On the design of unconventional testing machines for engineering testing – the case study of advanced joining processes unit

2.1.1 Commercial & unconventional torsion testing setups

Various commercial torsion testing machines exist that can test commonly used engineering materials. For instance, ZwickRoell GmbH & Co. KG have a range of horizontal torsion testing machines, where a servo-controlled electrical motor control can apply a torque loading of up to 500 Nm. Their testing machines also allow for the generation of a second axial load of up to 500 N [27]. Instron^® also provides a similar torsion apparatus series, named MT MicroTorsion Series, which can test specimens with a maximum torque capacity varying between 22.5 Nm and 5650 Nm. In their testing machines, one of the grips is fixed, while the other moves on a dual linear slide, thus offering high torsional stiffness, and a strain gauge torque cell is used to measure the torque being exerted [28].

These are just some examples of commercially available testing machines. In general, commercially available machines can deliver torque at different ranges, rarely going above approximately 5000 Nm. According to the literature, torsional testing of polymers may attain torques of up to 250 Nm, while metal specimens can attain much higher torque values. For the specific case of adhesive torsion testing, maximum torque values tend to be lower than 100 Nm for stiff adhesives, while the characterization of flexible adhesives requires much lower values [21].

Typical torsion testing machines need to have a relatively high axial stiffness [21], [27], [28], but precise alignment of the specimen may not be guaranteed, especially when the specimen has a complex mechanical behaviour, such as shape memory metal alloys or certain types of adhesives [21], [29]. Alternatively, special torsion fatigue testing devices have also been manufactured [30], [31]. In this manner, unconventional testing machines may be required.

Avila Ambriz et al. [30] developed a fatigue torsion machine where a servo-motor rotates one of the chucks fixing the specimen. This rotation is not directly transmitted to the specimen, but is transmitted with the help of a chain and a spocket. The developed setup also allows for bending loading with the help of a second linear actuator. Bustos and Vargas [31] have proposed a torsion fatigue testing machine that makes use of a flywheel. In this manner, one of the fixing chucks is connected to either a steel or an aluminium alloy flywheel, while the other is moved by an AC asynchronous motor that is commanded with the help of a variable-frequency drive. The electric motor is mechanically coupled to a flexible V-belt transmission and a two-stage gear reducer. Coupled to this gear reducer is a four bar linkage, with a geared rocker and crank that converts a continuous circular motion into an alternating semi-circular motion.

Regarding the development of special torsion apparatuses for adhesive joint testing, very few setups have be proposed. Brinz, Lewis and Geiger [29] patented an apparatus that can test adhesive joints under a combined tensile-torsion-shear load, which avoids unwanted displacement of the bonding surface. Goanţă et al. [32] proposed the use of a cam device to test specimens under a combined tension-torsion fatigue cycle. In this manner, the proposed device can be easily implemented in common uniaxial testing machines.

2.1.2 Proposed torsion testing architecture

The testing apparatus designed and manufactured by the AJPU research unit, which has a different machine architecture, is presented in Figure 2. This machine should be able to test bulk specimina, as well as butt joints and tubular joints with diameters varying between 5 and 30 mm, and lengths between 50 and 150 mm. The rotational test speed can vary between 0.05 and 1 rpm (which is equivalent to 5 ⋅ 10⁻³ and to 0.105 rad s⁻¹). The maximum torque required is of 60 Nm, while the resolution of the angular transducer is of 0.0005°. For the final machine project requisite, spurious tensile stresses must be below 1 MPa. These project requirements are of importance in defining the kinematic chain of the testing setup, since the structural kinematic chain must be stiffer than the specimen being tested. In this manner, the machine modes of vibration will not influence the control of the machine.

Figure 2:

Scheme of the developed torsion testing machine setup, with detailed views.

One must note that commercially available machines and most torsion testing apparatuses dispose their kinematics chain in a horizontal axis. However, with this strategy, the machine does not self adjust its grips in case of any misalignment, present in either the specimen, or in the assembled machine components. As such, normal torsion machines and specimens require tight dimensional and geometrical tolerances. Furthermore, the horizontal positioning of a long specimen will lead to an unwanted bending. The proposed architecture has a testing axis in a vertical direction, thus avoiding these issues [21].

