Modelling and Simulation

Enhancements in Power Transformer Arc-Resistant Specifications

Jean-Bernard Dastous

Abstract In 2007, Hydro-Quebec introduced arc resistance requirements in a new version of its power transformers’ technical specification. These requirements, based on earlier research investigations, featured arc energy levels to be contained, as well as an equation to establish a corresponding design pressure that the transformer had to withstand without rupture. In the process of its implementation, various areas in need of improvement were identified, prompting further research and analysis. These efforts culminated in the development of a validated and standardized methodology for evaluating arc-resistanttank design. This methodology is now prescribed for manufacturers and may be used by utilities to assess the suitability of supplied transformer designs. The key elements of this methodology, based on the use of nonlinear static finite-element analysis, were introduced in an updated version of our technical specification in 2020. Also, several improvements were introduced regarding the pressure design equation and the arc-resistant requirements of external components connected to the main transformer tank. This paper presents the key elements of our new arc-resistance specification, as well as some of the technical background upon which they rely, to lead to an understanding of their scope, expected variations, and limitations

Jean-Bernard Dastous

Short resume Jean-Bernard Dastous received the Bachelor of Science degree in mechanical engineering from the Université de Sherbrooke, Canada, in 1989, and the Master of Engineering degree from McGill University, Canada, in 1993. In 1989, he joined IREQ (Hydro-Québec research institute), Varennes, Canada, as a Research Scientist and has worked in the field of structural analysis of substation equipment and structures since then. He is the Chair of the IEEE working groups developing Standard 605 (bus design in air insulated substations) and Standard 1527 (seismic design of buswork between substation equipment). He is also a member of the IEEE working groups developing Standard 693 (seismic design of substation equipment) and Standard C57.156 (transformer tank rupture mitigation). He is presently convenor of a new CIGRE task force on Power transformer tank specification for passive protection against internal arcing. His areas of research interest include seismic design, design of bus structures, and arc containment in oil and gas insulated equipment.

Materials, Components and New Technologies

“Transformer digital twin – concept and future perspectives”

Patrick Picher (presenting), Sicheng Zhao, Zhongdong Wang, Sruti Chakraborty, Stephan Voss, Mohamed Ryadi, Tony McGrail, Nima Sadr Momtazi, Alexander Alber

Abstract The transformation of the power system via digitalisation brings new opportunities for innovation. For example, the digital twin concept has been studied extensively in the scientific literature of recent years, often for the virtual representation of manufacturing processes, but also for modelling of critical assets. Because of their strategic importance in electrical networks, transformers are already the focus of international efforts in power asset digitalisation and, therefore one of the top priorities for asset digital twin developments. The new CIGRE WG A2/D2.65, initiated in 2022, studies the concept and future perspectives of transformer digital twin. This paper reviews the state of the art of the digital twin concept, and presents some potential benefits and use cases.

Patrick Picher

Short resume Patrick Picher has been working as a researcher and project manager at the Hydro-Québec’s Research Institute (IREQ) since 1999.  His research interests are mainly focused on diagnostics, monitoring and modelling of power transformers. Since 2003, he was involved in international CIGRE working groups related to transformer Frequency Response Analysis (FRA), thermal modelling, intelligent condition monitoring, condition assessment indices, the influence of geomagnetically induced current and digital twin.  He was Secretary of CIGRE Study Committee A2 (transformers) from 2010 to 2016 and the Canadian representative on this committee from 2016 to 2022. He graduated from Sherbrooke University, Canada, in 1993 with a B.Eng. in Electrical Engineering and received his Ph.D. degree in Electrical Engineering from École Polytechnique de Montréal, Canada, in 1997.  Mr. Picher is a registered professional engineer and a member of IEEE (Senior Member), CIGRE (Distinguished member) and IEC TC 14 (Canadian mirror committee).

Transformer Life Management

“Application of Probabilistic Bayesian Networks on Transformer Condition Assessment”

Luiz Cheim (presenting), Alan Sbravati, Kumar Mani

Abstract Traditional transformer assessment techniques are developed based on the available data extracted from the transformers. Albeit aiming to estimate the risk of failure of a transformer, most conventional approaches were developed essentially centered on combining results from online monitoring system and offline test results. Thus, they focus more on ranking the units based on the data. The common strategies include, among others, weighted averages, criticality indexes and traffic lights, focusing more on maintenance prioritization, interventions, and budgets allocation. The method currently applied by the company represented by some of the authors already incorporates the estimation of the probability of failure, based on proprietary knowledge and experience. The current method expands the approach, allowing users knowledge and experience, as well as user-specific statistics, to be incorporated in the analytical process, adding a probabilistic layer to the typical tree of failure modes. Rather than the test results itself, the input data to the model is the “belief” that the data indicates the component or system will fail or not. For instance, abnormal results in a DGA result may impact the risk of failure in different components of the event tree, which will further impact the associated risks of the transformer to fail. Based on the concept of conditional probabilities in Bayesian statistics, this method allows inferring the expected impact / criticality of each type of issue (evidence propagation) on the continuous operation of the transformer. The likelihood of each failure mode can be estimated either based on the statistics of international transformer reliability surveys or on the experience of each asset management group. The Bayesian network analysis allows the bidirectional assessment of the system, both for checking the impact of each root cause on the transformer operation (inference) and to investigate the likelihood of a given cause, should a situation be identified (diagnostics). 

Luiz Cheim

Short resume Dr. Luiz Cheim has been with Hitachi Energy as a Sr. Principal R&D Engineer for a number of years, having over 30 years’ experience in the power transformers industry. His major activities as part of a global R&D team are in the development of transformers condition assessment and performance models and algorithms, as well as the development of new sensors and state of the art monitoring technologies. Dr. Cheim is the proponent of the new Hitachi Energy Transformer Inspection Robot (TXploreTM). In August 2018 Luiz was granted the Best Paper Award by the CIGRE organization in Paris, Study Committee A2/PS2 on the use of AI/Machine Learning techniques in support of transformer diagnostics. Luiz is in the editorial board of the new CIGRE Green Book on Transformer Life Management, a Task Force leader of CIGRE WG D2.52 AI Application and Technology in the Power Industry, responsible for Chapter 5 – Applicability and Maturity of AI Technologies. Luiz, together with a few members of the Cigre WG, gave a tutorial on the subject over the last Cigre Paris Session 2022. Dr. Cheim has filed over 20 patents in the last 10 years alone with Hitachi-ABB, including the most recently granted patent on AI Superminds. Luiz is also the Guest Editor of the Transformer Magazine Special Edition on AI and Machine Learning, November 2022 issue.