Conference PaperPDF Available


Content may be subject to copyright.
24th ABCM International Congress of Mechanical Engineering
December 3-8, 2017, Curitiba, PR, Brazil
Rodrigo M. Nogueira
Marco Antonio Meggiolaro
Jaime T. P. Castro
Pontifical Catholic University of Rio de Janeiro, Department of Mechanical Engineering, Rua Marquês de São Vicente 225, zip
22451-900, Rio de Janeiro, RJ, Brazil,,
Abstract. This paper presents the design and development of a fast rotating bending test machine for high-cycle fatigue
tests, which can rotate up to 18,000 rpm, generating cyclic stresses at 300 Hz. The bending force is applied by an electric
actuator, connected in series with a load cell. An Arduino board with a DC motor controller shield receives the force
feedback from the load cell, commanding the actuator to create the desired bending moment through proportional-
integral (PI) controller logic. It also commands the desired rotation speed for the main motor via a PWM signal to the
brushless motor speed controller. Another Arduino board is connected to a Hall-effect sensor, to count the revolutions
and to control the speed of the main motor. All data can be captured by a computer, via a USB connection. Pairing the
bending stress and number of cycles for failure creates a point for traditional stress-life plots, but due to its active
feedback, the machine can also apply load blocks needed for Gassner curves and even to simulate variable amplitude
loads. The design is versatile, can apply stresses up to 3000 MPa, is lightweight and portable, and costs a fraction of
standard commercial models.
Keywords: fatigue, testing machine, S-N method, variable amplitude loads, instrumentation, machine design
Fatigue is a mechanical failure mechanism caused by the repeated application of variable loads, and characterized by
the initiation and growth of a crack that may eventually lead the structural component to fail. Fatigue failures are localized,
progressive and cumulative. They depend on a number of features, such as the material and shape of the critical point of
the component, and on the applied load history. Fatigue failure costs are significant, estimated as about 3% of the US
GDP (Norton, 2007). Since the gradual initiation and growth of a crack usually does not alter the behavior of the structural
component, it can pass unnoticed until sudden failure, causing loss of human lives, equipment and structures. That is why
fatigue prediction methods are so important in practice. Among them, the most used is the traditional S-N method, which
correlates the elastic stress history (S) at a critical point of a component with the high number of cycles (N) needed to
initiate a macrocrack there. S-N design methods intend to avoid crack initiation, and they need fatigue properties that are
usually measured in rotating bending machines, such as the one schematized in Figure 1.
High-cycle fatigue tests are widely used by industries and research laboratories, as well as for educational purposes.
They are particularly important to properly measure real or practical fatigue limits. Even though such limits may be
associated with relatively short lives of 106-107 cycles in steels and other ferrous alloys, they may be associated to much
longer lives, up to 5108 cycles, in Al and some other non-ferrous alloys. Such long tests are only practical in fast and
energy-efficient machines. Servo-hydraulic testing machines are versatile, relatively easy to control, and may apply very
large forces, but they are very expensive, consume a lot of energy, and are not so fast. They are almost indispensable in
any serious fatigue lab, but they are not a good choice to collect a large amount of S-N data, in particular in the case of
non-ferrous alloys. Servo-controlled resonant electromechanical testing machines are faster and more energy-efficient, so
they may be a better choice for measuring S-N data, but they are even more expensive and cannot apply high loads.
Traditional dead-load rotating-bending fatigue testing machines, on the other hand, are energy-efficient and less expensive
than servo-hydraulic or servo-electromechanical machines, but since they are not servo-assisted, they can only deal with
constant amplitude fatigue tests with zero mean loads.
The machine presented in this work has been conceived and designed to eliminate the main disadvantage of the dead-
load rotating-bending machines, their lack of servo-control. Moreover, it has been designed to be very fast, capable of
rotating at up to 300Hz, or to generate almost 26106 cycles per day, using special bearings and a small electrical motor.
Its details are described next.
