Virtual Sensor opens a World of Efficiency for Large Motors
A virtual sensor developed by Siemens researchers calculates the temperature of a motor's interior without the aid of a single sensor installed there. The resulting information makes it possible to prevent unnecessary downtimes – an improvement that could dramatically lower operating costs.
by Aenne Barnard / Sandra Zistl
They are the size of rooms, but they are still hard to access: the rotors of large electric motors such as those used to compress gas. Such rotors experience high thermal stress when started and can suffer damage if overheated. As a result, their temperature must be monitored.
The motors that drive such rotors – so-called salient-pole motors – are huge machines that are used to pump tremendous volumes in the gas, oil and chemical industries. When being started, they generate an enormous amount of heat as they are connected directly to the electrical grid – that is without a frequency converter. If they were repeatedly started, their interior temperature could shoot to as much as 800°C, thereby potentially causing serious damage. The motors must therefore be allowed to cool before being restarted. The question is: how long? The temperature in critical areas of the motors’ interior cannot be directly measured. As a result, technicians have had to estimate the length of cooling time – that is, up to now. Normally, experts build in a safety buffer that rules out the possibility of damage. This frequently results in down times of up to 12 hours that are much longer than actually needed and results in significant lost operator income.
Now, thanks to work performed at Siemens Corporate Technology (CT), researchers can measure and monitor the interior temperature of a motor during operation using a virtual sensor – a development that could lead to significantly reduced downtimes and improved facility utilization. The mathematical model of the prototype virtual sensor is based on a digital twin – an exact, functional simulation of what a real sensor would do if it were possible to insert a sensor into a motor. A look through a HoloLens – an augmented-reality headset – at a demonstrator of a motor allows a viewer to see an exact simulation of the motor and its interior with a real demonstrator superimposed over it. Color codes from blue to red illustrate temperature levels.
The new models developed by Siemens researchers enable a valid conclusion to be drawn regarding the rotor temperature of a motor.
“We have drawn on work performed by our colleagues at Process Industries and Drives, particularly the motor factory in Berlin,” says Birgit Obst, a simulation expert at CT. “While developing motors, they use mathematical models that capture the geometry and material characteristics of drive units in order to create a digital twin of each component.” But these models are so extensive and complex that they normally cannot be used for real-time calculations. In an effort to resolve this challenge, Siemens researchers have made two significant strides: They have succeeded in using mathematical reduction processes and in deriving abstract models from them that may be less comprehensive but still yield essential results. These models can be calculated 1,000 times faster and with fewer and controlled deviations in precision than in traditional simulation tools used in engineering. The result has been the development of digital twins that can be constantly monitored during operation. Such twins are capable of providing a virtual image of reality at any time. Quantification of the precision of such virtual sensors is determined by comparing the data they generate to that of sensors on non-moving components.
The new models developed by Siemens researchers enable a valid conclusion to be drawn regarding the rotor temperature of a motor. “Figuratively speaking, you can think of this as a weather forecast,” says Dirk Hartmann, coordinator of CT’s Simulation and Digital Twins core technology. “We can now measure – in other words calculate, the temperature at particular places – in this case, the rotor. Furthermore, through the combination of available data from measuring points – the virtual equivalent of weather stations – we can develop a forecast for all areas; not just those being measured.”
This ability can save operators lots of money. “Such an optimized process that can prevent motor overheating and reduce the downtimes required during the cool-off phase can save up to €210,000 per hour,” says Artur Jungiewicz, a developer at Process Industries and Drives in Berlin. Another special characteristic of the simulations is their speed: The direction of temperature changes can now be monitored in real time and predicted.
A table-size demonstrator illustrates just how Siemens’ simulation system works: Two small electric motors are coupled via a spindle. The motor on the left slows the motor on the right and produces a continuous load in the process. Sensors measure the temperature outside the motor. At the same time, data regarding the length of operation and load are collected. The simulation system applies these input parameters and an associated mathematical model of the motor to calculate the temperature within the right-hand drive unit to forecast temperature progression.
“Our virtual sensor is really precise. It is almost as if it were measuring the temperature directly,” says Hartmann. “The prototype measures current conditions and predicts a point in time when the motor can be restarted.” The underlying models from engineering serve as “a valuable foundation of know-how that separates us from the exclusively data-based processes used by competitors,” says Obst. “The integrated chains of engineering models leading up to online simulation and calibration are unique selling propositions,” adds Hartmann.
Aenne Barnard / Sandra Zistl
Subscribe to our Newsletter
Stay up to date at all times: everything you need to know about electrification, automation, and digitalization.