Influence of heat treatment on hysteresis error of force transducers manufactured from 17-4PH stainless steel

Transcription

1 Measurement 39 (2006) Influence of heat treatment on hysteresis error of force transducers manufactured from 17-4PH stainless steel Bulent Aydemir a, *, Erdinc Kaluc b, Sinan Fank a a TUBITAK, Force Measurement Laboratory, National Metrology Institute (UME), P.K. 54, Gebze, Kocaeli, Turkey b Kocaeli University, Engineering Faculty, Kocaeli, Turkey Received 11 August 2005; received in revised form 20 March 2006; accepted 21 March 2006 Available online 24 May 2006 Abstract Different heat treatment processes can be applied on the spring element of a force transducer in order to obtain good and satisfactory performance. The study covers the attempts of different heat treatments on spring element using 17-4PH precipitation hardened stainless steel, which is regarded as one of the best and popular spring materials for force sensor applications. Heat treatments named as H900, H925, H1025, H1150 and S450 was applied, and different hardness values were reached. Especially, S450 heat treatment process was improved in this study. It was observed that heat treatments influenced the transducer performance, particularly hysteresis behaviour point of view. The results have shown that; hysteresis characteristics were improved with increasing hardness and sub-zero treatment process. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Force transducer; 17-4 PH; Heat treatment; Hysteresis 1. Introduction Strain gage based force sensors or transducers are used extensively in different field of industry. Several applications of force transducers in materials laboratories, manufacturing and control systems, where critical force measurements are strictly needed. In order to achieve low uncertainties in force measurements, the performance characteristics such as repeatability, linearity, hysteresis and creep errors influence the measurement uncertainty * Corresponding author. Tel.: ; fax: address: bulent.aydemir@ume.tubitak.gov.tr (B. Aydemir). of force transducer [1]. It is known that hysteresis error is one of the most important characteristics of force transducers [2 5]. It plays an important role on the measurement uncertainty of the force transducer. For this reason it must be as low as possible. It was determined that performance characteristics of force transducers were mostly affected by the heat treatment that was applied to spring element [2,6]. Spring materials may exhibit different hysteresis behaviour and the application of heat treatment on spring material causes important microstructural changes and respectively creates different hysteresis error performance in force transducers [1,7]. Various heat treatment procedures can be applied on the spring element of a force transducer in order to optimise performance characteristics. The study /$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi: /j.measurement

2 B. Aydemir et al. / Measurement 39 (2006) covers the attempts of different ageing heat treatments on spring element using solution annealed 17-4PH precipitation hardened stainless steel which is regarded as one of the most popular spring materials for force transducer applications [5]. Precipitation hardened stainless steel usually exhibits a higher hysteresis error than alloyed tool steel or aluminum [3]. Application of specially selected heat treatment reduces the hysteresis error. In this study, various heat treatments were applied on the specimens to obtain minimum hysteresis error from the produced force transducers. It is essential to select good performance materials, to perform good design, and to apply carefully controlled manufacturing procedures for force transducers [8]. The use of high accuracy electronic instruments in transducer outputs and the use of dead weight force standard machines in determining the performance characteristics of transducers are also very important to get reliable results. After application of all heat treatment and ageing processes, finishing process was applied to get the resulting specimens. Then special strain gages were bonded on the specimen surface. Finally, the selection and application practice of strain gages have been done properly. After completing the strain gage circuits of transducers, all of specimens are calibrated with dead weight UME (National Metrology Institute) Force Standard Machine having measurement uncertainties in a specified measurement procedure to obtain performance specifications of each specimen. 2. Materials and heat treatments of spring element of transducer Test material for the spring element of force transducers is 17-4PH steel mm diameter with steel bar was procured in condition A (Table 1). The Fig. 1. Force transducer specimens technical dimensional. tension type force transducer body specimens were first machined roughly to their dimensions, as seen in Fig PH steel is one of most popular spring element material for the force transducer application due to corrosion resistivity [1,9]. In order to change the microstructure in 17-4PH stainless steel specimens, various annealing heat treatments were applied after solution annealing process. Although H900, H925, H1025 and H1150 heat treatments could be found in literature, S450 heat treatment was specially improved for this study. The details of heat treatment, which were applied to 17-4PH stainless steel, are given in Table 1. These heat treatment processes were applied on the five groups of specimens. Each group has two identical specimens to determine effect of heat treatment on the hysteresis performance of force transducers. As a result of each heat treatment, hardness values were measured for each specimen using Zwick Z Table 2 Average hardness values of specimens Specimen code Hardness values (HRC) H H H H S Table 1 Heat-treating details of test specimens Specimen code (condition) Solution annealing processes Annealing temperature (K) Annealing time (min) Quenching medium Sub-zero treatment Ageing (precipitation) heat treatment Ageing temperature (K) Ageing time (min) Cooling medium A (standard material) Atmosphere H Atmosphere Atmosphere H Atmosphere Atmosphere H Atmosphere Atmosphere H Atmosphere Atmosphere S Water 1 h in liquid nitrogen Atmosphere