R. M. Nogueira, M. A. Meggiolaro, J. T. P. Castro
A fast rotating bending fatigue test machine
Figure 1 - R.R. Moore rotating-beam fatigue testing machine
2.1 CAD Design
The design process started using a CAD software to model the structure and commercial components, to make a good
estimate of fittings, weight and assembly procedure (see Figure 2). Commercial rotating-bending fatigue test machines
usually come with a heavy workbench, necessary for the dead weights to pass through the tabletop. This new design,
which uses an electric actuator, allows the machine to be much lighter (16 kg), and to be installed on any work surface.
Figure 2 Machine design on CAD software.
There are two pivoted bearing blocks that support the main rotation shaft. This allows the actuator to apply a force of up
to 50 kgf at two symmetrical points in those bearing blocks, generating the desired constant bending stress on the main
shaft section where the test specimen is clamped (see Figure 3). Since the main shaft is able to slide through the bearings,
this cancels out undesired parasitic forces that would arise from the turning of the bearing blocks, such as axial tension.
24th ABCM International Congress of Mechanical Engineering
December 3-8, 2017, Curitiba, PR, Brazil
2.2 Mechanical Solutions
Initially, the shaft couplings that hold the test specimen to the main shaft were collars, using screws on one side to
clamp the shafts. Since these components have a naturally unbalanced design, the high rotational speeds led to high
vibrations, causing errors in the load readings. It also damaged the brushless motor, since it has a very tight clearance
between stator armature and rotor magnets. For this reason, set-screw shaft couplings were designed and machined with
tight tolerances, in order to keep vibrations to a minimum (see Figure4).
On the initial design, the bending force was applied by a wire rope wound around a pulley, on the shaft of a high-
reduction-ratio gearbox. The system proved to be powerful enough to bend the test specimen; however, it lacked precision.
The gearbox backlash was large enough to reduce most of the applied load, making it impractical to control. In the final
design, another structural aluminum U channel was added to support an automotive electric DC actuator, connected in
series with a load cell, to apply and measure the bending load, as seen in Figure 5. This proved to be a much more reliable
and precise solution, since the actuator has a self-locking mechanism.
2.3 Control electronics
An Arduino development board (see Figure 6) with a DC motor controller shield receives the force sensing
feedback from the load cell amplifier, ranging from zero to 5 Volts, equivalent to zero to 50 kgf. Therefore, the board is
able to command the actuator to create the desired bending moment by means of a proportional-integral (PI) controller
logic. It also commands the desired rotation speed for the main motor, sending a PWM control signal to a brushless motor
speed controller.
Another Arduino board is connected to a Hall effect sensor, counting the magnetic pulses coming from the motor
magnets. When the SS411P Hall Effect sensor detects a positive transition, from zero to maximum magnetic field, it
switches digitally to “ON”. The number of magnets on the motor rotor (ten) is set in the Arduino’s code. Since the magnets
have an alternating pole configuration, when five transitions from south to north pole are detected, one revolution is
Figure 3 - Main structure components
Main shaft bearings
Block pivoting bearings
Test Specimen
Figure 4 Shaft couplings. To the left, a commercial shaft collar. To the right, the machined set screw precision collet.
R. M. Nogueira, M. A. Meggiolaro, J. T. P. Castro
A fast rotating bending fatigue test machine
registered, counting as one stress cycle on the test specimen. This data is captured by a computer via USB connection,
using a LabVIEW interface, registering the applied load, total revolutions or cycles, and current RPM.
2.4 Fatigue Calculations
A worksheet was created to relate the force applied by the actuator in kilograms-force and the machine’s dimensions
to the bending stress occurring on the test specimen, as can be seen on Table 1. Several conditions can be tested using
this machine, such as yield and ultimate stress, surface finish, size, load and, in the future, temperature. Pairing the applied
bending moment and total number of revolutions at failure creates a point of stress and number of cycles for the stress-
life plot. After multiple experiments, an S-N curve can be fitted.
The final design (Figure 7) is easy to assemble and to maintain. It uses low-cost commercial components and
structural parts that do not require CNC machining. The test specimen can be replaced in under a minute, and the main
shaft can be disassembled in about 15 minutes.
