I have no experience with ADV or turbulence study, but most stream and flood measurements from my discussion with others can be done in 15-30 minutes on moderate size streams, once set up.  But if you have noticed, there is some degree of flow pulsing and varying conditions in turbulent waters, standing waves, excess oxygen and perhaps sediment to contend with.  I have heard both oxygen bubbles and excess suspended sediment can affect ADV readings.  Whether it is worth the time to record low, mean and high or max velocities observed at a point over a specified time period or surge cycle, I don't know, it may depend on your objectives.  For stream flow measures, the mean point velocity is desired.  But if calculating forces and ability to dislodge and move particles, the max might be desired, along with density of water including temperature, barometric pressure, dissolved and suspended materials and as well bubbles.  In using stream pigmy and type AA stream current meters many times for channel discharge estimates, point readings may last 30 seconds or longer to obtain a desired number of rotations and varying units of time to use table.  So to assess a point turbulence with ADV is probably much easier in laminar flow, and In turbulent water, I would stay in place long enough to sample the variance through one or more surge cycles to gain what I need, perhaps a mean or a high.

I do not know turbulence studies of air or water. I guess the length to measure will change. 

The longer the more accurate the average value, however the velocity field will probably change during this period. I think it is a trade-off. 15 min seems to be widely selected by many researchers of boundary layer study.

Despite of measuring time, I would recommend you to calculate its uncertainty first. By having acceptable (1 - 2%) uncertainty, you would be able to estimate the number of recordings or minimum measurement time. 

Hi Praveen

It depends on your objectives and on the type of flow you are measuring. To obtain time-averaged velocities you may need only 30s but for 2nd (for example turbulent kinetic energy) and 3rd (for example Reynolds shear stresses) order moments you will need to measure for more time.

If you estimate the uncertainty and define the uncertainty you find acceptable for your study you can then define the time needed for your measurements.

I would first recommend you made a long time measurement to define the uncertainty and then decide on the time needed.

Check out these papers:

Buffin-Bélanger T & Roy AG. 2005. 1 min in the life of a river: Selecting the optimal record length for the measurement of turbulence in fluvial boundary layers. Geomorphology 68(1-2):77–94. doi:10.1016/j.geomorph.2004.09.032

Discussion of “Turbulence Measurements

with Acoustic Doppler Velocimeters”

by Carlos M. García, Mariano I. Cantero,

Yarko Niño, and Marcelo H. García

December 2005, Vol. 131, No. 12, pp. 1062–1073.

DOI: 10.1061/(ASCE)0733-9429(2005)131:12(1062)

Quaresma AL, Ferreira RML & Pinheiro AN. 2017. Comparative analysis of particle image velocimetry and acoustic Doppler velocimetry in relation to a pool-type fishway flow. Journal of Hydraulic Research. 55(4):582-591. doi: 10.1080/00221686.2016.1275051

Hope it helps.

Dear Praveen

As other said, the time frame is dependent on the objective and the available resources. The time resolution of 15 min is acceptable and you can change the resolution if required by looking up the data of 15 min and change it accordingly.  

I hope this paper might helpful for you

https://torroja.dmt.upm.es/pubs/2013/sillero_jimenez_moser_PoF13.pdf

In order to quantify most turbulent statistics, Buffin-Bélanger and Roy (2005) suggested an optimal record length (minimum sampling effort to achieve low standard errors) ranging between 60 and 90s with ECMs (20Hz) and ADV (25Hz). 

Depending on the statistics you are interested in, you probably need longer series at16 Hz.

Article 1 min in the life of a river: Selecting the optimal record l...