3 894 B. Aydemir et al. / Measurement 39 (2006) model hardness machine in order to determine relation between performance and hardness. Average hardness values in Rockwell (HRC) scale are given in Table Transducer design and strain gage application All transducers were designed for 100 kn capacity and produced in tension type and circular bar shape for easy strain gage application. Tension type transducer enables easy and exact centering in force standard machine since it has spherical apparatus, which eliminates the bending effect during loading. The basic elementary elasticity calculations were employed in the design of cylindrical tension type transducer. In order to eliminate end effect, the diameter to length ratio was taken greater than five. Diameter of tension bar is calculated as 20 mm for a strain level of 1500 lm/m for 100 kn capacity. After the application of heat treatments, specimen surfaces were finished to their exact dimensions by grinding to satisfy ±2 mm tolerance in all dimensions for each specimen. In order to get output signal from transducer spring element, N2A type 350 X transducer class (Measurement Group Co., USA) strain gages were used and bonded in this study. Its STC (self temperature compensation) number is 06 associated with exactly 17-4PH stainless steel for approximating the thermal expansion coefficients of the transducer material [10]. These strain gages are selected as 90 rosette to eliminate alignment errors. Correct application of strain gages is very important to obtain repeatable performance from the force transducers. Recommendations of strain gage producer were carefully applied to obtain better performance and quality bonding using M bond-610 adhesive for strain gage application. Since final transducer performances are greatly depending on the application of a uniform and repeatable clamping pressure, strain gages are bonded according to manufacturer recommendations using special clamping apparatus [8,11]. Two rosettes pair is bonded on opposite side with a 180 angle between them to establish a full bridge wheatstone circuit. Another pair of rosettes is bonded on the spring element to obtain another full bridge circuit. As it mentioned before that each group has two identical specimens. Although each specimen has a full bridge circuits, one specimen from each group has two full bridges circuit due to determination of the installation errors of strain Fig. 2. Completed force transducer specimen ready to performance test. gage application. A completed specimen prepared as force transducer was shown in Fig Calibration method for hysteresis error determination Hysteresis or reversibility error in force transducers is determined at each calibration procedure by taking the difference of force measurements in increasing and decreasing order of force applications (Fig. 3). In other words, the difference between the values obtained with increasing force and with decreasing force enables the relative hysteresis error (v), which can be calculated by using Eqs. (1) and (2) v 1 2 ¼ X X ð1þ X Nð1 2Þ v ave ¼ v 1 þ v 2 ð2þ 2 where v is the relative hysteresis error of force transducer, v 1, v 2 are the hysteresis errors in first and sec- Fig. 3. Schematic representation of hysteresis error on force transducer output graph.

