You are on page 1of 16

MECHANICS

RESEARCH COMMUNICATIONS

Mechanics Research Communications 33 (2006) 190205 www.elsevier.com/locate/mechrescom

A simplied method for the impact test of beams using a pseudo-dynamic (PSD) process
Francisco J.Q. Melo a,*, Joaquim A.O. Carneiro b, Cassilda L. Tavares c, Pedro P. Camanho c, Paulo T. de Castro c
c a Department of Mechanical Engineering, University of Aveiro, Portugal Department of Physics Engineering, University of Minho, Guimaraes, Portugal Group of Structural Design and Experimental Validation IDMEC-Pole FEUP, Institute of Mechanical Engineering, Faculty of Engineering of University of Porto, Portugal b

Available online 16 June 2005

Abstract The impact test in structural parts for dynamic applications is an essential procedure for their certication in the presence of time dependent loads. In the case of beam elements, either built with one material or as an assembly of dierent material members joined with recent developed bonding techniques, an impact test is of leading importance, once the dynamic resistance of the joints involved in the beam fabrication is assessed and evaluated. The pseudodynamic method is an alternative to dynamic analysis of structures, here oering to the researcher the possibility of examining with detail the specimen in test for the initiation and progress of eventual damage mechanisms arising in the beam element joints whenever included in the design. 2005 Elsevier Ltd. All rights reserved.
Keywords: PSD; Pseudo-dynamic techniques; Impact analysis; Pultruded beams; Composite beam structures; FRP; Fibre-reinforced plastics

1. Introduction 1.1. General procedures in dynamics of structures The behaviour of a structure under the impact of a moving body consists in converting kinetic energy into deformation energy, where the structure may undergo elastic or plastic deformations. This subject
*

Corresponding author. E-mail addresses: fqm@mec.ua.pt (F.J.Q. Melo), carneiro@sica.uminho.pt (J.A.O. Carneiro).

0093-6413/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.mechrescom.2005.05.003

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

191

is a particular eld of the extensive area of structural dynamics with participation of ecient numerical or experimental tools for the assessment of the structure behaviour under impact or generalized dynamic loads. Such techniques lead to accurate results, provided that realistic modelling of operating conditions in test specimens is observed. In the nite element method the dynamic equilibrium of the structure submitted to time variable loads and inertial eects is governed by the equation: _ MU CU KU Ft 1 _ where [M], [C] and [K] are, respectively, the mass, the damping and the stiness matrices; U; U and U are the structure acceleration, velocity and displacement vectors. The solution of (1) may be obtained either by modal superposition or using direct time integration techniques. For fast rate dynamics and non-linear structure behaviour, the modal superposition gives place to the more ecient direct time integration methods. Procedures in experimental dynamics usually include essential accessories as shakers, excitation tables, accelerometers and eventually high-speed data recording equipment. The task is performed as in an open loop routine, as the operator analyses the results only at the nal step. The initiation of a damage conguration can hardly be detected with inexpensive methods for thoroughly investigation. 1.2. Pseudo-dynamic (PSD) techniques This is a less conventional tool in dynamic analysis operating in a closed loop. At each time-step, a structure parameter status is recorded at an experimental task and fed back to a computer program running a direct time integration algorithm for the structure dynamic analysis. A new structure conguration is computed in a next time step, which is prescribed to the structure in test via the test rig actuators. The operating principles of pseudo-dynamic techniques impose some limitations to the type of degrees of freedom prescribed to the structure in the test rig. Fig. 1 exemplies a structure for dynamic analysis submitted to a base horizontal motion. The displacement eld consists mainly in horizontal relative motions between each storey and the base (Clough and Penzien, 1992). Horizontal inertial forces are assigned at each storey level as in Fig. 1(b). Stiness structure values are assigned to the columns with horizontal slabs rigid. The column pair of each storey cell is displaced in parallel, allowing a dynamic model in Fig. 1(c). This is compatible with the operating principles of pseudodynamic techniques, where the test-rigs are equipped with linear displacement actuators. Applications of pseudo-dynamic techniques have major incidence in seismic analysis, where the time dependent structure conguration is mainly dened along the horizontal direction (Fig. 1). This conguration can be prescribed in a test rig at each storey level. A set of linear displacement actuators impose the displacement vector while a load cell reads the structure internal reaction force vector at each degree of freedom level. The method deals now with a modied version of Eq. (1):

U2 U1
U0
a

m2 k 2 m1 k1
b

F2 F1

m2 m1 2k 2 2k1
c

Fig. 1. Multi-storey structure under seismic loads: (a) basic structure; (b) and (c) simplied dynamic models with lumped masses and horizontal displacements.

