What does correlation mean and how do you find a water leak by using correlation?

In water leak detection, correlation describes the measurement of time difference of sound waves via microphones. Leaking water or compressed air from a defective pipe generates a noise. The sound waves generated in the process propagate in both directions in the pipeline at a certain velocity.

The basics of correlation - parameters for a measurement - formula

Formula Correlation
  • D = Total distance, pipe length between the measuring points, can be re-measured with a measuring wheel
  • L = Distance to Leakage
  • V = Sound velocity in m/s, value usually based on empirical values
  • The time difference T is measured in milliseconds (ms).

Two microphones (sensors) can be connected to the leaking pipe, e.g. to a hydrant or slider. They detect the noise generated by the outflowing water or compressed air and transmit it to the correlator.

Video with sample data

The optimal conditions for the calculation of the correlator

The suspected leak must be between the two microphones and the length of the water pipe between the microphones must be known. The correlator measures the difference between two points in time, the arrival of sound at sensor 1 and the arrival of sound at sensor 2. This means that the time the sound takes to arrive at the microphone is measured. Speed and time difference are the quantities required to calculate the exact position of the leak.

Schematic with sample data

Correlation sample data

Note: If D/2 is displayed as the result, then the time difference is 0 and the leak is exactly in the middle of the measuring section.

What do I have to consider?

With metal pipes it is possible to place the microphones further away from each other than with synthetic pipes. This is due to the fact that the sound can propagate better in metal pipes than in synthetic pipes. The following applies: for synthetic, choose short distances for the microphones; for metal pipes, the microphones can be used easily for a longer distance.