4 B. Aydemir et al. / Measurement 39 (2006) Fig. 4. Force transducers under test in 110 kn dead weight force standard machine. ond series of readings, v ave is the average relative hysteresis error, X are the readings on the indicator with decreasing test force in first and second series, X 1 2 are the readings on the indicator with increasing test force in first and second series and X N(1 2) is the average reading on the indicator with maximum test force in first and second series. A schematic representation of hysteresis error is given in Fig. 3. Increasing and decreasing forces in 10 steps with 10% increment of 100 kn load were applied on the force transducer in a dead weight force standard machine in two series. Before beginning of the force application, three preloading in 100 kn as defined EN ISO 376 were applied on it [12]. Transducer output were recorded at the end of 30 s after application of each force steps using high precision indicating instrument that is DMP 40 S2 located in Force Measurement Laboratory of UME. Fig. 4 shows the force transducers that are under test in 110 kn dead weight force standard machine. These test are performed at very fine controlled laboratory conditions with a temperature of 21 ± 1 C and humidity 45 ± 5%. The short-term temperature control is better than ±1 C. Temperature variation is measured to be less than ±0.2 C during full test of transducer. 5. Results and discussion The hysteresis error of several heat-treated spring materials are presented in Figs In these figures, measured data are shown with symbols (D, h, s) while corresponding fitted curves are plotted with continuous lines. As it explained in transducer design section, that it was prepared two specimens for each group. One specimen has a full bridge circuits, other specimen has two full bridges circuit due to determination of the installation errors of strain gage application. In this case, three full bridge circuits which are named a, b, c on the graphs are used for H900, H925 and H1025. But, there are two full bridges on specimens H1150 and S450 due to some problem on the strain gages. Figs. 5 9 shows the hysteresis errors specimens represented with H900, H925, H1025, H1150, S450 coded, respectively. Average values of these data for each specimen groups are plotted to the graphs for comparison of the results to see effect of heat treatment on the hysteresis error of force transducers. The comparison of hysteresis error of several heat-treated spring materials are presented in Fig. 10. It is seen from Fig. 10 that the hysteresis error is strong function of the heat treatment and hardness. Even though, all the specimens have precipitation hardened martensitic structure, they exhibit different hysteresis errors. It is seen that the hysteresis error changes with changing hardness values. Similarly, hysteresis error is also a function of microstructure. These results can be attributed to crystal structure of the specimen forms after annealing and precipitation heat treatment. Heat treatment applied on the spring element influence the hysteresis error of force transducer according to measurement results [2,5,7]. It is found that the hysteresis error decreases at high hardness and increases as materials soften. Even in elastic region of the material, dislocation motion plays an effective role in hysteresis loop establishment [2,3]. In this case, each affects which decrease to movement of dislocation in metal crystal cause to decrease hysteresis error of force transducer. It is known that increasing dislocation density causes the dislocation locking and resist to dislocation movement in the grains [2,4,5,7]. As a

5 896 B. Aydemir et al. / Measurement 39 (2006) H900a H900b H900c Fig. 5. Hysteresis graph for H900 coded specimens H925a H925b H925c Fig. 6. Hysteresis graph for H925 coded specimens H1025a H1025b H1025c Fig. 7. Hysteresis graph for H1025 coded specimens. result of this resistance in high dislocation density material, the difficulties of dislocation motion cause to decrease hysteresis error under load. The average hysteresis error of the spring materials of the transducers which are H900, H925, H1025, H1150 and S450 coded specimens having precipitation

6 B. Aydemir et al. / Measurement 39 (2006) H1150a H1150b Fig. 8. Hysteresis graph for H1150 coded specimens S450a S450b Fig. 9. Hysteresis graph for S450 coded specimens H900 H925 H1025 H1150 S450 Fig. 10. Comparison graph for the hysteresis errors of all specimens. hardened martensitic structure were compared in Fig. 10 that the hardest materials having 45.3HRC hardness have the highest dislocation density and has the least hysteresis error. As a result of that the hardest specimens, S450 coded specimen show the best behaviour among all specimens or show the