192

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

_ MU CU Rst Ft

1a

where vector Rst is the structure internal restoring force vector read from load cells at the degrees of freedom. For structures where the displacement vector is dened along a common direction, the stiness matrix includes eorts associated to linear displacements, which is typical of multi-storey structures under seismic loads. Fig. 2 shows schematically the set-up for the method described with the operation sequence diagram. A set of linear actuators at the level of each degree of freedom prescribes a displacement vector. The load cells at each actuator axis read the internal reaction force vector. PSD techniques have reached success without major modications since the early experiments (Magonette et al., 1997; Molina et al., 2000). Innovation elds were attempted, not only in the design of test rigs but also in applications, as piping engineering (de Melo et al., 2001; Carneiro et al., 2001), where detail was focused on signal processor equipment and test-rig design to meet requirements assigned to geometry complexity and test specimens displacement eld. However, as this method is performed quasi-statically, the contribution of viscous damping forces in the dynamic behaviour of the structure practically is not detected. An upgrading of the method has consisted on operating the cycle of tasks depicted in Fig. 2 in a continuous process. This is known as continuous pseudo-dynamics and allows that contributions of damping forces arising from internal friction in the dynamically loaded structural members are included in the internal restoring force vector Rst (Eq. (1a)). In this work it is discussed the application of a pseudo-dynamic analysis to the investigation of the structural behaviour of a beam subjected to an impact load. The evaluation of the displacement of the structure at the contact point with a rigid striker is carried out as described next. This is an important step for the assessment of the structure performance, once it leads to the evaluation of the energy absorption capacity for the structure under impulsive forces. The displacement eld of a vibrating structure can be decomposed in a set of characteristic vibrating modes. The fundamental or rst vibrating mode corresponds to the minimum deformation energy associated to each of the superimposed modes dening the global structure motion; alternatively, it corresponds to a dominant exibility of the structure. Following this, in a dynamically loaded structure where it is expected that the rst vibrating mode is dominant, the approached analysis here proposed can produce results with a precision correspondent to the structure discretization into a single degree of freedom (SDOF). To minimize the errors arising from such a coarse nite element structure division, this analysis considers a

Reaction wall prescribe

Servo -actuators with load cells Test-specimen

Displacement transducers r ead

U (t+t)

U(t) and Rst

Actuator drive unit

Computer

Displacement trasducer record

Fig. 2. Closed-loop diagram for the operating principle of pseudo-dynamic techniques.

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

193

unique external force acting in a mode to induce mainly a fundamental vibration mode as described in the foregoing analysis. 2. Beam structures submitted to impact actions 2.1. Basic assumptions and validity of pseudo-dynamic techniques From previous discussion, it can be drawn that the exploitation of the potentialities of pseudo-dynamic techniques is limited to structural models where the nite element discretization refers to displacements of the same type of the actuators tted in the rig. Their number must be restricted, otherwise, the rig is expensive and its operation is complicated. An objective of the present work refers to the evaluation of the energy absorbing capacity of beam structures under impact loads. Truss or beam elements present simple damage mechanisms when they reach a collapse conguration. In the case of beam structures under multiple external loads, the collapse conguration can result from the combination of elemental damage mechanisms (Horne, 1979) as presented in Fig. 3. In the static analysis of framed structures a collapse conguration results from the combination of some elemental mechanisms as the represented in previous gure. If the dynamic rate of the external loads is not high, the expected collapse conguration is similar to the one statically generated. However, for fast rate dynamic loads it is possible that the previous congurations present an evolution similar to the presented in Fig. 4, these resulting from the combined action of the external loads and the inertial forces, which play an important role when high accelerations are present. 2.2. Set-up of a simple collision model It is understood that physically the collision between two bodies is a deformation process involving two elds, where these refer to a local surface distortion (resulting from contact forces) and a propagated deformation eld inducing bending or shear eects, as is the present case in discussion.

F
Plastic hinges

Fig. 3. Example of a portal frame and associated damage mechanisms for a single concentrated load: (a) basic design; (b) sway mechanism and (c) beam mechanism.

