Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 53, 3, pp. 745-755, Warsaw 2015 DOI: 10.15632/jtam-pl.53.3.745 ANALYSIS OF LATERAL VIBRATIONS OF THE AXIALLY LOADED HELICAL SPRING Krzysztof Michalczyk AGH University of Science and Technology, Faculty of Mechanical Engineering and Robotics, Kraków, Poland e-mail: kmichal@agh.edu.pl It is shown in this paper that the proposed concept by Krużelecki and Życzkowski (1990) of the equivalent rod can be applied in calculations of natural lateral vibrations of springs and that the obtained results will be nearer FEM results than the standardmodel based on the Timoshenko equivalent beam. The model, created on this base, allows one to calculate natural frequencies of the clamped-clamped spring. It is also shownthatmodels basedon the equivalentbeamconcept,whichare easier to apply than themodels treating the springas the spatially curvedrod,haveonlya slightly smalleraccuracy. It is also indicated that in themost commonpracticemaking ofmanufacturing end coils of springs, the natural frequencies differ significantly from the frequencies calculated bymeans of all testedmethods. The performed simulations show that differences between the first and the second as well as the third and the fourth natural frequency of the spring are small and, therefore, the axially symmetrical equivalent beam model can be used without a large error. The diagram allowing one to determine whether the desired frequencies are lower or higher than the cut-off frequency is developed for the presentedmodel. Keywords: helical spring vibrations, axially loaded spring,coil spring,Timoshenko equivalent beam 1. Introduction Lateral vibrations of helical springs constitute essential hazards for their operation safety, since they can cause high stress amplitudes and contribute to the loss of stability. On account of that, looking for ways of limiting natural vibrations of helical springs and the development of calculation methods allowing one to determine these limitations efficiency is essential. Equally essential is the accurate determination of natural frequencies. Problems related to determination the above value can be considered when the spring is treated as a spatially curved rod or using the concept of the equivalent beam.Analyses of spring vibrations treated as the spatially curved rod were presented, among others, by Love (1899), Wittrick (1966), Mottershead (1980), Pearson andWittrick (1986) Jiang et al. (1991), Yıldırım (1999)Becker et al. (2002), Lee andThompson (2001), Lee (2007), Stander andDuPreez (1992), Nagaya et al. (1986), Taktak et al. (2008), Yu andYang (2010). Theanalyses enabled a relatively accurate description of the effect, but due to problem complexity it was very difficult to apply and – depending on assumptions – to obtain analytical results in a closed form. For practical reasons much more suitable is the equivalent rod or beam model (Haringx, 1949; Della Pietra and Della Valle, 1982; Kobelev, 2014; Michalczyk, 2014). However, the beam model does not take into account several factors related to dynamic effects in springs. It assumes, among others, that the equivalent beam modelling the spring in the case of lateral vibrations is prismatic and has axially symmetrical parameters characterising stiffness andmass distribution. It can be expected that this assumption in the case of clamped-clamped springs (with fixed ends) – the most oftenmet case in practice – can be the reason of significant errors, since the spring lateral 746 K.Michalczyk stiffness is not the same in all directions and its distribution round the axis depends on the partial coil number.The problemof lateral vibrations of helical springs treated as the equivalent beamwas considered in the paper by Kobelev (2014). However, only the simplest case, such as the simply-supported beam, was solved in that study. A new concept of an equivalent rod, which could be applied to analysis