7 898 B. Aydemir et al. / Measurement 39 (2006) smallest hysteresis error. This behaviour may be attributed to dislocation density. Because changing hardness and changing yielding stress of the material due to different tempering temperature cause the change of dislocation density on the material [4,7]. Dislocation density increases up to cm 2 during martensitic transformation of the steel [4]. After completing the martensitic transformation, application of different tempering temperature cause the diffusion of carbon atoms from distorted lattice structure of martensite, then hardness and dislocation density of the material decrease with increasing temperature. Shortly, any treatment causing an increase in dislocation density results decrease in hysteresis error. Decreasing of hysteresis error with increasing hardness may also be explained by anelasticity. Detailed information about anelasticity can be found in Ref. [13]. Short information about anelasticity was given below. Anelastic deformation of a component of time dependent deformation which, in contrast to plastic deformation, is recoverable upon the removal of applied stress. For stresses below r E, Fig. 11, which can be called elastic limit, the material deforms in a fully elastic manner, without any measurable contribution of time dependent components. As one increases the stress above the elastic limit, one observes the formation of closed hysteresis r e t loops. This fully elastic behaviour is seen in the REGION I Fig. 11, where r is the applied stress, r Y is yielding stress, r A is anelastic stress, r E is fully elastic stress, e t is the total strain, e y is yielding strain, e A is anelastic strain and e E is fully elastic strain. The fully elastic stress is proportional to the yield stress of the materials. Generally it is between 10% and 20% of the yield point. Some numerical values are given in Table 3 to show the level of r E for some materials [13]. These closed hysteresis loops reflect the existence of recoverable time dependent strain, which is defined as anelastic strain shown in REGION II Fig. 11. These hysteresis stress strain loops remain closed up to a stress level r A which shall call anelastic limit. When the stress exceeds the anelastic limit the hysteresis loops will cease to be closed when the specimen is unloaded to the zero stress. The existences of nonrecoverable time dependent strain are shown in REGION III Fig. 11. Since the stress level involved is still considerably below the yield stress of the material, the observed nonrecoverable strain is usually associated with microplasticity [13]. According to above explanation and Fig. 11, Fig. 11. Schematic representation of stress strain regions for anelastic stress and hysteresis loop establishment under yield stress. Table 3 Elastic stress values of some materials depending on the yield stress [13] Material Yield stress (r Y ) 10 1 MN/m 2 Fully elastic stress (r E ) 10 1 MN/m 3 (r E )/(r Y ) Pure aluminium stainless steel Pressure vessel when the applied stress closes the elastic limit, hysteresis error decreases considerably. As a result, if any heat treatment increases the yield point of the material, it has beneficial effect on the decreasing of hysteresis error. In this case, hysteresis error can be combined with the change of yield point, which depends on the hardness of material. As it known that yield point of material increases with increasing of material hardness. If the yield point of the material increases, anelasticity limit of the material are also increases. Since applied forces (applied stress level) on the specimens are identical for all type of specimens during performance test in this study, possible change on yield point of material due to heat treatment will cause the change of hysteresis error according to Fig. 11. If the yield point of material increases, hysteresis error of specimen decreases proportionally without changing the applied stress. Shortly, any treatment causing an increase in yield point or hardness of material results decrease in hysteresis error due to increasing anelastic stress (r A ) of material. Anelastic deformation resulting from dislocation bowing