F F
Plastic hinges

Fig. 4. Example of a portal frame and the possible collapse mechanisms in a fast rate dynamic load: (a) basic design; (b) sway mechanism (plastic hinges in axial wave propagation) and (c) beam mechanism (plastic hinges in exural wave propagation).

194

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

As mentioned previously, it is adopted in this work a dynamic model with a single degree of freedom, assuming that it can practically condense the global dynamic behaviour of the structure at the structure/ striker contact point. This holds for deformation cases as the presented in Fig. 3, where the structure displacement at any point can be dened via simple expressions based on the deformed structure geometry and depending mainly on the displacement at the impacted section. In this section, the eect of the impact force of a striker against a beam structure is analysed under two aspects: The extended eect of the impact force in the beam structure, where this load is the cause for bending and shear deformation in a single element or for its propagation to adjacent elements integrating the structure. The local eect of the contact forces, which can be materialized for elastic or plastic surface deformations either in the striker or the still structure. The structure displacement eld from contact forces is much smaller than the one resulting from bending deformations; however the consequences of this localized eect have important repercussion on the time dependent structure displacement at the contact point, as it will be analysed next. The structure in analysis has a lumped mass distribution at the loaded point and remaining nodes (Fig. 5). The collapse mechanisms may involve sway or beam type, according to the external force orientation. From Figs. 4 and 5, it is observed that the external force F and the respective displacement are enough to dene an equivalent dynamic model, where the structure stiness Ks is calculated from the ratio of applied force F(t) and the consequent displacement at the contact point. The equivalent structure mass for this single degree of freedom discretization is calculated as resumed in Fig. 5. The equivalent stiness value Ks may be dened either in the elastic or plastic eld. Cases of plastic behaviour in pseudo-dynamic methods are experimentally evaluated at each load increment as referred in the foregoing analysis. To set-up the dynamic model of an impact problem using a pseudo-dynamic approach, a simple example is analysed next. The striker colliding with the structure is a virtual element only playing role in the algorithm in the pseudo-dynamic procedure as resumed in this sequel. Its contact force with the structure corresponds to the external load F(t) in Eqs. (1) or (1a). The following problem refers to a simple system composed by a rigid striker and a deformable structure as the presented in Fig. 6 where no damping eects are included. The dynamics of the striker/structure pair has two possible physical states:

mv /2 F c v mc /2 c

mv /2 mc /2

mv+ mc Ks

(a) - Basic system;

(b) mass lumping

(c) equivalent dynamic system

mv/4 v c (a) F mc /2 c (b)

mv /4 mc/2

2mv /4

Ks (c)

Fig. 5. Dynamic system equivalent to a structure under a single load: (a) basic system; (b) mass lumping and (c) equivalent dynamic system.

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205


Ms M

195

Striker at initial speed V0

System

K0

Fig. 6. Dynamic system for collision analysis in a single degree of freedom system.

The striker keeps in contact with system, at least during one time increment Dt and looses quantity of motion, impelling the system with the contact force, so increasing its potential deformation energy. In a non-contact situation, the striker and system form independent dynamic models, where the rst one travels at constant speed (eventually after a previous collision) as a rigid body and the second vibrates freely. These kinetic states are bound at each time step by the rst striker/system collision instant until the complete stop of both entities. The last state occurs when the striker kinetic energy is totally converted in the structure deformation energy. The mathematical expressions dening the previous kinetic states are resumed in the statements: (i) The contact/impact model in a linear-elastic structure deformation. The rst of previous assumptions is numerically dened from the dynamic equilibrium equation written at instant t + Dt as follows: Structure or system behaviour: M XtDt KXtDt FtDt Striker behaviour (from theorem of impulse) _ tDt _t FtDt Dt M s Xs Xs 2b 2a

where M, K, Xt + Dt, and XtDt are, respectively, the system mass and stiness matrices and displace_ tDt ment and acceleration vectors at time step t + Dt, while Ms, X tDt and Xs are, respectively, the stris ker mass, the displacement and the speed at time step t + Dt. In Eq. (2a) Ft+Dt stands for the system deforming force, coming from the impact of the striker having mass Ms. This, for its turn, reduces the speed according to the theorem of impulse, as by Eq. (2b). From the denition of constant acceleration and mean velocity between two consecutive time steps in the Newmark method, the following expressions are useful for the complete algorithm: _ _ XtDt Xt 0.5Dt XtDt Xt _ _ XtDt X t 0.5Dt XtDt Xt 2c 2d

