s.
determination of Tb requires identifying the lower envelope of the superimposed signal.
Fourier transform show that φ takes a value between 0.9π and 1.07π at p ¼0.1–0.3 MPa and v ¼1.1–6.6 m/s. Suppose that a body oscillates on a counterbody, moving with a velocity v as follows:
4.4. Junction thickness x ¼ A sin ð2πf t Þ For correct analysis of the temperature signal, it is necessary to know the thickness of the thermocouple junction in the direction perpendicular to the sliding surface. This thickness allows to estimate at what distance from the sliding surface the temperature is measured. In addition, it can be useful when calculating the thermocouple time constant [15]. The pair SFP04/steel S275 has a negative friction–velocity slope [21]. When the pliable beam is used to support the holder, frictioninduced tangential oscillation of the sample occurs [1]. The displacement x changes harmonically with the frequency f ¼126 Hz and amplitude A of the order of 100 mm. The signal Ts from chromel–alumel thermocouple, involved in friction, oscillates almost harmonically with the frequency f and amplitude of several degrees Celsius. An example of the harmonic behavior of x and Ts is presented in Fig. 15. It is seen that there is a phase shift φ between x and Ts. The calculations done by use of the discrete
ð1Þ
If the positive direction of x coincides with the direction of the counterbody motion and dx=dt o v at any instant t 4 0, then the sliding velocity vs is defined as follows: vs ¼ v dx=dt ¼ v 2πf A cos ð2πf t Þ The friction heat is generated with specific power q ¼ μpvs ¼ μpv 2πf Aμp cos ð2πf t Þ where μ is friction coefficient. The oscillatory term of q, i.e., 2πf Aμp cos ð2πf t Þ, leads to the periodic temperature rise T in the body, which decays in amplitude with distance z from the sliding surface [22] pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffi! pffiffiffiffiffi! αAμp 2πf k z πf 3π z πf T¼ exp pffiffiffi sin 2πf t pffiffiffi ð2Þ 4 K k k
106
O. Nosko et al. / Tribology International 88 (2015) 100–106
of the temperature determination increases with measurement duration. An accuracy of 5% is attained when processing the signal for an interval of the order of 100 s. 3. For unsteady sliding, multiple tests under the same conditions and combination of the temperature signals into a superimposed signal are required. With increase of the number of the tests, the lower envelope of the superimposed signal approaches Tb. The temperature signals from about 20 tests allow for accurate determination of Tb. 4. The thickness of the thermocouple junction can be calculated on the basis of the phase shift between the friction sample tangential displacement and temperature signal in oscillatory sliding.
Acknowledgements Fig. 15. Harmonic oscillations of x and Ts: SFP04, chromel–alumel, fs ¼2 kHz, no filter, p ¼0.1 MPa, v¼ 2.2 m/s, x and Ts are reckoned from their average values.
where k is thermal diffusivity coefficient (k ¼4.9 mm2/s for chromel tape smeared over the insulation layer); K is thermal conductivity coefficient; α is heat partition coefficient. From the Eqs.(1) and (2) we derive pffiffiffiffiffi 3π z πf φ ¼ þ pffiffiffi 4 k or
pffiffiffi k 3π z ¼ pffiffiffiffiffi φ 4 πf
ð3Þ
According to the Eq. (3), the phase shift φ¼0.9π–1.07π obtained on the basis of the experimental data corresponds to the distance z¼ 52–110 mm which can serve as an estimate for the junction thickness. Calculate the time constant τ of the chromel–alumel thermocouple. According to the heat conduction theory [22], if a surface of a body is instantaneously heated by T 0 degrees at t ¼ 0 and its temperature is then maintained constant, the temperature rise T in the body is governed by the following equation: z ð4Þ T ¼ T 0 erfc pffiffiffiffiffi 2 kt where erfcð U Þ is complementary error function. If we suppose that T at the distance z reaches 63% of T 0 at the instant t ¼ τ, then due to the Eq. (4) for z¼ 52–110 mm we have τ¼1.2–5.4 ms. This is in a good agreement with the estimate τ¼3.4 ms reported in [9] for chromel– constantan grindable thermocouple with hE150 μm and mica insulation. 