During the development process, the noise and vibration generated by the machine were greatly reduced, allowing it
to be operated in the same room where researchers are performing other activities (noise level below 85dB). Also, a more
rigid coupling system between the actuator and bearing blocks was chosen, using a bolt as a pivot to distribute the forces
evenly to both sides of the machine. Table 2 shows a comparison between the Instron® commercial machine and the S-N
machine developed in this work.
Figure 6 ARDUINO UNO, open-source electronics development board
Figure 5 - To the left, the original gearbox and DC motor. To the right, the linear actuator on the final design.
24th ABCM International Congress of Mechanical Engineering
December 3-8, 2017, Curitiba, PR, Brazil
Table 1 - Fatigue calculations for the S-N testing machine.
Al 7075-T6
AISI 1020
4340 steel
Distance from center to traction point
L1 (mm)
Distance from traction point to pivot
L2 (mm)
Distance from center to pivot
L (mm)
SF (10³)
Applied traction force
T (kgf)
Applied bending moment
M (N.m)
Test specimen’s radius
y (mm)
Test specimen’s moment of inertia
Test specimen’s maximum stress (MPa)
σ (MPa)
Number of cycles until failure
N (life)
Figure 7 Complete S-N fatigue testing machine.
R. M. Nogueira, M. A. Meggiolaro, J. T. P. Castro
A fast rotating bending fatigue test machine
Table 2 Comparison between the developed machine and a commercial product.
S-N machine by PUC-Rio
Instron® R.R. Moore
Bending moment range (kg.m)
0,01 - 3,23
0,25 - 2,3
Bending moment increments (kg.m)
Rotation speed (RPM)
60 - 20.000
500 - 10.000
Effective minimum force (kgf)
0,1 (100g)
Machine weight (kg)
Dimensions (mm)
620 × 470 × 120
990 × 330 × 510
Power supply
120 V, 50/60Hz
120 V, 50/60Hz
Approximate total cost (USD)
This paper presents the design and fabrication of a new mechanical fatigue testing machine based on rotating beams,
with the objective of testing high-cycle fatigue and service-life loads. The system is capable of very high speeds, up to
18,000 rpm or 300Hz, easily generating 1 million stress cycles within an hour. This makes the machine suitable for
industrial material characterization, where many high-cycle tests are needed, as well as for educational use, since tests
can be performed during a class. Variable bending loading can be controlled through a USB connection, allowing testing
under variable amplitudes without the need for several dead-weight changes, leading to several research applications with
real-life loadings. Its lightweight design and low cost, about one-tenth of the commercial machine price tag, encourage
the use of multiple machines, allowing many tests to be performed at the same time.
Possible future refinements include:
Use of a single board to control the whole process, such as an Arduino Mega board.
An end-stop switch to stop the count and shut down motor when test specimen breaks.
Replacing the current brushless speed controller with a sensored version, in order to have better control at low
rotation speeds and more torque
Using CNC machined ER-type collets to hold the test specimen, providing better grip, easier disassembly and a
wider range of test diameters.
Applying a low-pass filter to the force measurement to improve the proportional-integral control.
Variable load input interface, allowing for the automatic application of real-life loading blocks.
This work was supported by CNPq (Conselho Nacional de Pesquisa e Desenvolvimento) and PUC-Rio (Pontifical
Catholic University of Rio de Janeiro). The author would also like to thank Guilherme Rodrigues D.Sc., for his support.
Hendrickson, D., “Fatigue Failure Due to Variable Loading”, Department of Computer Science, Physics, and Engineering
- University of Michigan
Norton, R. L. Machine Design Integrated Approach, 4th Edition 2007, Bookman
Meggiolaro, Marco Antonio e CASTRO, Jaime T. P. Fadiga - Técnicas e Práticas de Dimensionamento Estrutural sob
Cargas Reais de Serviço: Volume I Iniciação de Trincas, 2009.
Sedra, A. / Smith, K. Microeletrônica 5 ª Edition, 2007, Prentice Hall - Br.
Shigley, J. E.; Mishke, C. R.; Budynas, R. G., 2004, Mechanical Engineering Design, McGraw-Hill, New York.