Substituting previous expressions in (2a), it is possible to get updated values of the structure acceleration at time t + Dt _ M 0.25Dt2 KXtDt FtDt KX t DtXt Dt2 Xt 3

196

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

The external force vector Ft+Dt in Eq. (3) results form collision with the striker. Two dynamic states may occur: (a) Striker and system are both in contact (at least during a time interval Dt). Solving (2b) for Ft+Dt and eliminating it in (3), an equilibrium expression for contact conditions results: Ms _ _t Xt Xs 0.5M s Xt FtDt Dt with : M M 0.25Dt2 K 0.5M s _ and : FtDt KX t DtXt 0.25Dt2 Xt MXtDt
t

_ Ms and Xs are the striker mass and speed at time t. The updated value of striker speed at instant _ t + Dt is the same as the system, XtDt , having assumed a total contact system/striker during Dt. (b) Striker and system are independent dynamic models. Here, the striker travels at constant speed _ _ XsttDt Xstt , while the system vibrates freely until next collision: 5 where : M M 0.25Dt2 K (ii) The contact/impact model in a generalized structure deformation. Modications are necessary in previous analysis, now replacing the product KXt on the right-hand side of (4) or (5) by Rs, the internal reaction force vector. This vector corresponds to the product KXt only in a linear-elastic behaviour, while in a generalized non-linear behaviour, the internal reaction force vector Rs, is evaluated at each time step from experimental measurement. In Eqs. (4) or (5) other terms also contain the contribution of the stiness matrix K. It is usual to delete them in seismic analysis, once the structure velocity and acceleration vectors are not large; furthermore, such terms are multiplied by quantities Dt and Dt2 which are generally small. However, in a higher-rate dynamic analysis, velocity and acceleration are not negligible; such terms are multiplying the stiness matrix K(t + Dt) in (4) or (5) at time step t. As the stiness matrix may change during the deformation, a linear approach to the calculation of the structure stiness results from a next internal reaction force vector at time-steps t + Dt; this allows the evaluation of a tangent stiness matrix at time t: Kt Dt
tDt Rs Rts XtDt Xt

_ MXtDt KXt DtXt 0.25Dt2 Xt

where internal structure reaction forces Rts and RtDt are read from load cells at the level of structure s degrees of freedom. It was considered that the structure deformation resulted practically from bending or the combination of bending and shear eects, according with the deformation model in the numerical approach. The contact

F limit

F limit

l l

Fig. 7. Local deformation on the structure from the eect of contact loads of a rigid striker. The constitutive relation associated to the structure response is approximated as rigid/perfectly plastic.

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

197

force of structure/striker pair was calculated after the theorem of impulse having assumed an elastic shock. The consequent displacement at the contact point was considered much smaller when compared with other structure displacements; therefore its contribution is neglected. To investigate the inuence of an eventual non-elastic shock in the global displacement of the deformed structure, it is now considered that the contact force can reach a limit value set from a constitutive law. Such relation can be set either experimentally or by a numerical approach via nite element techniques. In the present analysis, a simple approach consisted in a deformation model for the contact eects where it was adopted a rigid-perfectly plastic constitutive relation as sketched in Fig. 7. It is considered that the striker is rigid, while only the structure can undergo plastic deformations. The inuence of a rigid-plastic shock in the global structure deformation is analysed next. 3. Pseudo-dynamic procedures in a structure under an impact load At each time step, the Newmark algorithm calculates an updated structure displacement vector. In the operation mode of the pseudo-dynamic method, this displacement develops is prescribed to the structure in test, giving rise to the internal reaction force vector. This vector is fed back to the Newmark algorithm allowing for a new displacement vector, as described. The computer program using algorithms (4) or (5) calculates updated displacement, velocity and acceleration vectors for system and striker. In next time step, the system displacement vector is prescribed to the structure via precision actuators with load cells at the level of the structure degrees of freedom. The procedure is resumed on test examples reported in the foregoing analysis. Particular details are important in the performance of the method in discussion: Though the Newmark constant acceleration method is unconditionally stable, accurate values result from a time step smaller than a critical Dtc (Bathe, 1992): (a) For single degree of freedom r 2p M Dtc 7 10 K This is the natural oscillation period of the system (b) For multiple degrees of freedom, Dtc = 1/10 of the least period associated to highest system eigenfrequency. The striker was considered perfectly rigid . Eects of surface distortion from local plastic deformation in contact are not considered. 4. Numerical examples: cantilever beam with concentrated mass under transverse impact Next examples have as objective some insight in the numerical approach to the contact-impact between a moving rigid body and a deformable structure. The problems were solved with a numerical approach via the Newmark algorithm. For sake of simplicity, the structure was assumed to undergo linear elastic deformations. Example 1. The assessment of previous algorithm is performed with a simple example, where results are compared with data from solid dynamics theory. Fig. 8 shows the model used in the analysis, involving a single degree of freedom system and striker. The critical time step by (7) is Dtc = 0.0347 s. Smaller intervals lead to precision increased solutions, having been adopted Dt = 0.01 s. Fig. 9 shows the results for the displacement progression of the elastic

