The increasing digitalization of assets is resulting in an avalanche of data. The knowledge on these assets could be leveraged for making better informed decisions. When the present grid moves more towards smart grid, a more pervasiveness of information and communication technology could be expected. The increasing penetration of intermittent renewable energy sources and incorporation of utility scale energy storage systems in near future and technologies like electric vehicles adds in more uncertainties in the energy grid. The power transformers in the grid assume a very significant role in transmission and distribution of power. A failure of which could result in catastrophic consequences in terms of stability of the grid and its economic implications. The modern grid is not a mere interconnection of energy devices, it is also a network of measurement devices and communication such as Supervisory Control And Data Acquisition (SCADA).
According to a survey published by the International Council on Large Electric Systems (CIGRE) one of the major failures of the transformer was related with the insulation breakdowns. The transformer condition is monitored using oil tests, temperature measurements and onsite inspections. Various tests are in place for testing power transformer while it is in operation –like Dissolved Gas Analysis (DGA), partial discharge detector, Furfuraldehyde Analysis (FFA), temperature measurements, thermal imaging etc., many of which are available for online monitoring of transformers as discrete units as well as integrated solutions. With the expansion of the Industrial Internet of Things (IIoT), addition of more such online monitoring systems becomes obvious.
IoT and big data
Internet of Things (IoT) is an interconnection of physical ‘things’ that can interact with its physical environment with sensors/actuators along with computing and communication capabilities to exchange data with internet. The IoT devices have become ubiquitous in consumer electronics with the rise of smartphones, wearable health monitors, home automation controllers etc. IoT in the context of industrial environment is termed as Industrial Internet of Things (IIoT). The increased acceptance of smart grid technologies accelerates the deployment of IIoT in power system, especially in the advanced metering infrastructure, digital protection systems, phasor measurement units, intelligent electronic devices and asset monitoring systems. These produce enormous amount of data in the power system, which could be overwhelming for traditional decision making approaches. When considering an example of 50 Hz three phase voltage signal sampled at 128 samples per cycle, generates a dataset of 19200 samples per second. The collection of data, which is characterised by massive volume, high velocity and heterogeneous variety is termed as big data. The data from the big data analytics are applied to make sense of the collected data.
Digital Twins of transformers
The concept of Digital Twin was introduced in a presentation related to product life management by Dr. Michael Grieves of the University of Michigan in 2002. The name Digital Twin was put forth by John Vickers of NASA in 2010. The Digital Twin encompasses the physical system in the real world, its digital model and the communication linking the both. The digital thread concept forms the basic subunit of a Digital Twin, which is essentially the information that could trace back to the real world system. The expansion in the Digital Twin is fuelled by the Internet of Things (IoT) and reducing costs of computational and storage resources. The data from various sensors are aggregated by the monitoring system. The raw data collected from the sensors along with additional features when used with appropriate machine learning algorithms could be used to create a more realistic model of a real physical asset. A Digital Twin goes a step further to keep a digital footprint of the transformer throughout its lifetime. The data collected over time is analysed for trends of relevant features. This enables Digital Twins to be used for condition monitoring, predictive maintenance and remaining useful life estimation. Digital Twin creates a digital trace, so in case of a mishap it can be investigated with the digital counterpart. The Digital Twin concept has been researched in various industry segments such as manufacturing, healthcare, driverless autonomous vehicles and machine-to-machine interactions.
Digital Twins of transformer are virtual replicas of physical transformers, which engineers and consultants can use to run simulations and test scenarios before the actual devices are built, deployed or operated. Digital Twin technology has moved beyond manufacturing segment and into the merging worlds of the Internet of Things, Artificial Intelligence and Data Analytics. Digital Twin collects data from sensors and analyses to initiate the response through actuators.
