Mechanical Testing: Accelerated Protocol Development and Advanced System Design

Medical Device & Diagnostic Industry Magazine
MDDI Article Index

Accelerated benchtop testing plays a prominent role in bringing new devices to market. With a well-reasoned protocol in hand, the test engineer can focus on actuator and controls selection— the two most important elements in designing a powerful test system.

Medical device manufacturers know that mechanical testing is a critical step in both the product development program and the FDA approval process. The test program manager must ask the right questions to ensure that accelerated test protocols meet the company's and FDA's requirements. These questions can be broken down into four basic areas:

For mature products, there are existing test specifications that answer many of these questions. For orthopedic devices, this includes knees (ISO 14243-1), hips (ISO 9325/ISO 15032/ASTM 1612), and spine (ASTM F1717). For prosthetic devices, specifications for lower limbs include ISO 10328.

When it comes to new product designs, however, there are few if any established specifications for many devices. Among new device examples are stents, stent/grafts, intraaortic balloons, ventricle assist devices, septal patches, and pacemaker leads.

WHAT PARAMETERS SHOULD BE TESTED?

The first step in developing the test protocol is the failure mode and effects analysis (FMEA). The FMEA is done to identify all of the possible device-related failures that can occur because of implanting the device in the body. For example, Table I shows a partial list of potential failure modes and potential tests that were identified by the AAMI/ISO TG150, SC2, WG31 committee in developing their working document for endovascular devices.

Although an FMEA offers a somewhat negative perspective, it does provide a comprehensive means for determining the appropriate types of tests and their parameters.

HOW EXTENSIVE SHOULD TESTS BE?

A firsthand measurement of what happens physiologically is the best starting point. For orthopedic applications, this means combined force plate and motion data can be used to kinematically determine the applied joint loads. The literature has also offered cases where instrumented prostheses have been used to measure the loads in a more direct manner.

For intravascular applications, the use of x-ray angiography, pressure catheters, or intravascular ultrasound provides a good means for determining the radial dilation and pressures that occur during each heartbeat or extraneous movement. Combining measured data with finite element modeling provides a better understanding of the test parameters that must be generated.

A thorough literature search is paramount to the development of test protocols. Information provided on the FDA Web site about previously approved devices can be useful in developing test protocols. Published papers and articles about applied loading are abundant in the orthopedic area, while some articles are beginning to trickle out from the intravascular field.

Physician testimonials can also be used to characterize what happens physiologically. These highly trained individuals are often able to provide accurate and comprehensive observations about what takes place in the human anatomy.

Test protocol developers need to apply good engineering sense. Once the data have been collected and synthesized, a degree of judgment must be used to apply it to protocol development. Testing a device to the lowest end of the data spectrum will not ensure it makes it through the approval process.

The 9100 series stent/graft tester uses laser-based measurement technologies and a WinTest control system.

On the other hand, the protocol should not be made unnecessarily difficult. For example, there have been many arguments about the application of combined radial and bending loading to peripheral stents. Should considerable time and money be invested in developing a combined testing system, or should the modes be separated into simpler elements that can be tested on different testers? Depending on the engineering perspective, an argument could be made for either technique.

HOW MANY TEST CYCLES SHOULD BE RUN?

The number of cycles required for testing depends on the expected service life of the product and what could happen as a result of a structural failure. Some devices used to alleviate complications in terminally ill patients do not have a long expected service life. Generally their service life is measured in months. Most load-carrying devices such as orthopedic and intravascular implants, however, have an expected life of 10 years. Heart valves and pacemakers have an even higher expected life of 15 years.

What is the implication? Many load cycles are needed (10 million cycles for orthopedic, 400 million for intravascular devices, and 600 million for heart valves and pacemakers). In each of these examples, the historical precedent has already been set, and it would be difficult to argue that it should be changed. Many new medical devices are placed into the above categories because of their similarities.

WHAT SHOULD TEST FREQUENCY BE?

Opinions regarding the appropriate test system frequency range from at the physiological rate to 80 times the physiological rate. Because speed to market is often a critical factor for success, test frequency should be maximized while ensuring that physiological levels are maintained. Oftentimes a precedent has already been set. Table II identifies some common precedents.

For devices without appropriate testing precedents, at what frequency the test should be run remains a question. A list can be compiled to outline the factors that could limit the performance frequency of the test. Such factors might include:

Another consideration in determining the test speed is what the data will be used for. If the data are to be used for regulatory submissions, careful consideration should be given to test frequency. If, on the other hand, a new design's potential is being explored, faster speeds provide feedback that is more timely. The following methods can be used to verify test performance at higher frequencies.

COMPARISON OF STATIC TO DYNAMIC FORCE/DEFLECTION MEASUREMENTS

Figure 1 demonstrates the stress/strain or force/deflection response for a hypothetical medical device. The figure at left shows the specimen response to an applied load at a physiological rate. The graph indicates that the device behaves elastically at that rate. At a higher rate one begins to see a lag between the applied stress and strain. The dashed lines in the figure at right indicate a difference in stress, depending on whether the strain is increasing or decreasing. As long as the difference is less than a few percent of the applied variable, the test can be considered acceptable.

