Understanding accelerometers for vibration monitoring and predictive maintenance

Day-to-day manufacturing relies on equipment availability to ensure continuous operations with minimal downtime. Due to hostile operating conditions that generally involve high-pressure, high-temperature or magnetic environments, machinery failure incidents may happen and interrupt operations.

Machine maintenance regimes regularly schedule inspections for wearing equipment and to replace them regardless of their potential to operate for a much longer time. This approach proves to be extremely costly due to additional capital investment and production loss resulting from a total shutdown. There is also a possibility that machinery failure might occur between the scheduled maintenance times, which further adds to the cost.

With the availability of condition monitoring sensors and technology, we can better plan for maintenance by understanding the current state of the machinery. Predictive maintenance is a technique that involves regular monitoring of machine conditions such as vibration, temperature, oil condition, etc. It allows the operators to predict potential failures and helps them to prioritise the maintenance of relevant parts before they break. Predictive maintenance aims to predict when equipment failure could occur based on historical data, reducing the likelihood of failure, lowering maintenance costs and increasing the annual plant availability.

Any operating machinery generates vibration due to its linear or rotational motion. Problems arise when the measured vibration is trending higher than the baseline measurement. It generally indicates potential faults, which may include misalignment of drive components, load asymmetry or damaged bearings.

Vibration monitoring is therefore commonly done as part of the predictive maintenance approach as it is non-invasive and can identify faults at their early stage of deterioration. It can also diagnose the severity of the damage and find the source and leading cause of failures.

Vibration measurement sensors such as accelerometers, vibration switches and mobile vibration meters form part of a condition monitoring system for predictive maintenance. The complete systems consist of sensors, sophisticated data acquisition systems and powerful monitoring software with complete analysis packages integrated into the whole plant.

Types of accelerometers

There are two types of accelerometers: AC-response accelerometers and DC-response accelerometers. AC-response accelerometers are commonly designed from piezoelectric materials that produce an electrical output proportional to the acceleration. The sensing element looks like a source capacitor with a finite internal resistance which forms the RC time constant. For this reason, AC-response accelerometers are not suitable for measuring static acceleration such as gravity or centrifugal acceleration. DC-response accelerometers can respond down to 0 Hz, making them ideal for measuring static and dynamic acceleration.

Vibration monitoring systems for condition monitoring usually rely on AC-responsive piezoelectric accelerometers due to their wide range of frequency responses to capture almost all machine vibration trends. However, MEMS capacitive DC accelerometers have started to gain popularity due to their flexibility and lower cost.

Modern MEMS accelerometers are closing the gap with piezoelectric accelerometers in terms of operating bandwidth and dynamic range, making them suitable for predictive maintenance applications. Compared with piezoelectric sensors, they offer faster signal recovery after a shock event, extremely stable measurement, digital outputs and smaller size. MEMS accelerometers also have a built-in signal conditioning unit and potentially can be used to build intelligent, wireless, vibration-based condition monitoring systems.

Although MEMS accelerometers have potential as an excellent alternative to piezoelectric accelerometers due to their capability for future upgrading and signal conditioning capability, users should choose according to their requirements and after conducting successful validation tests.

Signal processing in a MEMS accelerometer

Vibration analysis starts with pre-filtering a slight DC bias from accelerometer and velocity data to suit the time-domain and frequency-domain analysis. The signals are then passed to the high-pass filter to isolate the dynamic component of the data and clear the DC components from the signals. A numerical integration technique can be applied to this acceleration data to derive the speed component in a few iterations.

Once the data is pre-processed, analysis of acceleration and speed data over time can provide insights on the machinery performance. Maximum peak value and RMS parameters can indicate warning conditions as they approach the threshold limit set by the equipment manufacturer.

Frequency domain analysis is commonly done using the fast Fourier transform (FFT) method. It can identify the root cause of the problem by examining the frequency and amplitudes of the relevant spectrum. For example, a high vibration amplitude on the rotation speed can represent motor imbalance, while vibration at specific frequencies can indicate bearing degradation and failures on the gear mesh.

Cabling and mounting

Cables and connectors are usually the weakest part of a measuring system. Choosing a suitable cable is vital to ensure reliable measurement. Bending or tension on a coaxial cable can generate a signal that cannot be distinguished from sensor output, which produces erroneous results, especially when measuring low vibration with charge-type transducers. This problem can be solved by using a low-noise cable or clamping the cable on the test object.

Knowing the proper mounting techniques is as important as choosing the right type of accelerometers for condition monitoring applications. Condition monitoring accelerometers generally yield the best results when mounted on a smooth and flat-machined surface. The transmissibility of high-frequency signals can be further improved by improving the contact between the sensor and the mounting surface by applying a thin layer of silicon grease.

Closures

Every aspect of the measurement chain in rotating components is potentially a vibration monitoring point. Vibration measurement as a condition monitoring approach can effectively diagnose potential failure events and indicate operational health in rotating machinery; however, accelerometers must be correctly installed and integrated with a suitable signal conditioning and measurement system to filter unwanted signals. For example, the sensors are installed on the housing of the bearings or nearby points for vibration measurement. They should be configured to allow 3-axis measurements at all bearings. The results should be referred to the reference value recommended by the ISO standards to assess the health of the machinery. The reference value differs for different machinery types: for example, the vibration of a reciprocating engine is generally higher than that of rotating machinery.

Image credit: ©stock.adobe.com/au/Monopoly919