CALIBRAÇÃO DE DADOS PARA ESTUDOS DE CONFIABILIDADE EM REDES DE DISTRIBUIÇÃO: MODELAGEM DA CONDIÇÃO DOS EQUIPAMENTOS E DOS ÍNDICES DE CONTINUIDADE NODAIS

Abstract

After the restructuring of the electric sector, the distribution utilities must maximize the reliability to avoid violation in the reliability targets at the minimal cost. This agreement between cost and reliability can be satisfied with the application of Predictive Reliability Analysis (PRA) in the planning of distribution networks. The PRA estimates the future performance of distribution networks, with regarding to energy supply interruptions, based on the failure data of the components and network topology. The PRA can delivery estimates for the following statistical reliability indices used in the distribution utilities: System Average Interruption Frequency Index (SAIFI), System Average Interruption Duration Indices (SAIDI), Connection Point Interruption Frequency Index (CPIFI), and Connection Point Interruption Duration Index (CPIDI). However, the PRA is rarely used by engineers during the planning of the distribution utilities. This fact is due to the existence of discrepancies between the indices estimated by the PRA and those measured by distribution utilities. These discrepancies are due to the lack of historical data to estimate the reliability parameters of the components: failure rates and repair times. In spite of the distribution utilities do not have a large amount of historical data associated with failures in their equipment, these utilities store historical data on system reliability indices (SAIDI, SAIFI, CPIFI and CPIDI). This information can be used to adjust the failure data of the components (failure rates and repair times) such that the reliability indices evaluated by the ACP models have nearly the same values as those measured by distribution utilities. This adjustment process of the reliability data in ACP models is named Data Calibration. Usually, the reliability data calibration is carried out through optimization techniques. However, the most of the existing methodologies ignores the nodal reliability indices (CPIFI and CPIDI) in the calibration of failure rates and repair times. Only the CPIFI index has been considered in the data calibration. Furthermore, it is not possible to assure that the SAIFI has the same value as its measured value when the calibration considers the CPIFI index. Nevertheless, the Brazilian Electricity Regulatory Agency (ANEEL) has established penalties for violations in the indices CPIFI and CPIDI. Due to this, the PRA models must accurately estimate the nodal reliability indices CPIFI and CPIDI. The main objective of this dissertation is to develop a calibration methodology of reliability data oriented to nodal reliability indices CPIFI and CPIDI. The proposed methodology uses nonlinear and quadratic programming models to calibrate the failure rates and repair times, respectively, in a decoupled structure. This decoupled structure allows the calibration of failure rates and repair times be carried out separately. Additionally, the utilization of equality constraints in the calibration models assures that the evaluated values of SAIFI and SAIDI indices are identical to their measured values. Furthermore, the proposed calibration model for the failure rates considers the equipment condition information obtained from inspection activities. The calibration models proposed in this dissertation were tested in a feeder of the power distribution utility of Maranhão (CEMAR). The tests results demonstrate that the proposed calibration models can significantly reduce the errors between the measured and evaluated values of the CPIFI and CPIDI indices