of the helical spring static stability was proposed byKrużelecki and Życzkowski (1990). That idea, in contrast to the model of Timoshenko and Gere (1961), took into account the influence of the lead angle of the helix line on its flexural, shearing and compression rigidities. In addition, that concept allowed the analysis of spring buckling in two planes. The aim of the hereby paper is the analysis of the possibility of application of theKrużelecki andŻyczkowski (1990) concept to thedetermination of lateral vibrations of axially loadedhelical springs. The results obtained on the basis of such formulated model will be compared with the classical model of the Timoshenko equivalent beam, with the numerical simulation results and with the results achieved by other authors. The presentedmodel additionally takes into account the non-rotational way of mounting of the end coils. 2. Analysis For an element of the equivalent beammodelling the axially helical spring loaded by a force P, as shown in Fig. 1, the following equations of motion can be written (Haringx, 1949) ∂N ∂ψ +Q =0 me ∂2y ∂t2 = N ∂ψ ∂x − ∂Q ∂x mer 2 g ∂2ψ ∂t2 = Nϕ+Q+ ∂M ∂x (2.1) The angle ψ originates only from bending, while the angle ϕ originates only from the equ- ivalent beam shearing ψ+ϕ =− ∂y ∂x ∂ψ ∂x = M α ϕ = Q β (2.2) Fig. 1. Model of the equivalent rod for analysis of lateral vibrations of the helical spring The coefficients α and β are up-to-date flexural and shearing rigidities of the equivalent beam, respectively. The spring mass for its length unit is marked by me, while the radius of gyration of the equivalent beam cross-section is marked by rg. Limiting the vibration analysis to small deflections only, it can be assumed that N ≈ P. Making use of (2.2) in (2.1) on this assumption, after transformations, we obtain β ( 1+ P β ) ϕ = mer 2 g ∂2ψ ∂t2 −α ∂2ψ ∂x2 −me ∂2ϕ ∂t2 +β ∂2ϕ ∂x2 = me ∂2ψ ∂t2 +P ∂2ψ ∂x2 (2.3) Analysis of lateral vibrations of the axially loaded helical spring 747 Eliminating ϕ from equations (2.3) and transforming, we obtain thewave equation of lateral vibrations of the spring under the axial static load (Haringx, 1949) ∂4ψ ∂x4 − (me β + mer 2 g α ) ∂4ψ ∂x2∂t2 + m2er 2 g αβ ∂4ψ ∂t4 + P α ( 1+ P β )∂2ψ ∂x2 + me α ( 1+ P β )∂2ψ ∂t2 =0 (2.4) Since we are considering themodel with undamped vibrations, separating variables in equation (2.4), the function ψ(x,t) can be written as the product of the amplitude Ψ(x) and the time function T(t)= sinωt. Analogous expressions can be also written for the functions ϕ and y ψ(x,t) = Ψ(x)sinωt ϕ(x,t) = Φ(x)sinωt y(x,t)= Y (x)sinωt (2.5) Substituting proper derivatives (2.5)1 into equation (2.4) and performing transformations, we obtain the equation for amplitudes in the form ΨIV + [(me β + mer 2 g α ) ω2+ P α ( 1+ P β )] ΨII − [me α ( 1+ P β ) ω2− m2er 2 g αβ ω4 ] Ψ =0 (2.6) Denoting for simplification b = (me β + mer 2 g α ) ω2+ P α ( 1+ P β ) c = me α ( 1+ P β ) ω2− m2er 2 g αβ ω4 (2.7) The characteristic equation for (2.6), after substituting Ψ(x)= Cerx, takes the form r4+br2− c =0 (2.8) Substituting q = r2, we obtain a quadratic equation q2+ bq− c =0, whose roots are q1 = 1 2 ( −b+ √ b2+4c ) q2 = 1 2 ( −b− √ b2+4c ) (2.9) The expression, which occurs in square roots in (2.9) can be transformed, using (2.7), into b2+4c = m2e α2β2 (α−r2gβ) 2ω4+2 me α ( 1+ P β )(P β + Pr2g α +2 ) ω2+ P2 α2 ( 1+ P β )2 (2.10) This expression, similarly as b, is higher than null for an arbitrary frequency ω, which means ∧ ω­0 q2 < 0 Thus, the solution form of equation (2.6) depends on sign of the root q1. The cut-off frequency value ωb at which the solution form of equation (2.8) changes, is calculated by equating the right-hand side of equation (2.9)1 to null ωb = √ β mer2g ( 1+ P β ) (2.11) Thus ∧ ω<ωb Ψ(x)=C1cosh(k1x)+C2 sinh(k1x)+C3cos(k2x)+C4 sin(k2x) (2.12) where k1 = √ 1 2 ( −b+ √ b2+4c ) k2 = √ 1 2 ( b+ √ b2+4c ) (2.13) 748 K.Michalczyk and ∧ ω>ωb Ψ(x)=C5cos(k3x)+C6 