Hi David, 

What exactly was the problem? Assuming the fluctuations are below the frequency response of the tubing system (which they should be since it sounds like you didn't use a tubing system) and below the response of the transducers, then all you do is take the mean of the pressure - as in, sum the data points and divide by the total number of data points. 

There are other statistics that you can use as well, including rms of the pressure time history and the variance of the pressure time history.

Hope that helps

Ric

Thank you Ric.

The measurement was taken in the exhaust line, so tubing is inherently part of the measurement. 

My problem with respect to taking the mean is how do i decide the range of data points(in this case the measurement time duration in the x axis)? This is trivial if the data points have repeating pattern (partly true for Idle and 1600 rpm). However, the data points at 2350 rpm and 3300 rpm did not have any well defined pattern.

I am suspecting that the choice of measurement duration will affect the results even if i take the mean.

Regards,

David 

Hi David, 

There are several approaches to take here:

1. you must remember that the sample mean is only an estimate of the true mean. Because it is an estimate, you can define confidence intervals of the mean estimate, which can be lowered by increasing the number of independent samples. If you are only measuring for a few independent samples, this may not be enough to get a good estimate of the mean. This means that you should sample for longer to get a better estimate of the mean. The longer - the better.

2. If you have non-stationary data, then you might have to take a different approach. Non-stationary data means that its statistical properties are variant with translations in time. So something like a moving average (low pass filtering) might give you a better estimate of the properties of the time series. BUT, in general, your data should be stationary, at least when viewed on a long time scale, so just take lots of samples for a long time. 

So in summary, if you sample for long enough, the sample mean should be an accurate estimate of the true mean. Consult Bendat and Piersol 2010 for more information. 

Hope that was helpful

Ric

Thank you very much Ric for the informations and explanations.

It is quite helpful indeed.

David