Electric motors are the workhorses of the industry. They are everywhere. Failure of an electric motor often means that an industrial process comes to a complete standstill. The monitoring system SAM4 from Semiotic Labs monitors the condition of motors and rotating equipment. It signals damage at an early phase so that customers can plan maintenance before production machines fail.
‘In 2015 I read about an algorithm that mimicked the brain. That got me thinking. I put some ideas on paper and discussed them with classmate Gerben Gooijers. We knew that an enormous amount of data would become available in the industry. We wanted to design algorithms to create value from that data, and then keep a share of that value for ourselves. We had it all thought out: dear customer, give us your data and, with our algorithms, we will make mountains of gold out of it. And then there will be world peace. But of course it didn’t exactly go like that.’ Simon Jagers, COO of Semiotic Labs, clearly enjoys talking about the company he founded with Gerben Gooijers.
‘In the beginning we had varying success. Not all data are valuable. Many customers do not want to make their data available; they see that as giving away business information. And there is a enormous variety of data. If you want to be active in such an environment, you have to position yourself as a consultant, while we really wanted to make products.’
Eventually the choice was made for condition monitoring of electric motors. ‘Such a motor either works or it doesn’t, and that’s no sensitive company information. It was not difficult to find partners who didn’t mind us monitoring their motors. And there are an incredible number of electric motors. Did you know that more than 40% of all electricity in the world is consumed by some 350 million electric motors? That is one really large potential market.’
Vibration versus current
There are various ways of collecting data about the condition of electric motors. The usual method is to measure vibrations. Deviations in frequency and amplitude often mean that there is a threat of something going wrong. At first, Semiotic Labs decided to work with existing vibration sensors.
‘Vibration sensors take quite a pounding. They must be installed on the motor. They also often suffer from environmental factors such as dust, moisture and other influences. And everything that sticks out from a motor can serve as a step for employees to reach something. I was driving around every weekend, replacing broken sensors. Vibration sensors offered valuable data, but there were many operational challenges.’
An alternative method for condition monitoring is analysis of the electric current. For Semiotic Labs, this proved to be more attractive. Simon Jagers explains: ‘By opting for current analysis, we have a unique system with many operational advantages. You don’t have to install the sensors on the motor, because they can be installed in a control box. There they are protected from external influences. It also offers added value – you can offer condition monitoring and provide insight into other parameters, such as runtime, energy use and the load of the motor and assets.’
Smart Asset Monitoring 4
The sensor Semiotic Labs has developed with Technolution forms the heart of the system of Smart Asset Monitoring 4 (SAM4). It measures the current and the voltage on the motor’s cables with a high frequency. The measurement data are sent to the cloud via an internet gateway. There, the self-learning algorithms of Semiotic Labs determine the status of the electric motor on the basis of fluctuations in the current and voltage. The status data are displayed on the user’s dashboard or forwarded to the existing control system. The user also receives insight into such data as operational time, energy use, load and capacity.
Data as raw material
For Semiotic Labs, it was important to maintain control of the sensor development.
‘With condition monitoring, you supply reliability. Peace of mind. Our strength is in Artificial Intelligence (AI). If we have good data, we can use AI to create valuable information from it. This allows us to predict motor failure and even determine the cause of the failure. This is why we wanted to develop our own sensor system. In this way, we maintain control over our own raw material – data – and we can guarantee the quality of the service and provide more reliability and peace of mind. After all, we are the creators of the raw material and the developers of the analyses. So it is essential that the signal of the sensors is of a high quality,’ says Jagers. ‘Anyone can measure electric current and voltage, but can you also do it with a very high frequency and purity? And with good manufacturability and realistic costs? For these reasons, we looked for a technology partner with in-depth knowledge of current analysis, one which could produce the required electronics quality and who could support us in areas where we had no knowledge, such as product life cycle management.’
Interesting first meeting
After some detours, Semiotic Labs came in contact with Technolution. ‘The first meeting was “interesting”,’ relates Simon Jagers with a grin. ‘I was talking with the CEO, Jan van der Wel. He questioned me rather critically. If you touch my baby, I can respond quite sharply, although Jan was just asking good questions. But, okay, people learn by doing. There was a click between us. When the specialists of Technolution finally came in, they had distinct ideas about the technology we were proposing. We were flabbergasted. They really knew what they were talking about. We felt that we could develop reliable electronics with Technolution. And that is what happened. Technolution provides us the peace of mind which we provide our customers.’
Despite – or maybe precisely thanks to – the turbulent start, the collaboration between Semiotic Labs and Technolution produced quick results. The system, SAM4 (see box), is running successfully at various large customers such as Vopak, Heineken and Tata Steel.
‘Our algorithms are continuously learning. We can now identify causes of faults which have not yet occurred. Eventually, we want to sell uptime: Asset Management as a Service. To do that, your system must be very reliable. You could even look at a Pay-per-Use model, where the customer doesn’t pay for the machine but per item which the machine produces.’
Simon Jagers sees plenty of opportunities. Semiotic Labs is collaborating with Technolution on the further development. Cheaper sensors, larger volumes and a broader market share are the key objectives.