sin(k3x)+C7cos(k4x)+C8 sin(k4x) (2.14) where k3 = √ 1 2 ( b− √ b2+4c ) k4 = k2 (2.15) The way of determining the cut-off frequency ωb for short, not loaded beams, was studied by Stephen andPuchegger (2006) andMajkut (2009). From the investigations of the author (which arepresented in the furtherpart of this paper) it results that even for relatively short springs, the first natural frequencies of lateral vibrations are lower than the cut-off frequency ωb. Therefore, in further considerations, expression (2.12) is used. The spring deflection angle resulting only from shearing, on the basis of (2.3)1, equals ϕ = 1 β ( 1+ P β ) ( mer 2 g ∂2ψ ∂t2 −α ∂2ψ ∂x2 ) (2.16) Making use of (2.2)3, (2.5)2,3 and (2.16) in equation (2.1)2, after transformations, we obtain the equation of amplitudes in the form Y =−    P ( 1+ P β ) +meω2r2g meω2 ( 1+ P β ) ΨI + α meω2 ( 1+ P β )ΨIII    (2.17) Using derivatives (2.12) in equation (2.17) and transforming, we finally obtain the equation of displacements in the form Y (x)= (−Ak1−Bk 3 1)C1 sinh(k1x)+(−Ak1−Bk 3 1)C2cosh(k1x) +(Ak2−Bk 3 2)C3 sin(k2x)+(−Ak2+Bk 3 2)C4cos(k2x) (2.18) where, for notation simplification, the following marks are introduced A = P ( 1+ P β ) +meω2r2g meω2 ( 1+ P β ) B = α meω2 ( 1+ P β ) (2.19) The boundary conditions, being the clamped-clamped spring, correspond properlywith the real operation conditions of a majority of springs applied in industry and, therefore, such a model will be considered. These conditions are 1) y(0, t)= 0 2) y(L,t)= 0 3) ψ(0, t) = 0 4) ψ(L,t) = 0 (2.20) where L is the real height of the spring. From conditions 1, 3 and 4, we obtain the following relationships C2 = −cosh(k1L)+cos(k2L) sinh(k1L)+ Ak1+Bk 3 1 −Ak2+Bk 3 2 sin(k2L) C1 C3 =−C1 C4 = Ak1+Bk31 −Ak2+Bk32 C2 (2.21) Analysis of lateral vibrations of the axially loaded helical spring 749 On the basis of the second boundary condition (2.20), we obtain the frequency equation 0= (−Ak1−Bk 3 1)C1 sinh(k1L)+(−Ak1−Bk 3 1)C2cosh(k1L) +(Ak2−Bk 3 2)C3 sin(k2L)+(−Ak2+Bk 3 2)C4cos(k2L) (2.22) Themodels proposedbyKrużelecki andŻyczkowski (1990) have beenused for the determination of the equivalent stiffness of the beammodelling the spring. The equivalent flexural rigidity and shear stiffness are expressed by α = 2EJ sinδ 2+νcos2δ β = EJ sinδ R2(1+ν sin2δ) 2πn φ0+2πn0 ∫ φ0 sin2φ dφ (2.23) where EJ is the product of Young’s modulus and area moment of inertia of the wire section, ν is Poisson’s ratio. The values δ and R are the lead angle of the helix and half of the diameter of the nominal spring for an arbitrary force P, respectively. The angle φ, occurring in (2.23)2, is the angle marked by the lead radius between the transverse force vector and an arbitrary cross-section of the spring wire. Since the transverse force direction may not coincide with the direction determined by the spring axis and the beginning of the helix line, the angle coordinate of the spring beginning φ0 can generally differ from zero. As compared to thewidely applied in the references equivalent stiffness, see e.g. Timoshenko andGere (1961),Guido et al. (1978), inwhich the lead angle of thehelix δ is assumedas constant andequal to zero, expressions (2.23) allowone to take into account anon-zerovalue of theangle δ dependent on the axial force P. The relationships given by Krużelecki and Życzkowski (1990) take also into account thenominal spring radiusR change aswell as the change of the active coils number n, under the influence of the force P. The authors assumed that the way of supporting the spring ends enables their free rotation. In such a case, during compression of the spring the number of its coils usually decreases (Michalczyk, 2009). Since the wire length is constant, it results in an additional increase in its nominal radius R. Its value as well as the value of the up-to-date helix line pitch angle δ can be determined by rearrangement of the relationships between the