198

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

Fig. 8. Dynamic model of an SDOF system under transverse shock load.

Max system displacement 0.157 m

0.15

1.0 0.8 0.6 0.4 0.2 0.0

System displacement (m)

0.10

0.05

Striker speed (m/s)

0.00

-0.2

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Time (s)

Time (s)

Fig. 9. (a) System displacement and (b) striker speed drop.

system and the successive striker speed drop as consequence of the of the shatter between the striker and system mass. The striker speed drop is evident in graphic (b), matching with time steps at displacement irregularities in graphic (a). The conservation of the total energy predicts that in a linear-elastic problem, the kinetic energy in the striker is converted in elastic deformation energy: 1 1 _2 M st X st0 KX 2max 8 t 2 2 This gives a maximum displacement Xmax = 0.1747 m for the system, while present result Xmax is 0.157 m, as shown in Fig. 9. This discrepancy can be analysed from Eq. (4), where the equivalent mass system is M M 0.25Dt2 K 0.5M s 4a

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

199

The equivalent mass (4a) increases with mass M of the deformable system, so determining in the next timestep smaller values of acceleration vector (4a) with incidence on the velocity and displacement in (2c,d). Physically this is understood from the fact of a moving body hitting elastically another one in a still position, a larger speed drop occurs with a larger mass of the still body. This inuence is inspected in the next case. Example 2. The striker mass is maintained at 10 kg but the elastic system mass is reduced to 0.25 kg (remaining parameters unchanged). The maximum displacement for the system is about 0.17 m (Fig. 10), closer to the theoretical value of 0.175 m from Eq. (8), still exhibiting some shatter from striker/system contact, as occurring in the ideally elastic contact between the striker and the structure. However, this is not realistic in real impact tests, where it is likely to occur a local smoothing eect from some plastic deformation at the contact surface. Next examples refer to the impact simulation of a beam with irrelevant end support design, as only an equivalent stiness value denition at the contact point interests. The beam is of an ideal material. The rst impact analysis considers that the contact force does not exceed a limit value, while in the second test that force could exceed it, leading to a local plastic deformation from the assumption of a rigid-plastic criterion. The problem data is as follows: Striker mass rms = 10 kgs. Initial striker velocity Vst = 1 m/s. Concentrated equivalent system mass rm = 1 kg. Equivalent structure stiness at contact point K = 300 N/m. Structure initial displacement = 0. Structure initial velocity = 0. Maximum contact force prior plastic deformation (rigid-plastic model). First problem Fclimit = 5000 N. Second problem Fclimit = 1500 N. Fig. 11 refers to the test where the contact behaviour of the structurestriker pair is perfectly elastic or exhibits a local damage.

0.20 1.0

System displacement (m)

Striker speed (m/s)

0.15

Max. displacement: 0.17 m

0.8 0.6 0.4 0.2 0.0 -0.2

0.10

0.05

0.00 0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Time (s)

Time (s)

Fig. 10. System displacement and striker speed drop for a system with an equivalent mass of 0.25 kg.

200
0.20

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205


0.20

Displacement (m)

0.10

Displacement (m)
0.05 0.10 0.15 0.20 0.25 0.30

0.15

0.15

0.10

0.05

0.05

0.00 0.00

0.00 0.00

0.05

0.10

0.15

0.20

0.25

0.30

time (s)

Time (s)

Fig. 11. (a) Evolution of the transverse displacement in a beam hit by a striker: striker mass = 10 kg; equivalent system mass rm = 1 kg; system stiness K = 300 N/m and limit contact force: 5000 N. In this problem the peak contact force was about 2500 N, not exceeding the limit contact force. (b) Evolution of the transverse displacement in a beam hit by a striker: striker mass = 10 kg; equivalent system mass m = 1 kg; system stiness K = 300 N/m and limit contact force: 1500 N. In this problem the peak force was exceeded (if the contact was perfectly elastic, the contact force reached about 2500 N). As the contact force is now limited to a prescribed maximum value, the maximum gap between the structure and the striker is now much smaller than in a rigid collision, which contributes for a smoothed aspect of the displacement graphic.