The Digital Twin may be deployed either on a cloud platform or on edge computing device. The Digital Twin deployed on the cloud platform can take advantage of higher computing resources and big data analytics. A Digital Twin on edge may be used for quick dynamic decisions and controls in order to avoid any latency in communication channels and information processing. Also, edge based solution is called for when considering privacy and information security. So, one of the strategies is to have an edge model for quick decisions and computationally demanding analytics on cloud.
Asset management and Predictive maintenance using Digital Twin
The Digital Twin concept provides a great choice for modelling complex systems such as a power transformer. A Digital Twin of transformer is a virtual image of the real world transformer, which captures its historical, static and dynamic characteristics. Any changes in the actual transformer parameters get reflected in its digital counterpart. The time based variation of various parameters of the transformer need to be captured as transformer health is influenced by ageing and maintenance operation. The model is initially loaded with a set of scenarios and while in operation its measurements from the transformer are used as feedback to keep the model updated. The Digital Twin of the transformer once made can be used for simulating the power transformer in all different conditions.
One of the popular yardsticks for transformer health is the dissolved gas content in the transformer oil. Traditionally dissolved gas analysis is performed in an external lab by taking an oil sample, now an online DGA kit is being set up along with the transformer. The condition evaluation criterion based on the DGA results are mentioned in IEC 60599 and IEEE C57.104.
Several studies could be observed in literature, which tries to estimate various transformer conditions based on DGA results. This includes traditional ratio methods, Duval’s Triangle method and intelligent methods like fuzzy logic, Artificial Neutral Networks (ANN), Support Vector Machines (SVM) etc. The DGA in itself does not give a complete picture of the transformer condition. DGA results are just one among several measurements associated with a power transformer. The data generated from multiple sensors need to be collectively analysed along with operational history to comprehend the condition of the transformer.
The monitoring device generates data continuously about the transformer under observation. This is apart from maintenance data, lab test results and equipment details. The Digital Twin keeps-on building digital thread of an asset right from its deployment, operation and finally up to its termination. The magnitude of the data makes it difficult for humans to understand, but an appropriate algorithm could comprehend. In a substation with multiple transformers even if all transformers were of same specifications, each becomes a unique entity to be cared for especially when it ages. So, each transformer in the substation would be mapped over to a unique Digital Twin. The operator can use the Digital Twin and its visualization real time monitoring, situational awareness, stability analysis, planning, fault identification, schedule upcoming maintenance and take better decisions. The Digital Twin of power transformer could also be used for controlling the actuators as power transformers are now being equipped with actuators like valves, cooling systems, dehumidifiers, tap changers etc. The utility company may use its collection of Digital Twins of its fleet of power transformers for contingency analysis, maintenance scheduling and budget allocations.
Conclusion
Digital Twin technology is promising for the future energy grid. The earlier the digitization rollout, it gives advantage over the amount of real time data harnessed from the asset. The Digital Twin is expected to bring in an efficient approach in design, operation and maintenance of power transformers. The confidentiality and integrity of data is important for implementing Digital Twin for a crucial piece of equipment such as a transformer. When we are leveraging data driven solutions and internet connectivity for operations of physical systems, it comes with its inherent vulnerabilities of cyber-attacks. The information and communication layers are on top of the power system, opening doors to cyber threats. In history, there have been several incidents of cyber-attack on energy grids. The cyber security of such cyber physical systems is still an active area of research. Regulations and policies are to be laid out taking into account data security and data management. For completely utilizing the Digital Twin models need to be interoperable across various platforms and the data exchanges should be maintained as per standards.
John A. R. is a Project Associate with the Department of Electrical Engineering, Indian Institute of Technology Madras, Chennai, India. His area of research interest includes Power Electronics based Power System and Application of Digitalization in Power Grids.
K. S. Swarup (S ’87, M ’92, and SM 2002) is with the Department of Electrical Engineering, Indian Institute of Technology Madras, Chennai, India. His areas of research are Computer Modeling, Simulation and Applications of Power Systems, Power System Automation, SCADA and Numerical Protection.
