Data Driven Model for Medium Voltage Cable Fault
The currently ongoing energy transition requires a significant increase in the resilience of electrical distribution grid. Moreover the failure rate upon the grid is growing in the last years, due to the increase in extreme weather conditions. Consequently, it is crucial for the electrical distribution system operator (DSO) to better understand the grid failure phenomena in order to optimally drive preventive interventions. In this project, we seek to carry out a model for the fault phenomenon which identifies its dependence on the grid’s constitutive parameters, operational variables, and external causes (weather, grid load, etc), exploiting the information collected in the urban area of Rome.
Category: Urban Infrastructure
Project Description & Overview
The Italian government authority promotes the continuity of the service through a system of rewards and penalties that it imposes on DSOs in order to direct them towards a continuous increase in the resilience of the electricity grids. In this scenario the DSOs are faced with unpredictability of the alternative-sources energy-flows, combined with the growing urban populations especially into an urban contest like the city of Rome, where many portions of electric grid are out dated and their replacement is difficult. The above combined with the climate extreme events increasing determines a significant fault number and impact growing. Consequently, it is critical for the DSO to better understand and model the failure phenomenon in order to best guide preventive interventions on the power grid.
In this project, we seek to model the fault phenomenon on the DSO electric- medium voltage cables and understanding its relations with constitutive elements, operating variables and external stresses that works on the cables of the grid.
We will integrate literature studies, information gathered in a structured way from the human experience of expert grid-operators, and obviously data available on DSO databases in order to build a failure model able to best classify the different vulnerability of cables of the grid. The final data driven model will take as input the constitutive characteristics of the cables (like: material, length, section, voltage level, type of installation); external causes registered when the fault occurs (like: weather conditions, electric loads). Finally the model must predict as output the expected fault rate of each grid-cable, under assigned external variables determining multiple scenarios.
Areti already owns huge amount of data already integrated on its analytics systems. The main data sourses available are:
• GIS System: which store constitutive and georeferenced information of the grid asset
• SCADA: which store topological and electric load information of the grid
• WHEATER PROVIDER: which supplies information about measured weather parameter on the upon the metropolitan area
• MDM system: meter data management system which elaborate all the electric consumption coming from the grid users
• FAULT DATABASE: historical information on previous fault
Information coming from structured survey of grid operation expert
-Data preparation and processing (preferably SAS & Python)
-Data analysis and visualization (using SAS Python)
-Programming (preferably Python, SAS)
– Machine learning techniques and data driven models knowledge
– Unstructured data processing
Learning Outcomes & Deliverables
1. Students will learn the data collection and pre-processing methods and their importance.
2. Students will be trained on the scientific method approach and hypothesis testing and discover data modeling for analysis purposes.
3. Students will learn to apply traditional tools of temporal analysis and information theory.