This page is about how to play with any hot-wire data. As far as I know, Matlab is the best and simplest tool to analyze the hot wire data. No matter in which format the hot-wire data is saved(such as ncid, hdf5, dat, txt etc), matlab is simple and flexible enough to read them all. Before going further you might wanna make sure that the hot-wire data you have is a good one. If your data is very bad, this could result in faulty calibration and the whole post-processing of hot-wire data will be useless. And the whole process is nothing but a total disaster. You can see the typical hot wire signal below shown in picture which is a good one. If you wanna know what exactly bad hot-wire data means you can check out the pics in this folder. Those sudden unwanted peaks in the signal are a cause of change of temperature or dust in the atmosphere or any other experimental problem.

If you have matlab installed on your computer then its a very easy task to have fun with the hot-wire data. You can find tons of articles on theoretical aspects of hot-wire anemometer, but here I am going to discuss about the post-processing of the hot-wire data. I will try to explain the things in as simple words as I can.

1. Calibration of Hot-Wire:

This is basically the most important and basic step. Typically you will have the hot-wire data in terms of Voltage. The purpose of calibration is to find the relationship between the hot-wire voltage and actual velocity of the flow(in other words 'a fit'). If you have gone through the theory of hot-wire, you might have came across the King's Law. For the sake of calibration you need to have the estimate of mean velocity of the flow at the hot-wire probe location. It can be done with the help of pitot tube or cup anemometer. Basically it can be done with the help of any velocity measurement device.

Kings Law says: E² = AVⁿ + B, E is the voltage from hot wire, V is the mean velocity measured with the help of any device such as pitot tube etc. A, B, n are the constants. n is the constant which is in the range of 0.45- 0.5. For this value of n, you can refer to literature or to the approximation you can assume it to be 0.45.

For the current discussion I am assuming it to be n=0.45. For the sake of calibration it is preferable to have as many data points as possible for accurate fit between E & V. Its better to explain this with the help of simple jet-flow example:

In the following figure, you can see a round hole in the big glass frame. Imagine the air is flowing outside the hole which we can call it as a jet-flow. You can also see the hot-wire which is mounted on the adjustable stand. Now for the sake of calibration, choose any suitable location of hot-wire probe next to jet. Now for each different operating speed of jet record the hot-wire signal. Say you record the hot-wire signal for 10 different operating speeds of jet. During the same experiment you have to measure the mean velocity of the flow with any device such pitot tube or cup anemometer etc

By this time, you must be having 10 hot-wire signals corresponding to 10 different operating speed of jet.

For 10 different hot-wire signals take the mean of each signal and save it accordingly.

In Matlab: For 1st signal - E(1) = mean(1^st hot-wire signal in voltage); V(1) is the mean velocity estimated from device

repeat the same procedure up to E(10) = mean(10^th hot-wire signal in voltage); V(10) is the mean velocity estimated from the device

After this constructing a fit in matlab is very simple.

p = polyfit(V^0.45,E²,1);

This command will return you two values in the workspace of matlab p(1) = A & p(2) = B

Use this A & B to transform the hot-wire voltage signal in to Velocity signal as follows.

V = ((E²-B)./A)^{(1/0.45) ;}

This is the instantaneous velocity signal you just got from the above equation. Now you are ready to extract any statistical quantities from this velocity signal.