torsion and curvature of the helix wire and the force P given by Haringx (1948) PR2(1+ν)cosδ EJ =sinδcosδ− R R0 sinδ0cosδ0 PR2 sinδ EJ =cos2δ− R R0 cos2δ0 (2.24) In order to realise the condition of free rotation of the spring ends, one of them has to be supported on a thrust bearing. Only in such a case equations (2.24) can be used for calculating R and δ. However, most often, springs are supported in a way making free rotations of the end coils impossible and, therefore, this casewill be considered further.Thenumber of active coilsn of the spring loaded by the force P is equal – due to the clamped-clamped condition – to the number of active coils n0 of the not loaded spring. On this assumption andmaking use of the geometrical relations, it is possible to determine the helix line pitch angle δ and the spring nominal radius R under load δ =arcsin ( ( 1− P γ0 ) sinδ0 ) R = R0 cosδ cosδ0 (2.25) The equivalent shear stiffness γ0 occurring in equation (2.25)1 can be expressed by the equation (Krużelecki and Życzkowski, 1990) γ0 = EJ sinδ0 R20(1+ν cos 2δ0) (2.26) 750 K.Michalczyk Knowing the nominal radius R of the axially loaded clamped-clamped spring, it is possible to calculate its equivalent radius of gyration rg. In the case of modelling the radius of gyration of cylindrical helical springs, the equivalent rod constitutes the thin-walled cylinder of the average radius equal to the nominal spring radius R and of the mass equal to the springmass. On this assumption, making use of (2.25)2, the equivalent radius of gyration rg can be written in the form rg = √ 2 2 R0 cosδ cosδ0 (2.27) Introducing (2.25)1, (2.26), (2.27) into (2.11), (2.22), (2.23), we can calculate cut-off frequ- encies and natural frequencies of lateral vibrations. 3. Simulations, results and discussion 3.1. Analysis of natural frequencies of lateral vibrations of not axially loaded springs Comparisons of the results obtained on the basis of numerical FEM simulations of models (given in references) in which the spring is considered as a spatially curved rod of the classic model of the Timoshenko equivalent beam (applied up-to-date in all investigations concerning lateral vibrations of the springmodelled as the equivalent rod) with the proposed in this paper model using the concept of the equivalent beam given by Krużelecki and Życzkowski (1990) is presented below. As the first example, the spring of the same parameters as those used in work by Lee and Thompson (2001) is analysed. These parameters are as follows: total spring length L =332mm, springnominal radiusR0 =65mm,wire radius r =6mm,density ρ =7800kg/m3, number of active coils n =6, Young’s modulus E =209000MPa, Poisson’s ratio ν =0.28. The first four forms of lateral vibrations of the clamped-clamped spring not loaded by a longitudinal force obtained by the FEM in the ANSYS software are presented in Fig. 2. Fig. 2. Forms and frequencies of lateral natural vibrations of the clamped-clamped spring As can be seen in Fig. 2, the frequency f2 is only insignificantly higher than f1. In the case of frequencies f4 and f3, the difference is even smaller. The comparison of natural frequencies of the spring not loaded by a longitudinal force ob- tained by means of FEM, the dynamic stiffness method (Lee and Thompson, 2001), transfer matrixmethod (Pearson, 1982), standardmodel of theTimoshenko equivalent beam (STB) and themodel presented in this paper (MTB) is presented in Table 1. In the methods presented by Lee and Thompson (2001) and Pearson (1982), the spring is treated as a spatially curved rod. The solution, in both methods, is obtained bymeans of a numerical method. Analysis of lateral vibrations of the axially loaded helical spring 751 It should be noticed, that in methods using the concept of the equivalent rod modelled as acylinder, all frequencies of lateral vibrations are doubled due to the axial symmetry. Table 1 f1 f2 f3 f4 FEMmodel (Fig. 2) 45.181 