Dynamic mechanical testing machines -
Material Property Data -
Material Property Data -
The authors are the only responsible for the printed material included in this paper.
... Uno de los dispositivos para ensayar este comportamiento es la máquina de pruebas de fatiga de viga rotativa, que se emplea para diversos estudios, por ejemplo, para estudiar tasas de crecimiento de grieta (Mayén et al., 2017), así como los efectos que tienen diferentes técnicas de mejoramiento de la vida útil en materiales de ingeniería (Gallegos-Melgar et al., 2020;Zuno Silva, 2016). La aplicabilidad de estas máquinas se ha extendido para abarcar ensayos con cargas a altas frecuencias en amplitud variable (Meggiolaro et al., 2017) y para probar materiales fabricados por impresión 3D (Slotwinski & Moylan, 2014). ...
Conference Paper
En las pruebas de fatiga realizadas en máquinas de viga rotativa se utilizan probetas con geometrías estandarizadas, las cuales presentan cambios de sección transversal para garantizar que la falla ocurra en una longitud de calibre. El cálculo correcto de los esfuerzos en la longitud calibrada de la probeta requiere que se considere un factor de concentración de esfuerzos, el cual no se ha reportado previamente en la literatura especializada. En este trabajo se utilizaron tanto el Método de Elementos Finitos como la Teoría de Vigas para el cálculo del factor de concentración de esfuerzos de una probeta con cambio de sección transversal complejo. Para ello, la mitad de la probeta se modeló en elementos finitos como una viga en voladizo y se determinó el esfuerzo en la zona de empotramiento. Analíticamente, se calculó el esfuerzo considerando a la máquina de pruebas de fatiga como una viga a flexión de cuatro puntos. Los resultados de ambos métodos se compararon para establecer el factor de concentración de esfuerzos. Se determinó que, si no se toma en cuenta el factor de concentración de esfuerzos, el esfuerzo en la probeta calculado con los modelos analíticos es 24% menor que el obtenido en elementos finitos.
... The magnitude of enhancement of fatigue resistance is higher at lower wt% of SiC (0-2 wt%); however, no significant change can be seen at higher percentages (4 and 6 wt%). Several studies have shown that increasing the wt% of MMC particles enhanced the fatigue strength [4]. ...
Full-text available
p>Fatigue is a process of progressive localized plastic deformation occurring in a material subjected to cyclic stresses and strains at high stress concentration locations that may culminate in cracks or complete fracture after a sufficient number of fluctuations. Fatigue testing is carried out using the ASTM D3479 with a notch or crack for investigating the initiation of crack. Several fatigue tests were conducted in tension-tension and/or tensioncompression loading at a frequency of 10Hz or sinusoidal wave’s frequency of 5Hz, and at constant-amplitude. The fatigue tests were interrupted by the researchers at regular intervals after a predetermined number of cycles to monitor crack advance and to observe the failure modes by various ways such as visual observation, digital camera, traveling microscope, CCD camera, etc.</p
Fatigue Failure Due to Variable Loading
  • D Hendrickson
Hendrickson, D., "Fatigue Failure Due to Variable Loading", Department of Computer Science, Physics, and Engineering-University of Michigan Norton, R. L. Machine Design-Integrated Approach, 4 th Edition 2007, Bookman Meggiolaro, Marco Antonio e CASTRO, Jaime T. P. Fadiga-Técnicas e Práticas de Dimensionamento Estrutural sob Cargas Reais de Serviço: Volume I-Iniciação de Trincas, 2009.
Microeletrônica -5 ª Edition
  • A Sedra
  • K Smith
Sedra, A. / Smith, K. Microeletrônica -5 ª Edition, 2007, Prentice Hall -Br.
Dynamic mechanical testing machines
  • J E Shigley
  • C R Mishke
  • R G Budynas
Shigley, J. E.; Mishke, C. R.; Budynas, R. G., 2004, Mechanical Engineering Design, McGraw-Hill, New York. Dynamic mechanical testing machines - Material Property Data - Material Property Data -