International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 120 Development of a Dropped Weight Impact Testing Machine Leonardo Gunawan, Tatacipta Dirgantara, and Ichsan Setya Putra Abstract This article presents the development of a dropped weight impact testing machine. The machine was developed as a facility to provide experimental data to validate numerical simulations of impact loads on crash boxes, parts of car structure that absorb kinetic energy during collision. The dropped weight impact machine was designed to produce impact load to a specimen that represents a crash box. The machine was equipped with sensor systems to measure the velocity of the impactor just before it hit the specimen and the force that crushed the specimen, and with a data acquisition system to record the crushing force for further analysis. The development process included the design, fabrication, and function tests of the machine. The function tests performed after the machine was built indicated that the dropped weight impact testing machine can fulfill the design objectives. Results of several experiments using specimens in form of columns with square, hexagonal, octagonal and circular cross sections showed that the experimental results are in good agreements with the results of impact simulations carried out using Finite Element Method. O Index Terms Crash box, design, impact load, experiments. I. INTRODUCTION ne of the main concerns in traffic safety nowadays is the improvement of the vehicle crashworthiness. To increase passenger safety, some parts of automotive structure known as crash boxes are designed to absorb kinetic energy during collision. These components are usually in the form of columns which will undergo progressive plastic deformation during collision. The crushing force, i.e. the force needed to deform the crash box, determines the deceleration of the vehicle during collision and indicates the capability of the crash box to absorb kinetic energy. The value of crushing force is determined by the geometry and the material of the crash box. Several researches on the analyses of buckling behavior of prismatic columns under low velocity impact have been reported for several years [1, 2, 3, 4, 5, 6]. These earlier studies explored columns of different materials and geometries which were unrelated from one to another. Hence, it is not easy to obtain certain relation between columns geometry and Manuscript is received November 10, 2011. LeonardoGunawan is with the Faculty of Mechanical and Aerospace Engineering, Institut Teknologi Bandung, Bandung 40132, Indonesia ( e-mail: gun@ae.itb.ac.id). Tatacipta Dirgantara is also with the Faculty of Mechanical and Aerospace Eng., ITB, Bandung 40132, Indonesia (e-mail: tdirgantara@ae.itb.ac.id). Ichsan Setya Putra is also with the Faculty of Mechanical and Aerospace Eng., ITB, Bandung 40132, Indonesia (e-mail: isp@aero.pauir.itb.ac.id). material with their capability to absorb impact energy. Practical researches that produced usable design rules are limited. Further, for columns made of aluminum, which is of importance in the future due to its high strength to weight ratio, only a few results have been published. From these considerations, systematic studies on energy absorption capacity of prismatic aluminum columns under low speed impact were carried out [7, 8, 9, 10, 11]. In these studies, behavior of aluminum columns subjected to axial impact load were numerically analyzed using explicit finite element method. Parameters considered in this study were cross section geometry and wall-thickness to perimeter ratio. Fig. 1 shows a result of numerical simulation of an aluminum circular tube subjected to axial impact load [7]. To verify the results of these numerical studies, an impact testing machine based on dropped weight principle was developed. The impact speed obtained from this principle is limited by the dropping height. However, the range of the impact speed still fulfills the low speed criteria. Moreover, its development and operation cost is low. This article presents the design and testing of this machine. II. DESIGN OF LOW VELOCITY IMPACT TESTING MACHINE A. Design Objectives The impact testing machine was designed to perform low speed impact tests, i.e. impact velocities less than 15 m/s. At each impact speed, the kinetic energy should be able to be varied. The machine should be able to accommodate impact tests on specimens of various cross sectional geometry with maximum outer geometry of 60 mm and maximum height of 170 mm. Other considerations in design were the safety and cost aspects. B. Conceptual Design An impact machine based on dropped weight principle was selected since it is able to provide impact velocity by using earth gravity. This machine is a cost effective solution compared to that using a gas gun. In the design, a specimen was fixed on top of a steel base. An impactor was elevated and then released at a certain height above the specimen. The impactor would hit the specimen with an impact speed that depends on the dropping height. The kinetic energy of the impactor was then absorbed by the progressive folding of the specimen wall, which reduced the kinetic energy of the impactor until it finally stopped.
