Earlier this week we fitted a couple of level sensors to some pretty large kerosene tanks, each containing 200,000 litres. The sensors have an accuracy of 0.5% and then the telemetry device we use to connect these sensors to our internet platform have the same. There will be errors in the calibration as well, such as measuring the tank and the specific gravity of the product (we are using pressure sensors and therefore the density matters). It's pretty fair to assume then that you'd expect the system to be 1% or more out, so about 2,000 litres.
I looked at some data earlier as we are recording data every hour so we can see how much the readings vary over time. This was easy for me to do even though the tanks are more than 100 miles away from me as we have internet connected sensors so I could jump online and see the readings securely via our web portal. Over 48 hours we saw a change of just 321 litres, or 0.16% of the tanks capacity, so what does this mean..?
First we need to understand the different terms you often hear associated to the performance of instrumentation and sensors:
Repeatability deals with how consistent a particular sensor is against itself. It describes the ability of a sensor to provide the same result, under the same circumstances, over and over again.
So what we have with this tank level monitoring system is excellent repeatability. In order to determine the accuracy then we would need to calibrate against a known source. Not easy but regular comparisons against a flowmeter either putting product into or out of the tank can give this. You then need to consider the calibration of the flowmeter, which opens up a whole new can of worms.