47.007 89.051 91.581 Dynamic stiffness method 45.135 46.951 88.976 91.586 Transfer matrix method 45.13 46.95 88.97 91.59 Equivalent Timoshenko beam (STB) 45.975 45.975 94.088 94.088 Model presented in the work (MTB) 45.764 45.764 93.62 93.62 Comparing the results given in Table 1 it is seen that the application of the modified Ti- moshenko beammodel (MTB) with taking into account non-rotational ends support and using stiffness given in the paper by Krużelecki and Życzkowski (1990) provides, in the case of the analysed not axially loaded spring, results very similar to those ones obtained on the basis of the standard Timoshenko beam model (STB). The largest difference between the frequencies obtained from the FEM simulation and the frequencies obtained from the two last models did not exceed 6%. Themost often occurring case in the engineering practice is the spring of end coils bent and ground off, since only this way of support (except cases of using special spring holders or spring guided on a pin or in a cylindrical sleeve) ensures its stable operation. It can be expected that theway ofmaking the end coils has an influence on the natural frequencies. In order to illustrate this problem, FEManalyses have been performed for a spring of the same parameters as before but with different shapes of neutral coils. The model of such spring and its first four lateral vibration modes are presented in Fig. 3. Fig. 3. Forms and frequencies of lateral natural vibrations of the clamped-clamped spring with typical industrial end coils Comparing data contained in Table 1 and 2, it can be noticed that in the case of the spring with endcoils bent andgroundoff, abetter similarity of theFEMresults for themodels proposed by Lee and Thompson (2001) and Pearson (1982) is not obtained (at least for the first two frequencies) than the results obtained on the basis of the models based on the equivalent beam concept. Table 2 f1 f2 f3 f4 FEMmodel with bent coils (Fig. 3) 43.298 43.798 85.846 87.291 752 K.Michalczyk 3.2. Analysis of natural frequencies of lateral vibrations of axially loaded springs Helical springs applied in the industrymost often operate under a certain preliminary axial load, usually caused by the supported machine, weight or preliminary tension, e.g. in overload clutches, valves or in variable-speed transmission systems. Thus, the influence of the axial force (approximately statical) on natural frequency vibrations of helical springs is essential. Diagrams of the first four natural frequencies of lateral vibrations obtained (as before) fromFEManalyses and from the solution presented in this study are presented in Fig. 4. The FEM simulations were carried out in two stages: in the first one, the statistical analysis was performed at the determined spring load, while in the second, themodal analysis was performed for such a loaded model. Analyses and calculations were performed for the following spring relative deflections: 0; 0.125; 0.25; 0.375; 0.5. Simulations were carried out in the ANSYS packet, Mechanical APDL module. As can be seen in Fig. 4, along with an increase in the axial force P and thus an increase in the spring relative deflection, the model presented in this study improves convergence with the FEM in relation to the standard model of the equivalent beam. Fig. 4. Comparison of the first four natural frequencies of lateral vibrations obtained from FEM simulations with the first two double frequencies calculated on the basis of the standardmodel of the equivalent beam (STB) and the model presented in this study (MTB) 3.3. Cut-off frequencies of lateral vibrations of helical springs according to the presented model As has been mentioned above, equations (2.12), (2.18) and (2.22) are applicable only in the case when the cut-off frequency ωb determined by equation (2.11) is larger than the looked for natural frequencies. Thus, from the point of view of analysis of spring