How Russian IT solution made power plants failure-free
Text: Marat Gazizullin (Izhevsk)
Russia’s first agreement on the implementation of the predictive analytics system developed by ROTEC JSC at network infrastructure facilities was signed in Izhevsk. Rosseti PJSC plans to introduce in Udmurtia a new software product called the PRANA System, which allows accurate predictions of equipment failures and thereby improves the power supply reliability in the region. The distinguished machine builder of the Russian Federation, Chairman of the Board of Directors of ROTEC JSC Mikhail Lifshitz explains to RG how this system works.
Photo: Denis Belozerov
- Any complex machine, be it a turbine or a transformer, breaks down periodically, and a control system activates when something is already broken. What are we doing? We discover this breakdown long before it happens, at the moment when it is only “planning to happen”, and we warn specialists about it. In practice, it turns out that the forecast horizon at the generation facilities is about three months.
When the system was first introduced in 2015, it was part of a service contract for the maintenance of generating equipment; gas turbines in particular. A few years before that, we began to think which monitoring system could be used for turbines. We looked at what Siemens, General Electric, and other market players were using. Our experts realized that the existing systems no longer meet modern requirements and decided to develop our own system.
In 2017, we removed it from the service contract, and started selling it as a digital service. There is a format called SaaS – “software-as-service”. We rent out a cloud solution and provide expert and dispatch support. The service can be applied on most manufacturers’ main types of equipment. Anyone can be a service provider. We now take any data exchange protocol available on the market – these are 25 manufacturers, but the client objectively sees everything that the service provider does. They see the state of the turbine before maintenance, the result, and what happened inside.
Are there other systems like PRANA?
- In terms of systems actually created and operated in Russia, ours is the only one. The system has been in development since 2011. It has been commercially available since 2015, when we connected the first four gas turbines. These were power plants in Kirov, Vladimir, Perm and Izhevsk. Now we are working with Mosenergo, Tatenergo, T Plus, Pavlodarenergo in Kazakhstan, with Gazprom Neft and many others. I hope that we will soon enter the Mongolian market. Globally, of course, there are competitors, such as General Electric and Schneider Electric.
What is unique about the system? Have you patented anything?
- Everything is unique. We have already received 10 patents and there are 20 more patent applications. The system is constantly evolving. In particular, we use machine learning, there is a lot of talk about it, but when you start to use it, things happen that you didn’t even anticipate in the beginning. For example, when a turbine is operated, its internal temperature is 1,200 degrees centigrade and its rotation speed is 3,000 rpm.
It is fitted with a lot of sensors, data is taken from the casing about vibration levels, temperatures and so on. But what should we do about the rotor, which lives its own life inside? Also, a “heat spot” forms at the turbine outlet which is not static, it is moving slightly all the time. So the machine, through learning, revealed the behaviour patterns of this heat spot, and our PRANA Predictive Analytics and Remote Monitoring System revealed the moment where the destruction of the blades began according to its behaviour. Nobody had ever done that before us.
We take a lot from our interactions with clients. At the beginning of our journey, there were several cases where an employee saw our forecast about a problem, looked at the screen, where everything was in order, and simply ignored the problem. And then alarm was triggered. The managers of the guy who ignored the warning said: we need a control mechanism so that our specialists can respond correctly to your reports. And so we implemented a digital online log inside the System.
The Technical Director and his or her experts have a tablet equipped with the System. PRANA makes all work absolutely transparent. The system shows how the station personnel reacts to changes in the state of equipment and what should be tracked. It is also a decision support system: when a repair is needed, how much money is needed for the next year, places where breakdowns could occur and what spare parts need to be ordered in advance.
Warnings of various scales are ongoing. There was an episode when we stopped launch operations. The client was sure that everything was fine. But in fact there was a malfunction in a gas turbine bearing and it would break down if launched, meaning a loss of 20 million euros.
One of your clients showed a 13-fold reduction in accidents and 16-fold reduction in equipment outages. How can we evaluate the effect of implementing the PRANA System?
- The effect is always determined by two budget items. These are “contingencies” and “emergency repairs”. From operational experience, it turns out that we simply “multiply by zero”, minimize these items in terms of key equipment, such as turbines, generators, transformers, compressors and boilers. We transfer these repairs from the state of “emergency” and “unscheduled” to the state of planned and predicted. We also help with decision-making – when it is necessary to make repairs and re-launch the equipment so as not to suffer damage from huge fines due to an unscheduled stop. Any emergency stop leads to an electricity generation break and any urgent repair is much more expensive. Furthermore, if these units are built according to the PDC program, then there are also significant fines.
Your business is at the crossroads of heavy engineering, energy and IT. Why has energy become the place of application for your mathematical business?
- The business itself is about mathematics. Everything that we have implemented and launched is based on our significant knowledge in power engineering and energy. (Mikhail Lifshitz – Chairman of the Board of Directors of The Ural Turbine Works, which produces power steam turbines).
The system is quite complex and expensive. As such, it is advisable to use it where the equipment is expensive, and dangerous. That is at infrastructure facilities, where generation takes place.
What is new about the project with Rosseti in Izhevsk?
- We have been connecting transformers to power plants for a long time. Before that, we regarded transformers as a piece of equipment. The novelty of the project is that we take not a single transformer, but a mathematical model of a whole section of networks. The first connected facility will be the Kalashnikov electrical substation, which supplies power to the city center. These are networks of the whole urban area, a very powerful branched network having more than a hundred pieces of equipment. The system builds a model of facilities’ operation and can catch the initial moment of failures in the city’s energy supply.
How does the PRANA Predictive Analytics System work in insurable events?
- The first money that we earned with the system was from insurance companies. They come to us after insurable events, bring digital archives and run them through the system. And when there is an analysis of “who is to blame”, it is very difficult to argue with a machine. Now we have agreements with three insurance companies which offer a discount on equipment insurance when the system is installed. The insurers trust us.
The PRANA System is installed on the main power generation equipment with a total capacity of more than 3.5 GW and a cost of five billion dollars. In Udmurtia, it will connect to the equipment of the Udmurt Branch of the Rosseti Center and Volga Region – 110 kV Kalashnikov Substation, 110 kV Pazely Substation and 35 kV Airport Substation. Moreover, there is a plan to connect Central Power Grids, Southern Power Grids and Glazov Power Grids in Republic of Udmurtia by 2023.