The structure displacement exhibits a graphic also obtained in previous examples, where the typical aspect arises from the rebound between the beam and the striker as consequence of the interaction between the elasticity and equivalent mass of the system at the contact point. Alternatively, when the limit value for the contact force is exceeded, the structure displacement graphic at the contact point appears as in Fig. 11(b), now much smoother as a result of a locally lower shock restitution factor compared with the rst case. On analysing the present results it is possible to note that, in spite of the dierent behaviour of both dynamic entities according to the contact characteristics, both the maximum structure displacement and the total elapsed time from rst impact until the complete striker stop did not experiment a signicant change. This result only applies naturally whenever the displacement at the contact zonal is much smaller than the one from the bending eect. If the objective of the experience is the evaluation of the maximum bending displacement that a beam under impact can withstand, it is possible to ignore whether there is a non-rigid contact in the impact process, once both the maximum displacement and the time for that value come practically unchanged.

5. Pseudo-dynamic simulation of the collision of a rigid striker against a simply supported beam Next experiment refers to the transverse impact of a rigid striker against a glass/polyester composite beam produced by pultrusion process with remarkable performance in the mechanical properties, this allied to an attractive aspect. Automotive and aeronautic industries are increasing the application of this type of elements as a partial alternative to lightweight pressed sheet metal hollow section elements. The aim of the next experiment is the evaluation of the shock energy absorption capacity for a pultruded beam element. The test was carried out in a multi-axial frame INSTRON machine as shown in Fig. 15. A single actuator set-up was used in the experiment. The test specimen has end joints bolted to the machine crosshead beams working as partially built-in edges (Fig. 12).

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205


Digital displacement gauge

201

Line feedback for displacement vector (optional)

Pultruded beam (test specimen) with bolted ends


Pull-down Load anchor

Crosshead beams

INSTROM Load cell Line feedback for internal restoring force vector

INSTRON Control unit

INSTROM Servocylinder with internal LVDT linear displacement transducer

computer

Next time-step displacement vector

Fig. 12. Schematic representation of servo-cylinder drive control in the INSTRON machine for pseudo-dynamic three-point pulldown test bending of a pultruded beam.

The beam end bolts support essentially shears forces, here assumed as equivalent to a single resultant reaction force and a bending moment. Dimensions and test-specimen properties are included in Fig. 13. Fig. 14 shows the design of bolted joint end supports with shear inserts fastening a composite beam to a metallic bracket. The adopted joint design has eight 4 mm-diameter holes drilled at each end of the composite beam supporting the material located at the exit of the drill to avoid damaging the composite. Following the procedure proposed by Camanhoe and Matthews (2000), metallic inserts with 4 mm and 6 mm internal and external diameters, respectively, were bonded into the holes using an epoxy structural adhesive. The use of the cylindrical bonded inserts leads to a redistribution of contact stresses in the composite, reducing the bearing stress concentration factor (Camanhoe and Matthews, 2000). The end distance and the adopted one between holes follow EUROCOMPs (1996) recommendations. Considering e as the edge and p the distances between the hole centres, ratios e/d = 3 and p/d = 6 were used (Fig. 14).

Length L=1.4 m

Transverse section

70 mm Thickness 3mm 30 mm

Fig. 13. Overall dimensions of a beam made by pultrusion process.

202

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

Fig. 14. Beam test specimen end support with a bolted joint including shear inserts.