In this manner, the torsion is imposed by the rotation of the NX 310EAPR7301 servomotor, which has a peak toque of 6.6 Nm and an angular speed of 2300 rpm at 230 V. This servomotor is coupled to a three stage planetary gear train, with a reduction ratio of 216:1, yielding a maximum torque of 80 Nm. The output shaft of the gear reducer is connected to the lower shaft (which is connected to the lower chuck holding the specimen) with the help of a torsionally rigid coupling. This coupling allows for axial, radial and angular misalignments, which is possible thanks to two duplex stainless steel diaphragms. Furthermore, to compensate for these misalignemets, both upper and lower shafts can have a deviation of up to 4°, relative to the vertical axis, thanks to two ball bearings, one for each shaft, as represented in Figure 3. Note that the machine’s mass center is placed above the ball bearing, with a huge contribution by both the lower shaft and the lower grip. In this manner, the normal functioning of the machine is unstable. To avoid structural dynamic issues, a counterweight is placed below the ball bearing.

Figure 3:

Schematic representation of misalignement compensation with both the shaft coupling and the ball bearing.

The upper shaft, which is fixed to the upper chuck, as shown in Figure 2, is connected to a torsionally rigid coupling. This coupling is also connected to the torque transducer, with a sensitivity of 1.0 mV/V and the capability of measuring torque of up to 100 Nm, which is above the required 60 Nm [21].

The torsion testing apparatus has four different testing modes: testing with controlled angular velocity, creep testing, stress-relaxation testing and testing with controlled torque. Each testing mode required its own controller for the servomotor [33]. The acquisition of the angular displacement is achieved with the help of a microscopic video camera.

3.1.1 Conventional tensile & compressive SHPB setups

A typical SHPB setup is composed by an actuation sub-system, the bar setup, and a data acquisition system. The most commonly adopted actuation method consists in launching a Striker Bar from a pressurized gas gun, which will hit the free extremity of the incident bar (also known as an input bar). From this impact, a compressive stress wave is generated, and moves towards the specimen, which is fixed between the incident bar and the transmission bar (or output bar). In the interface between the input bar and the specimen, part of the incident wave is reflected backwards, while the other part is transmitted to the specimen and, afterwards, the transmission bar. Finally, the transmitted stress pulse is absorbed by the momentum trap in the opposite side of the bar setup.

While this occurs, strain gauges connected to Wheatstone bridges, which are placed in the middle of the input and output bars, measure the three stress pulses and send the measurements to a high sampling frequency oscilloscope, or a computer with a DAQ board. Other auxiliary transducers may be present, depending on the characteristics of the SHPB machine in question, such as transducers to register the striker velocity before impact [38], [40].

Initial tensile SHPB setups did not follow a conventional architecture, unlike compressive SHPB machines, and various strategies were followed, such as the ones presented here. For a more detailed description of various tensile SHPB machines, please consult reference [38]. Lindholm and Yeakley [41] designed a top-hat specimen, which was positioned between a solid input bar and a tubular output bar. In this manner, an incident compressive pulse is generated, identically to that of compressive testing apparatus, but this wave is converted to a tensile wave while passing through the specimen. Nicholas [42] also proposed the use of a conventional SHPB compressive machine, where the specimen is fixed with threads, and a cylindrical collar is placed on top of the specimen. In this manner, the compressive stress pulse passes through both the collar and the specimen without hindering the integrity of specimen, until reaching the free end of the output bar, where it is reflected as a tensile wave. Afterwards, this tensile wave travels in the opposite direction, and is exclusively transmitted to the specimen. Harding and Welsh [43] proposed an alternative SHPB bar setup, where a transmission bar was placed inside a compressively loaded hollow input bar, and the test specimen was fixed with the help of a Yoke. Hauser [44] proposed a similar architecture where both the input and output bars are placed inside a hollow weightbar.

Despite the various solutions for tensile high strain-rate testing, the architecture defined by Nemat-Nasser et al. [45] has gained significant attention and is increasingly being adopted [46], [47], [48]. The proposed architectural solution consists in launching a Striker Tube, placed on the incident bar, towards an impact flange, which is placed in the free-end of the incident bar. The impact between both elements will generate a tensile wave that will propagate towards the specimen. The proposed SHPB testing machine architecture is based on this constructive solution.