International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 121 Fig. 1. Crushing mode and instantaneous crushing forces of a circular aluminum tube under impact load from numerical simulations [7] The crushing force of the specimen during the impact was sensed by using a load cell which was placed between the specimen and the steel base. The crushing force data was then recorded by a data acquisition system. A speed sensor was used to measure the speed of the impactor just before hitting the specimen. The crushing force data, the dropped weight mass and the impact speed were used in the validation of numerical analysis. From the design principle and requirements, the dropped weight impact testing machine was designed. Based on the design objectives and following [12, 13], several designs were considered and the best solution according to some previously set criteria was selected. Fig. 2 shows schematically the final design of the impact testing machine which can be divided into 4 subsystems, namely: the frame that consists of guide columns, base plate and concrete block; the impactor assembly that consists of impactor frame, projectile, roller, and weighting mass; the clamp and hoist mechanism; and the instrumentation. Frame The columns were designed to act as a guide of the moving impactor assembly. Two columns guide design was selected from three alternatives: no column, one column and two columns. The two columns guide design was selected due to its safety, moderate price and easiness of operation. The columns were made of 6 m long stainless steel pipe, with outer diameter of 11.4 mm and thickness of 6 mm. They were mounted on top of a base plate made of 3 cm thick steel. The base plate was fixed on top of a 1 m 1 m 2 m concrete block. The concrete block was half-buried in a square hole with width of 1.2 m and depth of 1.7 m on the floor. The hole was at first filled with 0.2 m thick sand and then the concrete block was put inside the hole. Finally, the gap between the hole and block was filled with sand. The sand would isolate the shock and vibrations during the experiment to the neighborhood. Impactor Assembly The impactor assembly consisted of an impactor frame, an impact head, weighting masses and rollers. The rollers were attached to the frame and each roller was equipped with a pretension spring that kept the roller always in contact with the guide column. This mechanism ensured that the impactor assembly always moved along the guide column during the experiments. The impact head was designed to be the part of the impactor assembly that hit the specimen. The mass of the frame, roller and impact head without the weighting masses was 20 kg. The mass of the impactor assembly could be adjusted by adding several weighting masses until the total mass of the impactor assembly was 150 kg. If there is no significant friction working on the rollers, the impactor assembly will hit the specimen with a speed of: v = 2gH Although the length of the guide columns was 6 m, the maximum dropping height of the impactor was less than 6 m since some space should be provided for the specimen, the load cell, the hoist, the clamp and the impactor assembly itself. The maximum effective height for the impactor was 5 m which corresponds to a maximum impact velocity of 9.9 m/s. The maximum kinetic energy of the impactor with maximum mass of 150 kg is thus 7350 J. Clamp Mechanism and Hoist The clamp mechanism was designed to clamp and release the impactor assembly with maximum weight of 150 kg. A hoist was used to lift the clamp mechanism together with the impactor assembly to a certain height related to the desired impact velocity. By a trigger from an operator, the clamp mechanism could release the impactor assembly at the desired height which then moved downward and hit the specimen. Instrumentation 1) Load-cell The crushing force of the specimen was measured by using a load cell which was specifically designed and built for this machine. The load cell was designed to produce linear output without any hysteresis. Therefore, the load cell was designed (1)
International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 122 Fig. 2. Schematic drawing and picture of the dropped weight impact machine to be loaded with a maximum stress at 10% of its yield strength. Buckling of the load cell was also considered in the design. AISI 431 Stainless Steel was chosen as the material of the load cell. It has a yield strength of 655 MPa (at 0.2% offset), a Young s modulus of 200 GPa, and a good corrosion resistance. Fig. 3 shows the design of the load cell. The basic part of the load-cell was a tube with outer diameter of 90 mm, thickness of 4 mm, and height of 35 mm. Both ends of the tube were integrally connected to 80 mm thick flanges which were needed to realize a linear strain distribution over the cross section of the tube. From the geometry and material properties, the capacity of the load-cell is thus 70 kn. Prior to the manufacturing, impact simulations were performed to check the performance of this design. The load cell was machined from solid stainless steel. After manufacturing process, two strain gauges were bonded to the load-cell tube at half height and at circumferentially opposite positions. The strain-gauges were then configured in a half Wheatstone bridge. Detailed design process of the load cell can be found in [14]. The load cell was calibrated by using Tarno Grocki compression machine. The load cell was loaded by using quasi
