The vast majority of Swedish utility network customers have for nearly a decade been supported by Advanced Metering Management (AMM) technology, including smart meters. Vattenfall´s modern smart meters enable a new level of monitoring for the LV network and improved MV network supervision. Thereby, improving power quality, fault detection and outage management functionalities.
One of the typical faults in LV networks is when the neutral conductor is broken or loose at either the network or the load side of the meter. The situation of lost continuity of the neutral conductor may damage the connected load or create hazardous touch voltage at equipment body. Such condition is not detected by the overcurrent protection devices, fuses or RCD. Moreover, currently used smart meters do not have the capability of detecting this disturbance automatically.
Vattenfall´s smart meters are already providing streams of data that are recorded. Some parameters are collected routinely by smart meters (energy, and disturbance events). Since there is a big potential to supervise the LV network with assistance of the end-customer smart meters, Vattenfall wants to take further advantage of such data. The value is to bring in event information from the smart meters in order to contribute to a better and more efficient monitoring of the LV and MV network.
Goal / Purpose:
The goal of this project is to analyse the behaviour of the LV grid under broken neutral conditions and propose an effective method (algorithm) to identify loose neutral situation based on end-customer meter readings (disturbance events).
A suitable algorithm would be needed at intermediary points so that, if broken neutral problems have arisen, clearly information can be sent to the operator about the type of fault and where the fault is located; in consequence, appropriate measures can be taken to address them, to avoid dangerous voltage levels, personal injury or fire hazard.
Several location possibilities should be considered if the neutral conductor is broken: close to the MV/LV transformer; close to main distributed-generators or large loads; at the far ends of lines or at the customer side (behind the meter). Other cases should be taken into account such as the grid conditions (strong, average and weak grid), the number of customers that are affected (several or just one) and different load conditions.
A general method to do the classification problem would be needed. This method should contribute to find explanations to why disturbances look the way they do, find a good state definition and set suitable threshold limits. Supervised and unsupervised machine learning methods such as multi-layer perceptron learning (MLP), Support vector machine (SVM), Convolutional neural network (CNN) or k-means clustering will be used to find the neural network. Such advanced analytics could allow Vattenfall to better understand the grid situation and also to support its decisions.
The study will be closely linked to Vattenfall DSO´s specific process, giving an analysis of our technical solutions in the field related with broken neutral detection. In consequence, current field test process during real faults situation will be evaluated to identify possible deficiencies that could limit the algorithm effectiveness, what needs to be changed and how. As a result, the project will support Vattenfall in improving of the whole chain process related to broken neutral disturbance, from early monitoring to field detection.
Expected outcomes & results for reporting and presentation are:
• Find a properly configuration of the neural network that could potentially, be used for the quickly finding of some obvious patterns for different fault states with large amounts of data, and further guide us to develop the algorithm.
• Propose and compare algorithms to detect broken neutral and identify where in the grid the problem is allocated, by using data from the smart meters.
• A simulation of such systems should be delivered, including the network models that should be used based on the above cases.
• A study on how this method can contribute to improve reliability indexes , give indications for maintenance and reduce utility’s costs.
The project is quite practical in application: with today’s meter collection and the algorithm it could be possible to automatically start to map the extent of a broken neutral impact. This method would allow Vattenfall to fast identify this kind of event and thereby, reduce corrective time and penalty for the utility.
Qualification, Application and Timeline:
It is preferable that the student has studied Master’s program in Electrical Engineering or Electrical Power Technology. However, as the proposed method will be implemented as a computational tool (using an agreed environment, e.g. Matlab, Python, etc) and applied to a few example networks, is likely to benefit from a person with mathematical interest and programming skills.
Please send your application to email@example.com . CV, cover letter, and grades are to be submitted in the application. Deadline for application is Nov 30th, 2018. The project´s scope is 36hp and it is planned to start in January 2019 or according to the agreement made with the student.
Supervisor, Contact person, Location:
The contact person and supervisor at Vattenfall R&D is Edel Wallin, Tel: 072 2198565, firstname.lastname@example.org This master thesis is done with Vattenfall R&D in cooperation with Vattenfall Eldistribution. The work place will be at Vattenfall R&D in Solna, Arenastaden.
Information for clients:
"We are convinced that diversity helps to build a more profitable and attractive business and strive to be good role models when it comes to diversity. Vattenfall is working actively to ensure that all employees should have the same rights and opportunities regardless of age, ethnic or cultural background, sex, religion / belief, sexual orientation or disability".