8 B. Aydemir et al. / Measurement 39 (2006) and pile up against barriers to dislocation motion [13]. It is known that microstructural changes such as changing amount of residual austenite, Cu particles and precipitates due to different aging processes cause the hysteresis error change in martensitic stainless steel [2,3,5]. Since the main reason of hysteresis error change was caused by restriction of dislocation motion in structure, some mechanisms influence the dislocation motion in the different microstructure of 17-4PH steel. Increase of strength in precipitated structure is due to fine Cu particle distribution. These fine distributed Cu particle cause to restrict dislocation motion; consequently, it causes to increase strength of material [2,5]. As a result of that existing precipitated Cu particles barrier to dislocation motion due to bowing and pile up dislocation and this cause to decrease hysteresis error in precipitated structure. The martensite type, residual austenite, shape, size and distribution of cementite and/or other carbides as well as dislocation density are the principal factors that are causing microstructural changes. Since the main reason of hysteresis error was mainly attributed to the restriction of dislocation motion in the structure, some factors and mechanisms should effectively influence the mobility of dislocations in different microstructures of 17-4PH steel. The increase of strength in precipitated structure is due to fine Cu and Nb particle distribution in the structure. Because, distributed fine Cu and Nb particles in the martensite matrix cause very effective restriction to the movement of dislocations and consequently it results a considerable increase in the strength of material [2,3]. Martensitic structure obtained by martensitic transformation as in condition A in Table 1 has approximately 30HRC hardness. If this structure subjected to ageing heat treatment, its hardness reaches to 45HRC as in specimen S450 due to precipitated structure after ageing process. Dislocation density is one of the most important factors dictating the hysteresis error characteristics of materials [4,5]. The existing Cu and Nb particle barriers to movements of dislocations are playing a strong role in bowing and piling up dislocations effectively and this will cause decrease in hysteresis error for martensitic structures in steels [5]. 6. Conclusions The conclusions of the study can be outlined as follows: (a) The hysteresis error of a force transducer using heat treated 17-4PH steel spring element, can be reduced by maintaining the hardness of spring element as high as possible. (b) Microstructural change plays an important role on the hysteresis error of 17-4PH steel spring element having the identical hardness values. In the other words, microstructural change results in hysteresis error change even the hardness are identical for the same material. (c) Increasing temperature and holding time during ageing heat treatment cause to increase hysteresis error due to coarsening of precipitates. Coarse precipitates have lower resistance to movements of dislocations. (d) Over ageing should be prohibited for the 17-4 PH steel due to harmful effect on the hysteresis error. References [1] A. Bray, G. Barbato, R. Levi, Theory and Practice of Force Measurement, Academic Press, London, 1990, p [2] T. Allgeier, Factors influencing the mechanical hysteresis in stainless steel load cells, Ph.D. thesis, Glamorgan University, England, 1994, pp [3] T. Allgeier, W.T. Evans, Mechanical hysteresis in force transducers manufactured from precipitation-hardened stainless steel, J. Mech. Eng. Sci. Part C 209 (1995) [4] S. Fank, Effect of microstructural characteristics of spring material on the performance of force transducers, Ph.D. thesis, Istanbul Technical University (ITU), Institute of Science and Technology, _ Istanbul, Turkey, 2002, pp [5] B. Aydemir, An investigation on the effect to the performance of the microstructural characteristics of 17-4PH precipitation-hardened stainless steel force transducers, Ph.D. thesis, Osmangazi University (OGU), Institute of Science and Technology, Eskisehir, Turkey, 2003, pp [6] M. Kawai, Problems raised by improvements in load cell accuracy, in: Proceedings of the 10th International Conference of IMEKO, Kobe, Japan, September 11 14, 1984, pp [7] S. Fank, Kuvvet Dönüsßtürücülerde Isıl ve Mekanik _ Isßlemlerin Histerisiz Hatasına Etkisi, _ ITÜ Dergisi, Seri d, Cilt 2, Sayı 2, Nisan, 2003, pp [8] M. Spoor, Improving creep performance of the strain gage based load cell, in: Proceedings of the 11th International Conference of IMEKO, Amsterdam, The Netherlands, May 12 16, 1986, pp [9] Strain gage based transducers, their design and construction, Measurement Group Inc., USA, [10] Transducer class strain gages catalogue, Measurement Group Inc., USA. [11] Technical Note, Strain gage installation procedures for transducers, Measurement Group Inc., USA, 1978.

9 900 B. Aydemir et al. / Measurement 39 (2006) [12] EN ISO 376, Calibration of force proving instruments used for the verification system of the uniaxial testing machine, BSI, London, [13] P.S. Alexopoulos, C.W. Cho, C.P. Hu, Che-Yu Li, Determination of the anelastic modulus for several metals, Acta Metall. 29 (4) (1981)