# How In-Line Bearingless Torque Sensors Can Help Increase Turbo Machinery Efficiency

The first phase of any effort to boost the efficiency of a piece of machinery in order to lower its operating costs and reduce power consumption is gathering accurate data on the machine’s initial performance, particularly torque data. The ability to measure torque reliably is crucial for applications ranging from determining the level of force required to lower an automobile’s window to testing a tugboat’s high-horsepower engines. The nature of turbo machinery can make it especially challenging to measure torque accurately. Electric motors, pumps, compressors and turbine engines can all generate high torques at high RPMs. In addition, torsional vibration can sometimes lead to premature failures in the driveline. Historically, this has created limitations on how torque can be measured both accurately and safely. This article addresses some of the basics associated with minimizing torque measurement uncertainty to allow making more informed modifications to a machine’s design or operating parameters. As every former first-year engineering student should recall, “torque” is defined as the amount of force needed to rotate an object about an axis, fulcrum, or pivot. The mathematical equation for torque is “force multiplied by distance.” To illustrate, imagine a foot-long lever arm attached to the center of a wheel; if hanging a one-pound weight on the end of the lever arm makes the wheel turn, the force needed to turn the wheel can be described as 1 lb-ft of torque.
However, ensuring accurate torque measurements isn’t always a straightforward process because a variety of physical influences are involved and there are multiple measurement methods, each of which has its advantages and disadvantages:
• Measuring Input Current to a Motor (approx. > 5% accuracy)—The main advantage of this method is that it is extremely inexpensive to implement; simply attach a digital multimeter to the cable powering the motor exerting torque. This is where the mathematics come in, requiring engineers to use the data acquired to calculate consumed electrical power, output mechanical power, angular speed, motor efficiency, and finally, the amount of torque produced. The biggest problem with this method is that it’s not particularly accurate, so it’s best used only for “quick-and-dirty” torque measurements.
• Strain Gauging a Rotating Shaft (approx. < 5% accuracy if calibrated)—As with the first method, this technique is inexpensive to implement because there’s no need to modify the driveline; basically all that’s required is attaching strain gauges to the rotating shaft. Acquiring the data from the rotating shaft typically requires a slip-ring assembly or telemetry system. The high number of variables and unknowns involved complicate calculating torque accurately based on strain gauge data.
• Reaction or Non-Rotating Torque Sensors (approx. < 0.5% accuracy)—The advantage of this sensor type is that there’s no need to cut the driveline in two to insert an in-line torque sensor, which makes it somewhat easier and less expensive to implement. However, the downside of this approach is that the measurements produced are far less dynamic than those produced by in-line rotating torque sensors because the data tends to be averaged by the mass of the dynamometer.
• In-Line Rotating Torque Sensors (approx. < 0.25% accuracy)—As might be obvious by now, this type of torque sensor offers the highest measurement accuracy and the most dynamic measurements, but it’s also the most expensive to implement because it has to be inserted into the driveline. In-line rotating torque sensors have been around in one form or another for roughly half a century, with a variety of technical advancements introduced to sensor design over the last few decades, including more choices on how to measure dynamic torque. Some of the newest in-line torque sensors support linearity and hysteresis errors of <0.03% of full scale with response times in excess of 3kHz.
The automobile industry has been using dynamic data acquired from in-line torque sensors for decades to improve powertrain performance, efficiency, and longevity by informing their design decisions. Over the last twenty years, bearingless in-line torque sensors (Figure 1) have become the standard for in-line torque measurement. A bearingless design allows for greater accuracy, higher RPM capability, and 24/7 operation with little or no sensor maintenance. A combination of digital telemetry and digital electronics helps to reduce noise, increase resolution, and improve torque sensor performance.
The use of titanium rotors allow for in-line bearingless torque sensor designs capable of up to 45,000 RPM, which would be well above the capabilities of a typical strain gauge-based torque sensor but necessary for applications like turbine fan blade testing. Today, custom torque sensor designs are helping to meet the growing demand for instrumentation suitable for characterizing high-horsepower turbine engines, with sensors capable of measuring torque greater than one meganewton meter (1MNm) now available.
Most bearingless torque sensors use foil strain gauges as the means of measuring stress (Figure 2). The strain gauges are configured and wired in what’s known as a Wheatstone bridge. A voltage (preferably an AC voltage to improve noise immunity) is applied to the circuit. This is known as an AC carrier frequency amplifier. When the sensor experiences mechanical torque, the resistance of the Wheatstone bridge changes, which outputs an analog voltage that is proportional to the torque applied. Because the rotor and stator are not in physical contact with each other, the analog voltage on the rotor is converted to a digital signal, which is then sent to the stator via a digital telemetry system. There are two antennas, one on the rotor that spins and one on the stator that is stationary. The stator receives the digital signal and converts it into any one of a number of usable outputs. If done correctly, this method of measuring torque can be very accurate and highly repeatable. However, there are limitations as to how fast the electronic parts can spin. Some bearingless torque sensors, such as the HBM T40B or T12, can also measure RPMs or angle of rotation with a reference pulse.
Another way of determining torque on a test stand (the reaction torque method mentioned previously) is the use of a lever arm and force sensor (Figure 3). As Newton's third law of motion explains, for every action there is an equal and opposite reaction. So, if the shaft is turning clockwise, there is an equal force (the reaction torque) that wants to turn counter-clockwise.
Figure 4 shows the contrast between torque data acquired using a lever arm and force sensor (blue line) and the dynamic data acquired by an in-line torque sensor (red waveform). The mass of the dynamometer acts as a 20Hz low-pass filter, generating what could be called an “average torque.” Essentially, dynamic data offers engineers more in-depth information on what is actually happening with the device under test. Comparing dynamic signatures after design changes are implemented allows gauging their effect on product performance and efficiency. The short end-to-end design of bearingless torque sensors makes higher torsional stiffness possible. Together with high electrical response times, that allows capturing torsional vibrations, which can be extremely useful in failure analysis, both in the test lab or on the production floor. When implementing a bearingless torque sensor, especially at higher RPMs, keeping the unsupported driveline as short as possible helps reduce the risk of hitting a critical speed, i.e., the point at which a rotating shaft becomes unstable (begins to vibrate harmonically). The goal is to avoid unwanted vibrations and shaft run-out, which can increase measurement uncertainty, as well as posing the potential for a catastrophic failure of the driveline. If the driveline is short and stiff enough, there might be no need for a support bearing (Figure 5).
Keeping most of the weight near the bearing of the power absorber reduces the “sag” in the driveline. If a torsional analysis shows a critical speed still exists in the testing RPM range, a support bearing may be needed. Given that a bearingless torque sensor is unsupported in the driveline, using a “dual flex” type coupling is generally recommended to remove any angular and parallel misalignments. It is very important to remove any parasitic loads that could damage the torque sensor during use. Parasitic loads, depending on the type and size, will also add errors to the torque reading. In summary, lowering operating costs, reducing emissions, and improving product quality are becoming more important every year, making the need for more accurate test data and more reliable test equipment critical. Fortunately, very accurate ways of measuring rotating torque in applications with higher operating speeds are increasingly available. Inserting a torque sensor directly into a driveline to capture and study dynamic data can help engineers improve the performance and efficiency of their turbo machinery. High response times make it possible to measure torsional vibrations that may lead to mechanical failures on the production floor. By eliminating parts that contribute to errors in the torque data, simplifying the driveline design can minimize test stand uncertainty. Bearingless torque sensors are one important new way to help to reduce data uncertainty while decreasing equipment maintenance requirements and, potentially, downtime.