Help with FFT for time-pressure signal

I'm studying various flow regimes in two-phase flow, like bubbly flow, slug flow and annular flow. It is expected that at each particular flow regime, the pressure and void fraction data is fluctuating at distinct frequencies. So my final goal is to have amplitude Vs frequency plots for each flow regime with distinct peaks. This way I can classify the flow regimes accordingly. An example image in attachments(fft).

I've a time series data for differential pressure (and void fraction) with following details --

Fs = 320, N = 3200,

The data sometimes fluctuates between +ve and -ve values. Almost all the times, the average is not zero. So when I take fft of the data (alternatively with hanning window and zero padding at 4096 points), I get huge zero frequency dc offset, which is always the average value of the signal. If I delete the dc offset with filtering, The amplitude data I get is not right. And also the peaks I'm getting are not distinct as I expect and as other researchers have obtained. Attaching an excel file (F02) with two columns, first being time and second the pressure.

Someone please help me! I've tried originlabPro as well and I keep getting the same results. If I'm missing any information about my problem, let me know. I'll submit is asap.

Thanks...

[Umair Khan]

Hello Dear,

Can you please contact me. [e-mail]

I am doing the same research.

Answer 1

[Sandro Lecci]

Hi, to remove the huge zero frequency dc offset I suggest you simply subtract the trace's average (in order to have a signal oscillating around zero), see the attached picture.

However, it looks like your signal is not as clean as the one of the other researchers. Do you know exactly what to expect from this dataset?

Best, S

[Vikrant Viks]

Yes, the signal is not really clean. But I expect, as shown in the attached image (fft), distinct peaks for individual flow conditions. In bubbly flow, the air bubbles are continuously distributed in water and flow at high velocity. So the pressure fluctuations are not very high, but more or less constant. In case of slug flow, a big slug of air passes intermittently inside water at high speed. So at every passing of this slug of air, pressure fluctuates rapidly. This intermittent fluctuation of pressure in case of slug flow should be visible as a peak in pressure fft. In case of bubbly floe pressure fft, it won't be present.

Also I was wondering if psd or dwelch are alternatives that I can look for fft?