International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 123 Fig. 3. Load cell design [14] Fig. 5. Speed sensor consists of two diode sensor and a bar fixed to the impactor Fig. 3. Load cell design [14] Fig. 6. Impact speed as function of dropping height static load. Outputs of the load cell with 0 to 45 kn loads at 5 kn increments were recorded. The measurements were carried out 4 times and the results are shown in Fig. 4. It can be seen that the relation between the load and the output voltage was practically linear and repeatable. 2) Speed Sensor To measure the speed of the impactor before hitting the specimen, an Autonics counter was used to count the elapsed time of the impactor assembly passing through two infrared diode sensors. The speed of the impactor was determined by dividing the distance between the two sensors with the elapsed time. Fig. 5 shows the two diode sensors and a bar fixed to the impactor assembly which represented the motion of the impactor. The speed of the impactor was then shown in a display unit. 3) Signal Processing and Data Acquisition Output signal from Wheatstone bridge was amplified, filtered, and stored into a PC by using a data acquisition card. Previous tests and numerical simulations showed that square aluminum specimen with cross section of 38 mm 38 mm and thickness of 2 mm was crushed in a very short time period, in the order of 20 ms. Hence the data acquisition card with the ability to sample the data with rate of more than 10 khz was used. By using the calibration factor of the load cell, curves of instantaneous crushing force versus time could be obtained from the output of the load cell. The mean crushing force of the column was calculated by using data of force as a function of displacement as follows:
International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 124 Fig. 7. Square and hexagonal columns before welding process Fig. 9. Deformation of a square column with perimeter of 112 mm from experiments and simulations Fig. 8. Instantaneous crushing force of a square column with 112 mm perimeter from simulation and experiment x 1 Pm ( x) = P( x) dx (2) x 0 To obtain data of force versus displacement, the displacement of the impactor during impact was calculated from the initial impact velocity and the travelled distance by assuming constant deceleration [16,17]. III. FUNCTIONAL TESTS Detailed design of the impact testing machine can be found in [15, 16, 17]. After the machine was built, several tests were carried out to check its functions. A. Impact Velocity For the dropped weight impact testing machine, the impact velocity depends on the height of the impactor above the specimen, as described by (1) if the friction between the impactor assembly and the guide columns is practically small. From several tests, data of impact velocities versus the dropping height were plotted and compared to the free fall case, as shown in Fig. 6. It can be seen that the velocities of the impactor were very close to the free fall velocities. This shows that the friction between the impactor and the guide columns was practically small. The data also shows that the impact velocities varied slightly from one test to another due to the fact that the height of the impactor could not be set exactly the same from one to the other test. Hence, the impact velocity in the experiment should be obtained from the speed sensor, not from the height of the impactor. B. Measurement of Crushing Force To check the function of the impact testing machine, experiments were conducted on columns with 4 cross sectional geometries: square, hexagonal, octagonal and circumferential. All columns were 150 mm long and they were made of 0.8 mm thick Al 6064-T4 with E = 68.9 GPa, ν = 0.33 and σ y = 145 MPa. They were made by using bending and welding processes. Fig. 7 shows the square and hexagonal columns prior to welding process. For each cross section, 7 columns of various perimeters were made, as listed in Table I. Validations of the impact machine were carried out by comparing the results of experiments with those of numerical simulations performed by using nonlinear finite element code.
