AC drives emerge as entry point for industrial digitalization
In ancient Greece and Rome, the pump was an essential mechanical technology, just as it is today. Yet even the very oldest technologies are being impacted by the emergent Industrial Internet of Things (IIoT) and the digitalization trends IIoT supports.
Pumps, as part of IIoT networks, and the motors and drives that move those pumps, are being equipped with more sensors, data-acquisition capabilities, and computational resources. But what is the best way for this to happen?
It’s been noted that the drive is the first point at which digital technology can be applied to a mechanical system, such as a pump, in a production environment. The drive can act as a kind of edge server or gateway that orchestrates the data that enables predictive maintenance or process optimization applications.
A variable frequency drive (VFD) controls the rotational speed of an alternating-current electric motor and thereby the flow and pressure of a pump, eliminating the need for a throttling valve.
The number-one cause of motor failure is the bearings. Motor bearing failure can bring just about any production process to a screeching halt. Maintenance-wise, for those in the know, just the sound of a motor can indicate to them an impending main bearing failure.
Motors can be equipped with vibration monitoring to detect impending bearing failure, as well as with the means to shut down a failing motor. However, because momentary vibration increases are not necessarily indicative of impending failure, reading only a single sensor value makes false positives possible.
Filtering can mitigate these type challenges, but even better, applying computer machine learning techniques can ensure not only that transients are eliminated, but that any significant trends detected are escalated for attention as soon as possible.
In other words, while management of pump and motor assets has been well understood for years, computing power helps decide when to take action, based less on the operators’ intuitive insight and more on quantifiable facts.
Other parameters such as how long the motor has been running, temperature—a very important indicator—and load, which may vary with the pumped fluid’s viscosity, can be factored in. The influence of these other factors can be calculated so that alarms are not strictly state-conditioned. The cost of the additional computing power is tiny compared to the cost of motor failures in an industrial setting.
Digital twin use is another way to move pump operations beyond strictly state-conditioned monitoring of sensor values. With equipment designs documented as a 3-D solid model, empirical operating data can be associated with it. The digital twin enables deeper insights into equipment operation. Technologies for building digital models and populating them with data are offered by at least several software suppliers.
Implementing IIoT can be data intensive. In practice, measuring parameters that include temperature, pressure, flow, vibration, and power, can generate 2.5 megabytes of raw data per second. Bandwidth for sending this data to the cloud can be expensive. Processing that data at the edge helps by compressing the data to a small fraction of features that can be easily sent to a local server or the cloud.
In today’s world, controllers detect when a parameter varies from its expected set point value. However, as mentioned, this can lead to transient alarms and unnecessary downtime. A more efficient method uses machine learning capabilities associated with the digital twin to recognize data patterns across parameters to determine whether anomalies are problematic and, if so, predict time to failure.
In a refinery with 150 pumps, installing servers or gateways at each pump can be expensive. Alternatively, next-generation drives can send pump data to an IIoT system, such as PTC ThingWorx. This reduces implementation cost and provides data to optimize fluid flow and reduce down time, while improving overall yield.
Industry has a good understanding of digitalization’s inherent possibilities, but consensus is lacking about how to proceed to implementation. Those well-versed in computing want to be involved in Industrie 4.0 but don’t necessarily have the required industrial domain expertise. To say it in plain English, IT departments aren’t always well-suited to dealing with remote operations in adverse environments nor do they necessarily know how to implement instrumentation for data acquisition.
As a result, in some cases IT providers seem to recommend users build a kind of redundant, secondary network devoted to acquiring the data needed for predictive maintenance and process optimization.
This is an expensive, high-risk approach, and doesn’t mark out a clear path from basic control to full-throated optimization functionality. On the other hand, drives equipped with extra computing power can be a major part of a “clean” solution, especially for pipelines, in process industries, or wherever motors are a big part of a plant’s energy consumption.
While some OEM equipment manufacturers balk at the cost of adding multiple sensors to their products, bringing motor-pump combinations into the IIoT world doesn’t mean applying a lot of extra sensors. The drive itself already has capabilities for analog measurement of pressure and flow, as well as 3-phase current transducers, with vibration inputs to be added in the near future. Many motor and load behaviors are understood from the current waveforms recorded in the drive.
Things like the ATEX protocol require six thermocouples to be installed in the motor stator. This provides a good window into a motor’s heat profile and the petrochemical industry has used these protocols for years in order to deal with applying electric motors in explosive atmospheres. Sensor and bearing manufacturers, recognizing the need for more complete information from the bearing, are innovating a broad array of solutions.
More bearing data is available than ever before, from MEMS, piezoelectric, and accelerometer-based vibration sensors that are installed externally, to some newer integrated designs. For bearing sensors, higher resolution analog-to-digital conversion is required, putting demand for more instrumentation and control resources. Integrating the needed interface into the drive directly is the most cost-effective and efficient way to build the bridge toward IIoT.
Operators tend to see the motor and drive as a black box: “you put electricity in and get mechanical work out.” At that level, no one really wants to deal with the complexity of what’s inside the box.
On the other hand, drives manufacturers like Danfoss have a deep understanding of the electrical and magnetic interactions that occur between a motor and drive.
By paying a little extra attention to the insights that additional drive data provides, significantly reduced cost of operations can result. And in the end, that is value that customers are looking for as we move into the IIoT connected world.