3.1.2 Proposed SHPB architecture

As previously mentioned, there is a wide variety of SHPB architecture setups for high strain-rate tensile testing, which may not always be compatible with SHPB compressive setups. In this manner, a machine architecture is proposed, where one can perform tensile or compressive high strain-rate testing with a reduced number of assembly alterations. In this manner, the architecture schematized in Figure 5 is proposed. Both loading setups can be divided into three main subsystems – a pneumatic actuator, a pressure bar setup, and a braking system designed specifically to stop the actuator.

Figure 5:

Simplified schematic representation of the proposed SHPB setup. (a) Compressive testing setup. (b) Tensile testing setup (only the pressure bar subsystem is presented).

The pressure bar setup for the compressive apparatus follows the same architecture as the one represented in Figure 4, given that it is widely used in this type of machines. For tensile testing, the pressure bar subsystem is similar to the ones proposed by Nemat-Nasser et al. [45] and by Gerlach et al. [46]. To generate the high strain-rate loading, the striker bar needs to impact the Input Bar at high velocities. Impactor velocities for SHPB testing are typically of 20–30 m s⁻¹ [38]. However, velocities can reach up to 50 m s⁻¹ [37].

To attain these velocities, a novel symmetric double-rod pneumatic actuator was designed, that can attain high velocities in a relatively short stroke. An initial functional simulation of the system reveals that, for a compressed air supply pressure of 8 bar, a maximum velocity of 23 m s⁻¹ is attained [39]. However, conventional pneumatic actuators can only reach much lower velocities and, therefore, a specially designed braking system is required to stop the actuator. This is done with the use of a transmission lever that is hit by a stopping car coupled to the actuator’s rod. This transmission lever is also connected to a set of disc springs (also referred to as Belleville springs) and two shock absorbers [49].

It is noted that, in the beginning of a compression test, the Striker Bar is inside of a striker holder, fixed to the rod of the actuator, and, when the rod of the pneumatic cylinder begins to decelerate, the Striker Bar leaves the Striker Holder, and is freely launched towards the Incident Bar. Likewise, for the tensile test, the Striker Tube is launched inside a striker holder, and attains a free motion inside the tensile striker holder, while the rod is being decelerated by the braking system. Afterwards, the Striker Tube hits a Flange attached to a Loading Bar, similarly to what was proposed by Nemat-Nasser et al. [45], thus generating a tensile pulse. During these motions, optical sensors indirectly register the velocity of the impactor during its free motion, and strain gauges register the pulse wave through the Incident and Transmission Bars.

3.2.1 Complex pneumatic high-speed actuator

Tenreiro et al. [39] performed an initial requisite study for the SHPB apparatus, and determined that velocities of up to 25 m s⁻¹ may be required for adhesives and adhesive joint testing. Furthermore, tensile and compressive stress wave generation was required. In this manner, a symmetric double-rod pneumatic cylinder, that can move horizontally with high acceleration and with reduced friction, is considered as the best actuator solution for this apparatus. To achieve this, the actuator subsystem requires:

1. No physical contact between the rod and piston of the actuator, and its static components. This means that no piston seals, rod seals or guiding elements are present in the assembly, and that the rods are supported by a contactless method, such as air bushings (which are the blue rectangles in Figure 5a). Note that without the seals, there will be air leakages;

2. A reservoir, to maintain a high air mass flow without fluctuations, thus guaranteeing a continuous acceleration movement;

3. The maximum air flow be to used at the start the acceleration movement, during the first instances of movement. This requirement is essential, since, as previously stated, no seals are present in the actuator, and air leakages will inevitably exist.

Figure 6:

Detailed view of the actuator and corresponding braking system [39], [49].

Given the complexity of the architecture, a functional model of the actuator dynamics is required. A thorough description of the dynamic mathematical models to estimate the behaviour of the high-speed actuator is outside the scope of this work. However, one must point out that this required the modeling of the air reservoir, directional and quick exhaust valves [54], both pneumatic chambers [55], [56], the leakages between each chamber and the atmosphere, as well as air leakages between both chambers [57], and the actuator mechanics. For more information, please consult reference [39].