International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 125 Fig. 10. Comparison of mean crushing force, P m, of square, hexagonal, octagonal, and circular columns with various perimeter obtained from simulations and experiments TABEL I Mean crushing force, P m, of columns with various cross sections obtained using numerical simulations and experiments Square Hexagonal Octagonal Circular L* P m [kn] L P m, kn L P m [kn] L P m [kn] [mm] N ** E *** [mm] N E [mm] N E [mm] N E 113 4.09 3.54 109 6.37 5.33 112 5.80 5.72 112 6.49 4.97 129 4.32 3.62 124 6.20 5.93 124 6.11 6.54 124 6.71 5.34 139 4.23 4.31 135 6.18 5.77 136 6.39 6.54 136 6.92 6.00 158 4.32 4.01 156 6.40 6.40 148 6.58 6.86 148 7.18 6.42 168 4.76 3.70 159 6.42 6.39 160 6.87 7.80 160 7.35 6.32 180 5.10 4.24 173 6.57 5.58 172 7.11 7.22 172 7.48 7.31 191 5.30 4.50 181 6.61 6.11 184 7.22 7.85 184 7.80 8.25 * L : length ** N : numerical simulations *** E : experiments In FEM analysis, the columns were modeled by using Belytschko-Tsay shell elements. Based on the convergence tests, element size of 1 mm 1 mm was used to obtain detailed numerical results at acceptable processing time. To make the model as close as possible to the specimen, the elements of the welding area were treated separately from the other part of the column by introducing a different Young s modulus value that was calculated from the hardness data of the welding area [16, 17]. Fig. 8 shows a typical graph of instantaneous crushing force versus time from one impact test and the corresponding numerical simulation for a square column with perimeter of 112 mm. The data shows that the crushing force fluctuated during the impact which was related to the fold formation of the column. In the first fold, two sides of the column bent outward and the other two bent inward. The force increased significantly and then reduced to a local minimum where the column sides were fully folded. This folding mechanism repeated until all kinetic energy of the impactor was fully absorbed by the folds. Fig. 9 shows that the folding modes and the number of folds in the experiment and in numerical analysis were in good agreement.
International Journal of Engineering & Technology IJET-IJENS Vol: 11 No: 06 126 C. Calculation of Mean Crushing Force Crushing forces for all columns were measured and presented as mean crushing forces by using (2). The experimental data were then compared to those obtained from numerical simulations as shown in Table I and also in Fig. 10. It can be seen that the mean crushing forces obtained from experiments and numerical simulations were in acceptable agreements. Some differences between numerical simulations and experiments were addressed to imperfections of geometry and materials of the specimens, and also to numerical models used in the simulations. Some improvements still have to be carried out to minimize the differences between numerical and experimental results. However, the results already indicated that the developed dropped weight impact testing machine can be used to validate the results of low speed impact simulations. IV. CONCLUSIONS The dropped weight impact testing machine has been successfully developed. The machine can produce impact load to a specimen with maximum speed of 10 m/s and variable mass up to maximum 150 kg. During the test, the impact speed and the time history of crushing force can be measured and recorded for further analysis. Impact tests performed by using this machine are able to produce data that is comparable to those obtained with FEM simulations. The impact machine can be further improved by equipping it with displacement sensor to measure the displacement of the impactor during impact. The force versus displacement curves can then be