adjusted vertically to an elevation in the water column specified for each
experiment. For example, in some test cases the vertical elevation was mid-
depth, so in a distorted version of the experiment the water depth increased, and
the probe position needed to be raised.
Before commencing measurements, a box was placed around the laser beams
to create a condition of zero flow velocity. Measurements were then taken to
establish "offsets" in the system due to local temperature effects on the analog-
to-digital converter cards. These values were recorded in the experiment log, and
the offset test was repeated at completion of the experiment to see if any
significant drift in the electronics had occurred during the test. If the difference
was not too great, an average was taken, and the offsets were applied to the
measured time series of turbulent velocity fluctuations. Greater variation in the
offsets necessitated repeating the test.
For each test, flow was initiated on the flow table and sufficient time was
allowed for the flow to reach a quasi-steady state condition before beginning data
acquisition. The laser Doppler system was programmed to record two
components of horizontal velocity at each grid point. Time series data of
velocity fluctuations were collected for 10 sec at a rate of 100 Hz per channel.
At completion, the traverse moved to the next point, waited by 0.5 sec for
movement vibrations to dampen, then repeated the process. At the end of the
experiment, data were converted to engineering units and moved to a more
modern computer system for subsequent analysis and plotting.
Inherent in this measurement scheme is the assumption of quasi-steady flow
conditions. Previous experiments with this facility had shown that the time-
averaged flow components in the nonturbulent portion were reproducible. Initial
testing at the start of this project examined the duration of sampling needed at a
point to get a reasonably stable mean value, and 10 sec was found to be well
above the threshold necessary. However, in the far field well away from the exit
structure, slowly oscillating, large-scale gyres develop which would require a
much longer sampling duration to establish a repeatable mean value.
Before each experiment, the surfaces of the flow table glass bottom were
cleaned of minute particles that might interfere with the laser beams. However,
sometimes a small particle in the flow would deposit at a grid location and
partially block one of the laser beams. When this happened, either no data were
acquired, or the measured data were significantly smaller than actual. Where
measurement "dropouts" occurred, it was usually obvious in the vector plots
because all the surrounding values were uniformly larger.
The following sections provide details about each of the test series
conducted. Representative results and comparisons are given. Complete results,
mostly in graphical form, are included in Appendix A, Appendix B, and
Appendix C.
Case 1: Flow Separation at Vertical Edge
Free Jet
The first set of tests was performed on a free jet constrained by a vertical
boundary with a sharp edge as illustrated in Figure 18(a). In this scenario, it was
expected that only strong horizontal turbulence components would be generated,
40
Chapter 5 Turbulence Scale Effects Experiments
Previous Page