lateral vibrations, it is important to find out whether the looked for vibrations form is above or below the cut-off frequency ωb. This problem was analysed by Guido et al. (1978), where the authors indicated that for slenderness L0/D0 = 1 (or less) natural frequencies always correspond to solution (2.13). However, they did not provide neither slenderness values nor relative deflections atwhich the successive natural frequencies can be calculated from (2.12). On the basis of the studies performed at various material and geometrical parameters, it can be stated that the spring slenderness and its deflection decidewhether the given natural frequency is higher or lower than the cut-off frequency. The diagram allowing one to determine whether equation (2.12) or (2.14) should be used in calculations of natural frequencies of lateral vibrations is presented in Fig. 5. The up-to-date spring length L1 related to the length of a not loaded spring L0 is marked in the vertical axis, Analysis of lateral vibrations of the axially loaded helical spring 753 while the not loaded spring slendernessL0/2R0 in the horizontal axis. If, e.g. we are interested in natural frequencies of lateral vibrations of the clamped-clamped spring of a slenderness being 2.5 andcompressed to0.6 of its initial height (pointA inFig. 5), then thefirst twodouble frequencies ω1,2 and ω3,4 can be calculated using the solution in form (2.12) and (2.22), while in order to find the parameters of higher frequencies, the solution in form (2.14) should be used. Fig. 5. Curves separating the zones in which the cut-off frequency ωb is higher than the successive frequencies of lateral natural vibrations 4. Conclusions The modified model of the equivalent beam, presented in this study, allows one to determine form (2.18) and frequencies (2.22) of lateral natural vibrations of the axially loaded spring. The results obtained on the basis of the presented model, using the concept given by Krużelecki and Życzkowski (1990) are slightly closer to the results of the FEM analysis than the standard model based on the Timoshenko equivalent beam. The developed model allows calculation of natural frequencies of the axially loaded spring and supported in a waymaking free rotations of its end coils impossible. It is also shown that themodels based on the equivalent beam concept, easier in applications than models treating the spring as a spatially curved rod, have only insignificantly smaller accuracy. The largest difference between the frequencies obtained from the FEM simulation for the spring without bent end coils (Fig. 2) and the frequencies obtained from the presentedmodel is a little bit above 5%. Comparing the frequencies obtained from the FEMsimulation for the springwithoutbent end coils (Fig. 2)with the frequencies obtained from theFEM simulation for the spring of typical industrial end coils (Fig. 3) can be noticed that the differences are even larger – themaximumdifference for the second frequencyexceeded 6%.Thus, theway ofmaking the end coils significantly influences the natural frequencies. In themost often met in practice case of spring end coils being bent and ground off, the natural frequencies differ significantly from the frequencies calculated by all tested methods. The performed simulations indicated also that the differences between the first and the second as well as the third and the fourth natural frequency are small for springs and, therefore, the axially symmetric equivalent beammodel can be applied for calculating (without large error) these frequencies. The diagram (Fig. 5) allowing one to determinewhich formof the solution to equation (2.6) shouldbe applied in calculating frequencies and forms of natural vibrations of the spring of the given slenderness and static deflection has also been developed for the presented model. 