A load was applied at the beam mid span section using the INSTRON machine actuator for evaluation of an initial stiness value to be added to the previous structure data in Fig. 13. In this experiment the limit contact force was supposed high enough to induce an elastic contact. The pseudo-dynamic process performs the following tasks during each time-step: 1. Prescription of a small displacement at the load point and reading the internal reaction force to calculate an initial SDOF stiness K0 = F0(load cell)/Dl in a three-point beam bending. The evaluation of the structure stiness was achieved dening a static forcedisplacement diagram on operating very slowly the machine in a set of prescribed displacement increments, pausing at each increment to read the structure reaction. This procedure practically suppresses the contribution of internal friction forces and allows a structure adaptation prior a next load increment. 2. Calculation of the equivalent structure (or system) mass at the contact/impact structure section, m % qAL/2, where q is the specic mass, A is the transverse area and L is the beam length. 3. Calculation of the critical time step Dt (from (7)). 4. Use of Eqs. (4) and (2a,b) for contact striker/system during increment Dt. 5. Analysis of gap between striker and structure (contact or non-contact). 6. Choice of Eqs. (4) or (5), depending on the contact status. 7. Reading from load cell of internal reaction force form test specimen and feedback it to Eqs. (4) or (5). 8. Calculation of next system acceleration, speed and displacement for system and striker; send data to actuator control unit and repeat procedure from step 5. Fig. 15 shows the experimental set-up used in the experiment. The INSTRON machine has an actuator with a built-in internal LVDT (linear voltage displacement transducer) along the hollow cylinder ram. Fig 16 shows the displacement at midspan contact section past successive time steps until the striker stop. It is noted the rebound between the test specimen and the striker as a consequence of the elastic contact predicted by the numerical model. The maximum transverse displacement experimentally measured at the contact section is about 0.00795 m as depicted in previous graphical results. This value is in good agreement with a theoretical result from (8) where the kinetic energy of the striker is 10 J. The elastic energy absorbed by the beam is: Ue = 300,000 (0.00795)2/2 % 9.48 J. The kinetic energy in the striker is Uk = 10 J. This discrepancy, though small, may be due to a discontinuous system kinetic state, with successive strikersystem rebound until complete stop. In fact, the experiment cannot be included in a linear analysis, once its geometry and inertial parameters change with time. It is noted here that the time scale in graphical results of Fig. 16 refers to virtual values. The procedure dealt with a modied (non-automated) equipment to cope with the operating method of pseudo-dynamic techniques, each time step carried out in the numerical/experimental tasks of the technique have in fact included delay pauses, this making the method largely dissimilar of the continuous pseudo-dynamic

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

203

Fig. 15. Test specimen installation in the INSTRON machine for three-point bending.

9,00E-03 8,00E-03 7,00E-03 6,00E-03 5,00E-03 4,00E-03 3,00E-03 2,00E-03 1,00E-03 0,00E+00 0,00E+00

Transverse displacement at

1.2 1 0.8 0.6 0.4 0.2 0 -0.2 0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016

contact point (m)

Striker speed drop (m/s)

5,00E-03

1,00E-02

1,50E-02

-0.4

Time (s)

Time (s)

Fig. 16. Displacement at impact point and striker velocity drop.

techniques. The real time of the test took about one minute; that is to say, a value in excess about 6000 times of the predicted time for the complete striker stop in the collision with the structure in test. This fact naturally excludes the contribution of internal friction forces in the deformed structure acting as a structural damping component.

6. Conclusions and discussion for technical applications The problem was formulated with a single degree of freedom, where concepts of dynamics of rigid or deformable bodies played role in the algorithm of the structure equilibrium. The method is proposed as an alternate procedure in the evaluation of the shock absorbing energy of structures under impulsive loads, as for example, test pendulum used in the certication of separate parts for specic application structures, as transport vehicle bodies (Dagson, 2001). The eect of an impact load from an external striker on simple structures as cantilever or simply supported beams can be simulated with acceptable accuracy with the present method, even when a contact surface damage can occur. This last case was approached having used a quite simple model, where the local structure behaviour was assumed a rigid-perfectly plastic constitutive model. The tool is an economic alternative for a real impact analysis with the attribute of being performed in a set of virtual time steps. The process can be interrupted when desired for test-specimen inspection, seeking for collapse mechanism initiations in bolted, welded or bonded joints if used in the design.

