Using location intelligence to flag meter tampering by paying customers

Using location intelligence to flag meter tampering by paying customers

Claire van Zwieten

The City of Ekurhuleni (COE) has identified an efficient and effective way to identify paying customers who are tampering with their meters and recording purchases to avoid detection through no purchase reports. The method enables revenue personnel to identify cases where there is a high probability of tampering taking place so that corrective action can be taken.

Two possible approaches are discussed below.

Approach One

The first approach involves auditing stands with low consumption, by tariff, from lowest average monthly consumption towards highest average monthly consumption. (Higher income consumers are on Tariff B and a better return on effort is expected from curbing high-end tampering.) This approach can result in stands needing to be audited not being grouped together through the selection process, which results in resources not being optimally utilised.

Approach Two

The second approach involves assigning stands a priority index based on average monthly consumption divided by property value multiplied by the number of units on the property. This process results in higher value properties with low average monthly consumption receiving a smaller index value. Suburbs are then assigned a priority index based on the sum of stand priority indexes with a value of 0.0005* or lower within a suburb. (*This value can be modified to fine-tune results.)

After this, the suburbs are graded from those with the highest index and providing the greatest return with least effort (in terms of number of properties) to those with the lowest index. A GIS desktop analysis of the resulting dataset can then be used to flag which stands within the dataset should be audited.

The flagged stands are then audited in the field to determine if tampering is taking place. Sometimes false flagging occurs (e.g. a meter linked to a stand on which a high value property exists that is being used to power a gate and guardhouse) but often enough actual instances of tampering are detected.

(Words 327)


 Questions for Stephen Delport for Zoom Interview 

  • I understand that the City of Ekurhuleni has been investigating effective and efficient solutions for identifying meter tampering by paying customers. Why is this important for COE?
  • The rising cost of electricity is resulting in an ever-greater number of customers tampering. This increases the non-technical losses of CoE with a resultant reduction in revenue. The tampering needs to be curbed to prevent loss of revenue and to get the message across to other would-be tamperers that there are consequences for tampering.
  • Your department has identified two possible approaches to tackle this problem. Please could you describe how Approach 1 works? (Slides 4 to 10)
  • The average monthly consumption is calculated for a property over a year to allow for seasonal differences in consumption. This exercise is done per tariff (A (IBT) or B (fixed charge + flat rate) and the properties are sorted from lowest to highest average.
  • Is there a downside to Approach 1? What is its success rate? (Slide 10)
  • The downside is that the order that properties are sorted in means that the properties are not grouped logically for audit purposes, particularly if the data is for the whole of CoE. This can be mitigated to a certain extent by dealing with one suburb at a time but the methodology provides no method of prioritising which suburb will provide the greatest return for effort expended. The success rate is > 50%, which is the same as for Approach 2, but it will not be as efficient in terms of resource use as there is no way to prioritise suburbs.
  • Approach 2 appears to be very effective and efficient. Could you describe how Approach 2 works? (Slide 11)
  • Approach 2 calculates a priority index for a stand from the average monthly consumption divided by the property value. The lower the index, the greater the theoretical return on effort expended. The sum of the indexes per suburb is then used as an indicator of the suburb priority.
  • How do you generate a priority index for stands? (Slide 12)
  • The metering, property and customer data resides in a PostgreSQL database that is spatially enabled through PostGIS. The calculation is performed using a join from the spatially enabled data to the monthly average consumption with the venus code (stand number) used as the key.
  • Explain the process of generating a priority index for suburbs? (Slide 13)
  • A spatial query is used to sum all the stand indices within the boundaries of each suburb and divide the sum by the number of stands within the suburb.
  • How is the filter set up to identify the low consuming data layer? (Slide 14)
  • The value to be used as a filter is selected through examining the range of the calculated stand indices and selecting a suitable value. The filtering can be performed using a SQL query or Excel can be used to connect to the database data and the filter can be performed in Excel.
  • Describe the process of sorting the suburbs from highest to lowest priority index? (Slides 15 to 16)
  • The sorting can be performed using a SQL query or Excel can be used to connect to the database data and the sorting can be performed in Excel.
  • How are the flagged stands assessed to determine whether tampering is indeed taking place? (Slides 21 to 24)
  • This is not a prerequisite step but can be carried out to potentially reduce the number of stands requiring a field audit. Using GIS with recent aerial photographs allows identification of properties with valid reasons for lower than expected consumption, e.g. solar panels installed on the roof.
  • What success rate does Approach 2 have in identifying instances of meter tampering?
  • As with Approach 1 the success rate is greater than 50% but resources are used more efficiently.
  • Would you recommend Approach 2 to other metros and municipalities?
  • Yes, it is an approach that can be followed to utilise resources efficiently and effectively to target tampering. However, the approach CoE is taking is constantly evolving and being improved with the intention of increasing the success rate.


If you would like to watch a video recording of the interview, click here: