The importance of data is now starting to seep through to the manufacturing industry; they can therefore provide companies with a lot of valuable information. Semiotic Labs goes a step further and uses data combined with artificial intelligence to predict machine failure. Maintenance can thus be done when it is really needed and that makes a huge difference to costs.
It costs companies a lot of money when production processes have to be shut down for preventive maintenance of machines or fixing breakdowns. Therefore, companies benefit greatly when downtime is reduced as much as possible. This is possible with technology from Semiotic Labs. By combining smart sensors and self-learning algorithms, this startup predicts when and why machines or machine parts are going to fail. In this way, maintenance can be scheduled at the ideal time: before machines fail or when performance deteriorates, but not before. The solution is suitable for machines driven by electric motors.
Algorithms
Founder and COO Simon Jagers found inspiration for Semiotic Labs in US electrical engineer Jeff Hawkins' book On Intelligence, about an algorithm that mimics the brain. "I then started looking around to see what possibilities this could offer," he says.
Jagers did not just jump into the deep end; before setting up a company for good, he first called a number of companies to ask if they would be at all interested in the technology he had in mind. They all turned out to be keen to sit down with Jagers. Then he started fleshing out the concept with his business partner Gerben Gooijers and they started working on building the algorithms.
Service
Semiotic Labs provides a so-called 'end to end' solution for smart condition monitoring that consists of a sensor, algorithms and an online user interface. Based on the solution, customers gain insight into the condition of their machines. Together with partners, they provide a full-service solution in the field of (preventive) maintenance, combining the monitoring solution with partner maintenance services that provide proactive maintenance when the condition of assets calls for it. "We therefore call it Uptime as a Service. In practice, in a nutshell, it works like this: we install a sensor in a clean dry place (control cabinet or engines); this takes about 15 minutes per engine. We then send the data by wifi or cable to the cloud. Time is then needed to learn what the specific quirks of the engine are. That takes about a month now but that time is getting shorter. From then on, we can tell the customer what is the condition of the motor and the rotating system behind it."
"The problem you often see with electric motors is that they are maintained based on time: The 'mean time between failure'. It looks at the average time between failure events. That average is then used to schedule maintenance intervals. This prevents a lot of unplanned downtime. But it also ensures that a lot of maintenance is carried out far too early. Moreover, opening up 'healthy' machines carries risks: performing maintenance itself can also cause damage to machines. Our technology determines the condition of assets very accurately and if damage has occurred, we try to estimate when this translates into unplanned downtime. This allows you to schedule maintenance at the right time; when there is damage or when performance deteriorates, but not before. There is less unnecessary maintenance and you avoid unplanned downtime because you are ahead of damage."
Pay per use
In fact, all companies for whom the cost of unplanned downtime can be high can benefit from Semiotic Labs' technology. Jagers: "The biggest gain is in increasing operational reliability at lower costs. The condition of machines is accurately determined, which means you do less frequent, but more relevant maintenance.
It also suits the current phase the industry is in; it is increasingly looking for pay per use models. A customer pays the supplier not because he has a machine, but for its use. For the supplier, this means that the risk of downtime falls on him. He wants to reduce that risk by being able to better determine which machines need maintenance and when. An additional advantage of this model is that the interests of the supplier and the asset owner become equal, because the supplier no longer gets paid for providing services or parts, but for a correctly functioning machine."
"Because we use artificial intelligence, we can also say more about the relationships between machines. We get to know the specific quirks of motors and rotating equipment and automate the unique making of the baseline of healthy behaviour for each machine. So, for example, we know that a particular machine breaks down if you use more than a certain amount of power. We also learn more about pump and motor combinations and, for example, the influence of climate. The sea climate in the Netherlands has a different effect on a machine than a cold mountain in China. So in the end, the more systems we monitor, the better the solution will be for everyone."
A number of companies are now using Semiotic Labs' service. Based on these experiences, the software and hardware will be optimised. "We are now testing the technology in the field on some scale, to demonstrate efficacy in practice as well as to further improve the product. In early 2018, we will roll it out as a commercial product."
By: Kelly Bakker
Source: Semiotic Labs
