Automatic Identification of Poorly Performing Substations and Meter Devices. The Future of District Heating Analysis

Abstract: The district heating sector in Sweden is today facing several challenges due to competition from other heating alternatives as well as decreasing heat demand in buildings. To increase its competitiveness, the district heating companies have to find ways to keep their production costs at a more or less constant level. One way of doing this is to increase the efficiency of the district heating systems. To be able to do this, the district heating companies need to identify substations which have a negative effect on the overall efficiency of the system. An example of these substations are substations with poor cooling performance which means that they do not extract as much heat from the district heating system as they are supposed to do. To compensate for this, more hot water needs to pass through the poorly performing substations. As a result, the district heating system will be inefficient and this leads to increased production costs. To be able to find the substations with poor cooling performance, the meter reading data of the substations has to be investigated and analysed. Today, most district heating companies perform these analyses manually which is both timeconsuming and ineffecient. Many companies are now interested in developing automatic methods to identify poorly performing substations. The purpose of this study is to develop a substation analysis program which automatically can identify poorly performing substations out of a total number of 3 000 district heating substations. The large amount of substations generate a large data set, and in order to be able to perform correct analyses, the data which is analysed has to be of good quality and not contain any abnormalities. Because of this, this study also aims to develop an investigation of data program which can identify, and handle, potential abnormalities. To be able to identify abnormalities in the meter reading data, hourly meter readings are used since these contains a large amount of information about the meter devices’ performance. A number of common abnormalities are identified and handled according to their impact on the data set, before converting the hourly meter readings into daily consumption values. These values are then used in the substation analysis program in order to identify the poorly performing substations. The first step to identify substations with poor cooling performance is to create a reference case based on the daily consumption values for substations with good cooling performance. The daily consumption values for each substation are then compared to the reference case. If the values differ with more than a prescribed tolerance, the substation is declared as poorly performing. This procedure is performed for three different signatures based on energy, cooling and return temperature values. The output from both programs are lists containing ID numbers for poorly performing substations and meter devices respectively. The lists containing poorly performing substations compiles the result from the three analysis signatures and rank them according to their overflow. The overflow is a quantity which describes how much excess water passes through the substation in question due to the poor cooling performance. The lists containing poorly performing meter devices are presented for each identified abnormality and ranked according to the total number of abnormality occasions. The output lists show that the programs can identify poorly performing substations and meter devices. Due to the large amount of data, it has not been possible to manually validate the entire result. Instead, some samples have been investigated which shows that in most cases, the abnormalities identified in the data investigation program are correctly identified. However, this investigation also shows that the programs can not cover all different types of abnormalities and difficulties of the data set. This study identifies poorly performing equipment of one district heating system. The results from the programs should be considered as an indication of what equipment should be investigated further in order to improve the system performance. It does not investigate the overall efficiency of the impact of the equipment on the system, and neither analyse the economical benefits which may arise from an overall improvement of the system performance.