generated by combining the time history of force and displacement data. velocity axial impact load, in Proceeding of The 5th International Conference on Numerical Analysis in Engineering, Padang, Indonesia, 2007. [8] I. Wirayudhia, S. Sindhu, A. Jusuf, T. Dirgantara, L. Gunawan, I. S. Putra, Impact analysis of thin-walled prismatic columns, in Proceeding of the International Conference on Advances in Mechanical Engineering, Shah Alam, Malaysia, 2009. [9] M. Mora, A. Jusuf, T. Dirgantara, L. Gunawan, I. S. Putra, Low velocity impact analysis of foam-filled double-walled prismatic columns, in Proceeding of the International Conference on Advances in Mechanical Engineering, Shah Alam, Malaysia, June, 2009. [10] Annisa Jusuf, Fajri Syah Allam, Tatacipta Dirgantara, Leonardo Gunawan, Ichsan Setya Putra, Low velocity impact analysis of prismatic columns using finite element method, in Proceeding of the 8th International Conference on Fracture and Strength of Solids, Kuala Lumpur, Malaysia, 2010. [11] S. S. Hendradjaja, L. Gunawan, T. Dirgantara, I. S. Putra. Parametric study for circular and octagonal cross section tubes subjected to low velocity impact loading, in Proceeding of the Regional Conference on Mechanical and Aerospace Technology, Bali, Indonesia, 2010. [12] G. Pahl, W. Beitz, and K. Wallace, Engineering design: A Systematic Approach. Springer, London, 1996. [13] M. Navarrete1, F. A. Godínez and F. Serrania, Design and fabrication of a low speed impact tester, Journal of Applied Research and Technology, 2004, pp.249-254. [14] L. Gunawan, T. Dirgantara, I. S. Putra, and V. C. Thanh. Development of load cell for low velocity axial impact testing, in Proceeding of the Regional Conference on Mechanical and Aerospace Technology, Bali, Indonesia, 2010. [15] S. Siahaan, Design of a low speed impact testing machine with maximum load of 150 kg (in Indonesian), Undergraduate final project, Dept. Aeronautics and Astronautics, Institut Teknologi Bandung, Bandung, Indonesia, 2007. [16] S. S. Hendradjaja, Parametric study of prismatic columns with octagonal and circular cross section subjected to low velocity impact loading, M.S. thesis, Dept. Aeronautics and Astronautics, Institut Teknologi Bandung, Bandung, Indonesia, 2011. [17] B. Rabeta, Parametric studies of low speed impact loading on square and hexagonal columns using experimental and numerical method (In Indonesian), M.S. thesis, Dept. Aeronautics and Astronautics, Institut Teknologi Bandung, Bandung, Indonesia, 2011 ACKNOWLEDGMENT The authors gratefully acknowledge the support of the Directorate General for Higher Education of Republic Indonesia through Competitive Research Grant 2009-2010 and Competence Research Grant 2009-2011. REFERENCES [1] J. Alexander, An approximate analysis of the collapse of thin cylindrical shells under axial loading, Quarterly Journal of Mechanics and Applied Mathematics 13, 1960, pp.10-15. [2] T. Wierzbicki and W. Abramowicz, On the crushing mechanics of thinwalled structures, Journal of Applied Mechanics, 50, 1983, pp.727-734. [3] W. Abramowicz and N. Jones, Dynamic axial crushing of circular tubes, International Journal of Impact Engineering 2(3), 1984, pp.263 281. [4] W. Abramowicz and N. Jones, Dynamic axial crushing of square tubes. International Journal of Impact Engineering 2(2), 1984, pp. 179-208. [5] W. Abramowicz and N. Jones, Dynamic progressive buckling of circular and square tubes, International Journal of Impact Engineering 4, 1986, pp. 243-270. [6] W. Abramowicz and T. Wierzbicki, Axial crushing of multi corner sheet metal columns, Journal of Applied Mechanics 56, 1989, pp.113 120. [7] L. H. Anh, I. S. Putra, T. Dirgantara, D. Widagdo, H. Homma, and Kishimoto, Behavior of thin-walled prismatic columns subjected to low