1. LOSS ESTIMATION IN LV CIRCUITS USING INTELLIGENT TECHNIQUES - THE RGE EXPERIENCE
Carlos César Barioni de Oliveira Enerq/USP
André Méffe Enerq/ USP
Hernán Prieto Schmidt Enerq/ USP
Mauro Augusto da Rosa RGE

2. Objectives
Determining types of LV circuits representing all of the utility´s LV circuits by means of:
Attributes: (1) distribution transformer rated power (kVA), (2) transformer type, (3) rated voltage (kV) and (4) type of circuit location (urban or rural).
Typical daily load curves for each consumer class and consumption range
Developing techniques and tools to classify a specific network for one of the pre-set types of LV circuits.
Estimating losses in LV circuits whose topological data is unknown.

3. Classification Process
Out of networks, 187 representative networks were selected.
Besides the data available from Data Bases, the network topological information was obtained.
LV circuits:
Type of transformer
Type of area
Rated power
Rated voltage
187 networks:
Loss coefficient
Network length

4. Classification techniques and tools
Classification techniques (Hierarchical and Self Organizing Map-SOM), allowed grouping the sampled networks into categories having similar attributes.
Upon comparison of both techniques, the Hierarchical Classification technique proved to be more appropriate for the objective of this study

5. LV Network Modelling
The LV network analysis is based on the losses coefficient (LC) and average length (L) of the corresponding representative network.
The losses daily curve is then obtained from the transformer daily load curve:

6. Load model validation

7. Computational Tool

8. Computational Tool

9. Conclusions
New methodology for estimating demand and energy losses in LV circuits - reduced set of circuits and the extrapolation of some parameters (loss coefficient and circuit length) to the whole LV circuit population
It overcomes the lack of information normally associated with such circuits.