5. Conclusions The present study demonstrates the capabilities and limitations of grindable thermocouples as applied to polymer materials sliding on metal. On the basis of thorough analysis of the temperature signal Ts from the grindable thermocouple, the following results have been obtained. 1. Generally, Ts overestimates the sliding surface background temperature Tb due to the friction of the thermocouple on the disc. The lower bound of Ts coincides with Tb. 2. In the case of steady sliding, Tb can be determined as the lower envelope of Ts. This envelope is identified by finding the local minimum points of the temperature signal and removing the temperature peaks associated with these points. The accuracy
The authors wish to thank Mr. Hisao Miura at Tokai Material Co., Ltd. for supplying friction materials and Dr. Hideo Shibutani at Saitama University for his assistance in obtaining thermocouple micrographs. References [1] Kragelskii IV. Friction and wear. London: Butterworth; 1965. [2] Newcomb TP. Transient temperatures in brake drums and linings. Proc Inst Mech Eng Automob Div 1958;12:227–44. [3] Kennedy FE. Thermal and thermomechanical effects in dry sliding. Wear 1984;100:453–76. [4] Bowden FP, Thomas PM. The surface temperature of sliding solids. Proc R Soc Lond Ser A 1954;223:29–40. [5] Floquet A, Play D. Contact temperatures in dry bearings: three dimensional theory and verification. J Lubr Technol 1981;103(2):243–52. [6] Tian X, Kennedy FE, Deacutis JJ, Henning AK. The development and use of thin film thermocouples for contact temperature measurement. Tribol Trans 1992;35(3):491–9. [7] Kennedy FE, Frusescu D, Li J. Thin film thermocouples arrays for sliding surface temperature measurement. Wear 1997;207:46–54. [8] Peklenik J. Ermittlung von geometrischen und physikalischen Kenngrössen für die Grundlagenforschung des Schleifens. [dissertation]. Aachen: TH Aachen; 1957 in German. [9] Nee AYC, Tay AO. On the measurement of surface grinding temperature. Int J Mach Tool Des Res 1981;21:279–91. [10] Nosko AL, Kimura Y, Mizuguchi S. Measurement of surface temperature of sliding materials under sparse lubrication. In: Proceedings of the Japan international tribology conference. Nagoya, 1993. [11] Nosko AL, Kimura Y. Surface temperature measurement with tape-type thermocouples. In: Proceedings of the 6th nordic symposium on tribology. Uppsala, 1994. [12] Rowe WB, Black SCE, Mills B, Qi HS, Morgan MN. Experimental investigation of heat transfer in grinding. CIRP Ann 1995;44(1):329–32. [13] Babic D, Murray DB, Torrance AA. Mist jet cooling of grinding processes. Int J Mach Tool Manuf 2005;45:1171–7. [14] Batako AD, Rowe WB, Morgan MN. Temperature measurement in high efficiency deep grinding. Int J Mach Tool Manuf 2005;45:1231–45. [15] Lefebvre A, Vieville P, Lipinski P, Lescalier C. Numerical analysis of grinding temperature measurement by the foil/workpiece thermocouple method. Int J Mach Tool Manuf 2006;46:1716–26. [16] Lefebvre A, Lanzetta F, Lipinski P, Torrance AA. Measurement of grinding temperatures using a foil/workpiece thermocouple. Int J Mach Tool Manuf 2012;58:1–10. [17] Romashko AM. Study of heating of lift-and-transport machines disk brakes. [dissertation]. Moscow: Bauman Moscow Higher Technical School; 1979 in Russian. [18] Nosko AL. Development of method to calculate heating of lift-and-transport machines disk brakes with account for thermal contact resistance. [dissertation]. Moscow: Bauman Moscow Higher Technical School; 1985 in Russian. [19] Lebedev VG, Kochetkov AM, Zhabokritskiy RA. Assessing ground-surface contact temperature using grindable thermocouples. Sov J Mach Tools 1973;9:29. [20] Chichinadze AV. Polymers in friction assemblies of machines and devices. New York: Allerton Press; 1984. [21] Nosko O. Effect of temperature on dynamic characteristics of a pad sliding on a disc. [dissertation]. Saitama: Saitama University; 2013. [22] Carslaw HS, Jaeger JC. Conduction of heat in solids. 2nd ed.. Oxford: Clarendon Press; 1959.