754 K.Michalczyk References 1. Becker L. E., Chassie G. G., Cleghorn W. L., 2002, On the natural frequencies of helical compression springs, International Journal of Mechanical Sciences, 44, 825-841 2. Della Pietra L., Della Valle S., 1982, On the dynamic behavior of axially excited helical springs,Meccanica, 17, 31-43 3. GuidoA.R., DellaPietra L., DellaValle S., 1978, Transverse vibrations of helical springs, Meccanica, 13, 2, 90-108 4. Haringx J.A., 1948,Onhighly compressible helical springs and rubber rods, and their application for vibration-freemountings, I,Philips Research Reports, 3, 401-449 5. HaringxJ.A., 1949,Onhighly compressiblehelical springsand rubber rods, andtheir application for vibration-freemountings, II,Philips Research Reports, 4, 49-80 6. Jiang W., Jones W.K., Wang T.L., Wu K.H., 1991, Free vibrations of helical springsm, Transactions of ASME, 58, 222-228 7. Kobelev V., 2014, Effect of static axial compression on the natural frequencies of helical springs, MMMS, issue 3 8. Krużelecki J., Życzkowski M., 1990, On the concept of an equivalent column in the problem of stability of compressed helical springs, Ingenieur-Archiv, 60, 367-377 9. Lee J., 2007, Free vibration analysis of cylindrical helical springs by the pseudospectral method, Journal of Sound and Vibration, 302, 185-196 10. Lee J., Thompson D. J., 2001, Dynamic stiffness formulation, free vibration and wave motion of helical springs, Journal of Sound and Vibration, 239, 297-320 11. Love A.E.M., 1899, The propagation of waves of elastic displacement along a helical wire,Trans- actions of the Cambridge Philosophical Society, 18, 364-374 12. Majkut L., 2009, Free and forced vibrations of Timoshenko beams described by single difference equation, Journal of Theoretical and Applied Mechanics, 47, 1,193-210 13. MichalczykK., 2009,Analysis of helical compression spring support influence on its deformation, The Archive of Mechanical Engineering, 56, 4, 349-362 14. Michalczyk K., 2014, Influence of the elastomeric coating on parameters of steady state vibra- tions of coil springs in the resonance and outside it, Journal of Theoretical and Applied Mechanics, 52, 2, 507-518 15. Mottershead J.E., 1980, Finite elements for dynamical analysis of helical rods, International Journal of Mechanical Sciences, 22, 267-283 16. Nagaya K., Takeda S., Nakata Y., 1986, Free vibration of coil springs of arbitrary shape, International Journal for Numerical Methods in Engineering, 23, 1081-1099 17. Pearson D., 1982, The transfer matrix method for the vibration of compressed helical springs, Journal of Mechanical Engineering Sciences, 24, 163-171 18. Pearson D., Wittrick W.H., 1986, An exact solution for the vibration of helical springs using a Bernoulli-Euler model, International Journal of Mechanical Sciences, 28, 83-96 19. Stander N., Du Preez R.J., 1992, Vibration analysis of coil springs bymeans of isoparametric curved beam finie elements,Communications in Applied Numerical Methods, 8, 373-383 20. Stephen N.G., Puchegger S., 2006,On the valid frequency range of Timoshenko beam theory, Journal of Sound and Vibration, 297, 1082-1087 21. Taktak M., Dammak F., Abid S., Haddar M., 2008, A finite element for dynamic analysis of a cylindrical isotropic helical spring, Journal of Materials and Structures, 3, 641-658 Analysis of lateral vibrations of the axially loaded helical spring 755 22. Timoshenko S., Gere J.M., 1961,Theory of Elastic Stability, New York and London: McGraw Hill, p. 140 23. Wittrick W.H., 1966, On elastic wave propagation in helical springs. International, Journal of Mechanical Sciences, 8, 25-47 24. Yildrim V., 1999, An efficient numerical method for predicting the natural frequencies of cylin- drical helical springs, International Journal of Mechanical Sciences, 41, 919-939 25. Yu A.M., Yang C.J., 2010, Formulation and evaluation of an analytical study for cylindrical helical springs,Acta Mechanica Solida Sinica, 23, 1, 85-94 Manuscript received August 13, 2014; accepted for print March 13, 2015