Given the required high velocities needed, there is a need to brake the actuator in a relatively short portion of the actuator’s stroke, which corresponds to 0.3 m of the total stroke of 1.8 m. In light of this, and given that the initial air pressure before movement will define the velocity of the striking element of the SHPB, an actuator braking system is required. This subsystem is composed by:

1. A Transmission Lever, whose rotation axle is placed such that a reduction of velocity is achieved;

2. A Belleville spring set, that can withstand the high generated loads in the deceleration stage, with minimal displacement, and that will absorb part of the velocity;

3. A set of Shock Absorbers, that dissipate energy for linear velocities of up to 5 m s⁻¹ [58].

3.2.2 Gripping system for tensile testing

An important aspect of tensile SHPB testing is the method in which the specimen is fixed. In compression testing, the specimen is simply placed between both bars with a slight tightening. In fact, to avoid overestimation of the specimen’s strength, caused by the interfacial friction of this tightening [59], lubricant and grease may be used [60], [61]. However, for tensile testing, this is simply impossible to perform, since, when the tensile wave passes up to the specimen, the Input Bar will inevitably suffer a displacement fromward the specimen, which will then fall. As such, the specimen needs to be clamped, glued or threaded to both Incident and Transmission Bars [62], [63], [64].

Based on the work of Ledford et al. [63] and of Constantino [65], Bernardo et al. [51] designed a gripping system, where an SLJ specimen is fixed on both ends with a grip, which will be threaded onto both the Input and Output Bars. Each grip consists of two components, one of which will be screwed on top of the main grip component, that has a cavity specially designed for specimen fixture. A necessary step of this work was the optimization of the thread shape, with three posssible shapes being studied: a quandrangular thread shape, a typical triangular ISO-thread shape, and a trapezoidal thread shape. This last thread profile was chosen, with a nominal diameter of 14 mm and a pitch size of 2 mm, which are present in international standards on trapezoidal threads [66].

4.1.1 Alternative drop-weight apparatus architectures

Normal Drop-weight impact machines follow a typical behaviour where an impactor, normally referred to as drop mass, is initially positioned at a certain height, from which it is released. The test specimen, which is placed below, is hit by the drop mass, and thus, the energy being transferred to the specimen’s deformation originates from the kinetic energy of the drop mass, which is a function of the mass, m _d , and of the initial height, h ₀ [71].

Ideally, this type of testing machines should have a anti-rebound mechanism or energy absorption device, to guarantee that no rebound of the drop mass occurs, and that no subsequent impact on the specimen occurs. However, not all drop tower machines have such a mechanism [72], [73]. One such testing apparatus with this shortcoming is the commercially available ZwickRoell Amsler MIT230F drop tower model, which can launch a drop mass of 23.5 kg from a height varying between 110 and 1000 mm, attaining a maximum velocity of 4.4 m s⁻¹. However, this apparatus does not have a setup avoiding drop mass rebounds [74].

Ambur et al. [75] proposed an apparatus where a drop mass assembly, that is initially lifted with the help of an electrical motor and a halyard, is released from an inputed height, until hitting a specimen. A damping unit consisting of a dashpot and a helical spring (a shock absorber), which is present inside the Drop-weight, functions as a anti-rebound system, thus avoiding subsequent unwanted impacts. Zhang et al. [76] developed a drop-tower machine to test concrete specimina. In this manner, this machine has a height range of up to 2595 mm, and two possible drop masses: an aluminium one, whose mass is of 18.60 kg, and a steel impactor, with a mass varying between 60.55 kg and 315.55 kg. Accelerometers are placed on the specimen, while a piezoelectric force transducer, which can measure loads of up to 177.92 kN, is present inside the drop mass. A magnetic strip is laced in one of the columns to determine the height of the drop mass. Adachi et al. [77] developed a small Drop-weight apparatus, where a thin aluminium drop mass, which has a magnet inside, is dropped from a tubular glass and hits the specimen being tested. The magnet passes through two coils, thus generating an electromotive force in each coil, given that these coils are placed at specific heights, the signals being measured allow the estimation of the velocity of the drop mass.

4.1.2 Proposed drop-weight apparatus

The proposed drop tower, which is schematically drawn in Figure 7, was projected and built by the AJPU research unit with several project requirements in mind. Firstly, this apparatus should be able to release a drop mass of 56 kg from heights of up to 1.27 m, reaching an impact velocity of 5 m s⁻¹, which yields a maximum energy of 700 J. For the maximum working velocity, a minimum impact energy of 50 J is also required, which translates to a weight impactor with a mass of 5 kg. Finally, the drop mass can be positioned at a specific inputted height with a resolution of 1 mm.