204

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

This method used has, however, important drawbacks resumed as follows: The simple formulation of the method, with only one degree of freedom, masks the presence of more deformation modes inherent to bending wave propagation when high impact speeds are involved. The performed experiments suggest that this approached analysis can be applied to truss or portal frames structures whenever the displacement eld can be dened with reasonable accuracy from a unique section, precisely the one at the contact zone with the striker. Essentially, the displacement eld of the structure in test is resumed to the rst (or fundamental) bending eigenmode, the one involving the minimum deformation energy. This constitutes a limitation for the technique when the structure displacement eld is too complex, making the method inapplicable in the characterization of the damage models as an alternative to true impact tests. The procedure was carried out in a sequence of prescribed displacements in each time increment, but eectively it corresponded to a long real time delay during each step of the process. As the internal restoring forces, resulting from the imposition of the updated displacement at each time step, were read after the structure in test was still, the contribution of internal friction forces in the structure deformation process was neglected. This suppressed an important contribution for a realistic dynamic behaviour, but for elastic deformations (which was the case of the composite beam analysed in this work), this omission may not be very detrimental in the precision of the results. The case of large deformations with non-linear material structure behaviour has a more relevant importance in the prediction of the dynamic behaviour of structural elements subjected to impact loads. The procedure here proposed needs the inclusion of an additional force vector resulting from damping eects in Eq. (1) of this work. As an accurate and complete characterization of the damping forces in dynamically stressed structures is not possible, an alternative to this consists on evaluating an equivalent viscous damping factor as described by Clough and Penzien (1992). With the equipment used in this work, a procedure to estimate an equivalent viscous damping factor in presence of large plastic deformations is proposed as follows: (a) Dene a structure stiness K as described previously (with small, slowly imposed and paused displacements to read the internal structure load). The procedure applies even with plastic deformations, where a tangent value for K can be calculated. (b) Change the structure deformation rate in the INSTRON machine, running the actuator ram at constant and increasing velocities to plot a set of displacement/force results read from the LVDT and the load cell. The evaluation of an equivalent viscous damping factor to be used in Eq. (1) of the dynamic equilibrium of the structure appears after subtracting in the structure internal forces read from the load cell the component resulting from the product of the stiness K with the corresponding displacement. It is noted here that, the maximum deformation velocity available in a INSTRON machine remains well lower than values expected in impact tests. The process here proposed is naturally an approach with a predictable accuracy highly dependent on the material properties and structural constitution of the test-specimen, Nevertheless, it appears as a simple process to avoid the set-up of complicated algorithms in the method. References
Bathe, K.-J., 1992. Finite Element Procedures, second ed. Prentice-Hall, Englewood Clis, NJ. Camanhoe, P.P., Matthews, F.L., 2000. Bonded metallic inserts for bolted joints in composite laminates. Journal of Materials Design and Applications: Proceedings of the Institution of Mechanical Engineers Part L 214, 3340. Carneiro, J.A.O., Pereira, J.P.T., Rocha, A.J., de Melo, F.Q., 2001. Structural validation of motor vehicle seats by pseudo-dynamic analysis. Paper 2001-01-3462, ATT/SAE, Automotive and Transport Technology Congress and Exhibition, Barcelona, October 2001.

F.J.Q. Melo et al. / Mechanics Research Communications 33 (2006) 190205

205

Clough, R., Penzien, J., 1992. Dynamics of Structures, second ed. McGraw-Hill, New York. Dagson, N., 2001. Inuence of the forming process on the crash response of a roof rail component. Master Thesis, Department of Solid Mechanics, Linkoping University, Sweden (Reg. No. LiTH-IKP-EX-1836), March 2001. de Melo, F.J.Q., Carneiro, J.A.O., Lopes, H.R., Rodrigues, J.F.D., 2001. The dynamic analysis of piping systems using pseudodynamic techniques. Journal of Strain Analysis ImechE 36 (5), 441451. EUROCOMP, 1996. In: Clarke, J.L. (Ed.), Eurocomp Design Code and Handbook. E&FN Spon. Horne, 1979. Plastic Analysis Design. Pergamon Press, Oxford. Magonette, G., Molina, F.J., Taucer, F., Verzeletti, G., Renda, V., Tognoli, P., 1997. Contribution of the JRC ISPRA to the intercomparison of analysis methods for seismically isolated nuclear structures. In: International Post-SMIRT Conference Seminar on Seismic Isolation, Passive Energy Dissipation and Active Control of Vibrations of Structures, Proceedings International Symposium, Taormina, 2527 August 1997, Italy. Molina, F.J., Verzeletti, G., Magonette, G., Taucer, F., 2000. Dynamic and pseudodynamic responses in a two-storey building retrotted with rate-sensitive rubber dissipaters. In: Proceedings of the 12th World Conference on Earthquake Engineering, New Zealand.

You might also like