Figure 1. Specimen response to an applied load is illustrated on the left; the dashed lines of the figure to the right indicate a difference in stress, depending on whether the strain is increasing or decreasing.

MATERIAL RESPONSE MEASUREMENT

Table III shows the response of a viscoelastic material subjected to sinusoidal loading at different frequencies. As the test frequency increases, phase lag between the applied stress and the resulting strain also increases. E* is the measured modulus while E' and E'' are the real and imaginary components that can be calculated using the measured phase lag. The table indicates that there is little difference between the measured and real modulus at the higher frequencies. This implies that the specimen's properties do not change much over the applied frequency range.

PHYSIOLOGICAL VERSUS STROBOSCOPIC MEASUREMENTS

In tests involving a measurable motion, it can be useful to compare the motion of the accelerated test with that of the physiological test using either stroboscopic measurements or high-speed overlays. If the high-speed motions can be shown to be equivalent to the physiological-rate motions, an argument can be made for running the test at the higher speed. Higher test rates have been justified using this verification method.

After a test protocol has been determined, the appropriate test instrument must be selected. This process entails selection of the correct actuator type and control considerations for the test system design.

ACTUATOR SELECTION

There are a number of actuator types that have been developed for test systems used in a wide array of applications. These include servohydraulic, servopneumatic, electrodynamic, servomotor, and stepper motor. There are advantages and disadvantages to each approach.

Servohydraulics. This type of actuator has been the primary driver for fatigue-testing applications since the early 1950s. With this approach, high-pressure (3000-psi) hydraulic oil is supplied to a servovalve that ports the oil to either side of a hydraulic cylinder, causing motion.

An advantage of the technology is that it can be scaled to a wide range of forces (several thousand tons). Although the typical frequency response of most hydraulic systems is in the 30-Hz range, some high-end systems can be run to several hundred hertz. Disadvantages of using servohydraulics are infrastructure requirements (for hydraulics, cooling, and power distribution), high energy requirements, and maintenance. In some high-cycle biomedical applications, seal and rod wear and the potential for oil leaks may pose problems.

Servopneumatics. These actuators were introduced to test applications about a decade ago to provide higher resolution a lower price than servohydraulics. With this approach, air (100 psi) is ported through a servovalve to a pneumatic cylinder to cause motion.

Advantages of the approach compared with hydraulics include reduced infrastructure (runs from existing shop air systems), cleanliness, and higher efficiencies for low-frequency and low-force applications. The systems are also more portable and seem less intimidating because of their lower operating pressures. Disadvantages include lower performance (less than 10 Hz) and, in high-cycle applications, the same rod- and seal-wear issues associated with hydraulics. There is also a crossover point at which pneumatic drives become less efficient than hydraulic drives for high-frequency, large-displacement applications because of the compressibility of the air medium.

This servopneumatic actuator has been designed specifically for low-force fatigue testing.

Electrodynamics. Although linear electrodynamic drives have been used over the last 50 years for vibration and shaker applications, they have been applied more recently to fatigue testing. Field windings are used to generate an electromagnetic field. When placed in the presence of a secondary, fixed magnetic field, the windings generate a force that can be used for mechanical testing. Many new possibilities are being explored for applying electrodynamic actuators to test applications.

There are many advantages to using this approach rather than servosystems, among them higher-frequency performance, better force fidelity, elimination of the oil and cooling infrastructure, cleanliness, longer endurance, minimized maintenance, and portability. The only present disadvantage is that the force level is limited to several hundred pounds. With the advent of higher-efficiency amplifiers and advanced motor designs, force capacities are expected to increase.

Rotary Servomotors. These actuators operate on the same basic principles as electrodynamic drives. Most often they are to ball screw drives to create linear motion. In that they can be found in many tensile testing systems. A growing aspect of fatigue testing involves the use of rotary in direct-drive applications to create high-frequency testing devices.

Although rotary servomotor screw-driven systems offer economical force generation, they are typically slow and do not perform well in cyclic applications. Rotary servomotor direct drives have high frequency response and long endurance. Because they are optimized for high RPM speed, typical output torque is limited. Newer high-torque, low-RPM drives have been developed that hold great promise for rotary fatigue testing.

Rotary and Linear Stepper Motors. These types of actuators use an electromagnetic design that is similar to the rotary. Whereas a servomotor uses six field poles that are commutated using analog voltages, a stepper motor has 50 poles, which are commutated using binary signals.

The primary advantage of using stepper motors lies with their resolution (1 revolution = 50,000 counts). They can rotate with great precision and are well suited to applications involving microscopes. Stepper motors also require no additional infrastructure and are very efficient to operate. One disadvantage is that they do not develop much speed or torque. This precludes their use in cyclic or high-speed applications.