Figure 7:

Scheme of the developed drop tower setup.

In this manner, a gearmotor is used to rotate a bronze barrel with helicoidal profile, where the steel cable will be coiled during the ascending movement of the drop mass. It is noted that this gearmotor has a gearbox coupled to it, which performs a reduction of 1:80. To guarantee that the steel cable remains in a vertical position, an Axial Connector moves the barrel throughout the shaft in which is it mounted on. To help this, a second shaft, with a bronze screw, is also mounted in the Lifting System, and rotates accordingly with the help of a gearbox, thus enabling the winding of the steel cable. A three-dimensional rendition of this subsystem is presented in Figure 7. The electric motor, which has an encoder included, is controlled with the help of a PID controller [33].

The carriage connects the drop mass with the steel cable, thus allowing the drop mass to released from a specified height. This carriage has a release clamp where an upward shaft of the drop mass is fixed onto a sleeve-ball carrier and a compressed spring. Afterwards, the drop mass is released with the help of a pneumatic cylinder. A detailed view of this mechanism is shown in Figure 7.

While the drop mass falls, a fork attached to the impactor will pass through an optical detector, to determine the velocity of the drop mass. In this manner, the first tine of the fork will trigger a counter, while the second (and last) will stop said counter, thus allowing one to measure the velocity of the impactor right before the impact. However, before the test starts, a subroutine is triggered, where the positioning of the optical detector is made, by first zeroing its position (hence the micro-switches).

A Rebound Capture System was also implemented in this apparatus, with the aim of preventing subsequent rebound, after the first impact. As such, a system based on a simple lever mechanism was proposed. Initially, an arm, placed on each side of the drop mass, is in a horizontal position, and a fast-acting pneumatic actuator (one per arm) is actuated by a directional valve, forcing the rotation of the arm, thus lifting the drop mass. Shock-absorbers are also present to absorb the energy caused by the impact. This system is presented in Figure 8a. Before the mechanical design of this system, a functional simulation was developed, where several dynamic phenomena were included, such as the directional valve supplying the actuator [54], the active pneumatic chamber of the actuator [55], [56], the arm mechanics, and a LuGre Friction Model modeling the friction between the piston and the body of the pneumatic cylinder [78], [79], [80].

Figure 8:

Constructive details of AJPU’s drop-weight testing machine. (a) Rebound capture system. (b) Velocity acquisition system.

While the impact occurs, a piezoelectric load transducer, placed inside the drop mass, acquires the load being generated during the impact. An accelerometer is also present inside the drop mass, thus acquiring the acceleration during the fall and during the impact [81].

5.1.1 Proposed creep testing apparatus architectures

The proposed creep testing machine contains three testing stations, as shown in Figure 9, each one with two wedge action grips. The load capacity should be of 2.5 kN, with a maximum grip displacement of 300 mm. As mentioned in the beginning of the section, temperature conditions influence the creep behaviour of adhesive materials, and, therefore, a climactic chamber is also required so that tests can be performed at a temperature range between −100 °C and 200 °C.

Figure 9:

CAD rendition of the multi-station creep testing machine with detailed views.

The specimen is fixed with the help of a mechanical wedge action grip, which is detailed in Figure 9, and whose moving body design allows it to fix a specimen without altering the vertical position of the Jaw Faces. In this manner, the specimen is fixed at a pre-selected and specific position, without undesired compressive forces that cause specimen buckling. When the Body moves downward (or upward, in case of the upper grip), the inclined sides push against the matching side of the Jaw Faces, with the help of a tensile spring, forcing its simultaneous movement relative to the grip centerline, ensuring correct specimen alignment. The specimen does not suffer relative movement, since the Jaw Face surfaces in contact with the specimen are serrated with a diamond cut teeth. The Bushing guarantees the vertical movement of the Body during the grip tightening. Inside the Control Nut, that is fixed with the help of Handles, a Spindle exists and is responsible for fixing the Shoe position, and, consequently, the Jaw Faces holding the specimen.