CONTROL CONSIDERATIONS

Most advances in mechanical testing over the last two decades have resulted from the application of computer controls to testing. Expensive minicomputer system add-ons have given way to fully integrated WinTest-based systems. Although there are many possibilities for control system design, the following areas should be considered when selecting a control system for a particular test application.

Transducer Readout. Most control systems have been designed to interface with transducer systems that provide analog outputs. Examples of analog transducers include load cells, LVDTs, extensometers, and potentiometers. Many transducers are now available with digital output. This type of transducer optical encoders, laser measuring devices, and temperature transmitters. A control system that can interface easily with these devices provides advantages when configuring test systems.

Data Acquisition. Traditional data acquisition methods the "snapshot"—derived from a series of data points taken over a defined period of time. While this technique is useful for verifying the fidelity of the output signal, long-term trending is better determined by sampling peaks and valleys of the waveform.

Other advanced data acquisition approaches include taking data when a predefined signal level is crossed or a preset event occurs (load dropoff during specimen failure, for example). Timed data acquisition on a logarithmic cycle count is also useful because many work-hardening effects occur during the first several load cycles. There is then little or no change until specimen failure occurs. A control system that provides level crossing and event triggering and logarithmic cycle counting provides the most power when performing a test.

Controlling the System. Most control systems feature proportional, integral, and derivative (PID) control. While PID loop closure is adequate for static or low-frequency tests, high-performance tests often require special control algorithms to achieve greater accuracy.

In test applications where more than one actuator applies load to the specimen, crosstalk can occur. A control technique where the feedback from one actuator is fed into the control loop of other is useful in canceling the cross-coupling effects. In other high-velocity applications, it is desirable to forward feed some of the command signal into the control loop. This provides the system with an extra "kick" when high accelerations are required.

Another element of control is waveform compensation. In this approach the controller automatically adjusts the applied waveform to compensate for test system changes. An advanced technique involves compensating across varying end levels that occur in physiological or random load sequences. This requires the controller to track and categorize the signal changes according to start and end level. A control system that provides advanced control algorithms and compensation routines will provide greater performance for test applications.

Waveform Generation. The waveform generator defines the load-time profile that is applied to the specimen. Simple waveforms include the sine, triangle, square, and ramp. More complex or physiological waveforms can be created by connecting individual waveforms together in a block that can be repeated as required. More advanced simulations often require the controller to play back a previously recorded signal. To facilitate this, the signal is processed to create a digital file that replicates the original signal by using frequency and end levels. A control system that can read a drive file in addition to generating simple and block waveforms provides greater flexibility for test system development.

Driving Multiple Actuators. As mentioned earlier, test systems can be built using a variety of actuators, including hydraulic, pneumatic, and electric drives. The control system needs to be adaptable to accommodate each of these different actuator technologies.

Advanced tests also involve subjecting the specimen to multi-axis load states. Specimens requiring multiaxis loading include tissue-engineered heart valve leaflets, septal patches, bone screws, and other devices. Control and synchronization of multiple actuators is a requirement in performing these tests. While many tests are satisfied with only two actuators, more-accurate simulation is often provided with a greater number of actuators. A controller that interfaces to multiple actuator types provides greater flexibility in performing accurate simulations.

Application Programs. Most accelerated tests are performed on a checklist basis. The operator runs through a series of sequential steps to set up and run the test. Once the test is running, the operator sets the system to capture data for subsequent analysis and report generation using a secondary program (such as Lotus, Excel, or MatLab). To expedite the checklist, more advanced control systems enable the operator to save the test setup to a file that can be quickly recalled to repeat the test. A higher-level approach involves prompting the operator to perform certain actions, such as energy source activation and specimen insertion, which cannot be stored in the test setup. This makes it easy for inexperienced operators to accurately repeat the test without forgetting some minor step along the way.

The highest form of operator assistance involves the development of an application program that performs the test system initialization, prompts the operator to load the specimen, automatically records data throughout the test, and performs analysis and report generation upon test completion. To achieve this, the control system requires an embedded programming language that enables the operator to quickly generate customized application programs. A control system with this type of environment will increase power for advanced testing applications.

CONCLUSION

There are many steps in developing protocols and systems for accelerated benchtop testing. Test protocol development begins with an understanding of the physical parameters that must be tested followed by a determination of how rapidly the test can be accelerated without compromising data integrity. Once protocols have been established, a test system can be built using a variety of actuator and control technologies. Understanding the advantages and disadvantages of each available actuator technology is critical in determining whether it is appropriate for the test application. Once the actuator technology has been chosen, it must be joined with a control system that provides the power and flexibility required by the test application. Following this multistep approach to protocol development and test system design will provide an end result that is optimal and provides high value for years to come.

Kent S. Vilendrer is the president and CEO of EnduraTEC Systems Corp. (Minnetonka, MN). The company manufactures fatigue systems and provides laboratory services for testing the durability of medical devices and advanced biomaterials.