The upper grip is fixed to an upper coupling rod that is also connected to the load cell, thus guaranteeing that load is measured during a creep test. This load is connected to the upper beam of the Test Frame with the help of a rod that has an Axial Spherical Bearings, as detailed in Figure 9. If both grips were locked in rotation, torsional loading would be unintentionally induced on the specimen. To prevent this issue, the articulated grip should be initially tightened [95].

The lower grip moves, since it is attached to a servo pneumatic actuator with an integrated displacement encoder, thus making it possible to acquire the piston rod displacement. The cylinder model DDPC-Q-80-300-QA, manufactured by Festo, has a stroke of 300 mm, an operating pressure range varying between 4 and 12 bar, and a theoretical force of 2721 N at 6 bar. Note that the actuator’s rod has a square cross-section to guarantee that no rotation of the rod occurs. The pneumatic circuit associated with each pneumatic actuator is presented in Figure 10 [95].

Figure 10:

Pneumatic circuit controlling each pneumatic actuator.

An Air Filter and Pressure Regulator unit is present in the beginning of the pneumatic circuit to supply the three actuators (one for each set of grips), and a 5/3 directional valve with closed center to define the movement direction of the actuator. A pressure regulator valve is required in the supply of the upper chamber of the actuator, to control the pressure in said chamber, and therefore, the load being applied. The 3-way proportional pressure regulator valve model VPPX-8L-8-1-G14-0L10H-S1, manufactured by Festo, with piloted diaphragm design, was chosen for this task, given that it has an integrated closed-loop control system. This control system compares the load condition defined in a computer software by the user, with the load value being measured by the load cell [95], [96]. Finally, two flow control valves in meter-out are employed to avoid sudden movement when the specimen fails, thus guaranteeing a smooth cylinder motion as well as regulating the piston speed [95].

5.1.2 Proposed climactic chamber setup

As previously mentioned, a thermal chamber is required to perform creep tests between −100 °C and 200 °C. However, the load transducer works only in a small temperature range, from −10 °C to 40 °C. Therefore, a heat insulator is required for each shaft connecting the upper grips with the load cells. This heat insulator consists of a three-component stainless steel cylindrical casing, in which the modified upper rod extension connects to the cylindrical base plate with the help of two nuts. Between the base of the cylindrical insulator and the bottom lid, a ceramic insulator is lodged, while the top part of the modified extension connects to the top lid of the insulator. Note that the empty space inside the casing is filed with rock wool, thus avoiding heat transfer by radiation, and maximizing insulation. Numerical thermal simulations show that the outer point at the upper lid of the insulator should be at a temperature of 40 °C, when the thermal chamber is at 200 °C. Conversely, when the thermal chamber is at −100 °C, the outermost connection of the insulator to the load transducer is at a temperature of 15 °C [97].

A Thermal Chamber was also designed to perform tests at various temperatures, which is presented in Figure 11. This chamber is composed of two half parts, and each can rotate relative to the lateral structural shafts of the test frame with the help of a Crank. Each part of the test chamber is composed of three 0.5 mm austenitic stainless steel sheets, that separate the inner chamber and the atmosphere. These sheets create two intermediary layers, the inward layer having Rock Wool, which insulates the heat. In the outward layer, forced air convection is created with the help of two fans, which are mounted on the back half of the chamber. A window is also present to help visualize the experimental tests. Stainless steel angles are mounted around the opening of the window, with a rubber band, to seal the thermal chamber. Two electrical resistances of 2.5 kW each are present in the back half of the frame. A hole connection is also present in the back half, thus allowing the injection of a cryogenic coolant and, consequently, performing tests at negative temperatures. The dimensioning of this chamber is done using heat transfer concepts, as specified in Ref. [98], thus yielding a heat transfer rate from the chamber to the atmosphere of 530 W, or a loss of approximately 10 % of the total heating power.

Figure 11:

Designed thermal chamber for the proposed creep testing machine.

A thermocouple type T temperature transducer was chosen to measure the temperature inside the thermal chamber, since this element can work at a temperature range between −270 °C and 400 °C [99]. An industrial temperature controller is adopted to control the temperature inside the chamber, and this controller makes use of time-proportional control to emulate a PID controller. In this manner, a relay is switched at high frequency, with the help of a solid-state relay that endures a high number of commutations. This industrial controller can also control a cryogenic solenoid valve, thus defining the flow of coolant entering the chamber [97].