Partially Mixed Region in Stratified Fluids


experiment image
Research performed by Amenda Chow and Tyler Pittman,
May - August 2006,
Department of Mathematical and Statistical Sciences, University of Alberta

Background

A stratified fluid is a fluid with varying densities. Furthermore, a stable stratification arises when lower densities always occur above heavier densities. Our oceans and the atmosphere are well known examples of stable stratified fluids. But what happens when an oil spill dumps into the ocean or a toxic gas is accidentally released into the atmosphere? More importantly how do these toxins behave in their respective stratified environments. Because of the lack of scientific research done and obvious importance, our experiments attempt to model partially mixed regions in stratified fluids.

The partially mixed region is an isolated uniformly dense area in a stratified fluid much like the oil spill or toxic gas in the ocean or atmosphere, respectively. The movement of this mixed region through the stratified fluid is known as an intrusion. Our main focus is how this intrusion affects the ambience. We concentrate heavily on how the internal waves created by the intrusion disturb the stratified fluid. Results from this can help environmentalists determine how fast the oil spill spreads or how long it takes for the toxic gas to become harmless.


Experimental Setup

The experiments were performed in a rectangular glass tank measuring 48.4cm in height, 17.4cm in width and 197.3cm in length. Using a standard double bucket system, the tank is filled with uniformly stratified fluid to a maximum height of approximately 33.0cm.

After the tank is filled and ample time is given for the stratified fluid to settle, a conductivity probe traverses vertically through the tank providing us with a density profile of our uniformly stratified fluid.

A gate is inserted inside the tank 18.5cm from the edge to the bottom of the mixed region and this section of the tank is now known as the lock. The gate acts as a barrier preventing fluid from entering the lock. By lowering the gate only to the bottom of the mixed region, this reduces the waves caused by the pulling of the gate and allows us to be confident that the waves created below the mixed region is a result of only the intrusion.

A square screen of horizontal black and white lines is placed behind the tank just after the lock. The screen is required for synthetic schlieren. This digital image process records the distortion of the image of horizontal lines due to the bending of light through various density gradients.

Tank Setup of a Partially Mixed Fluid

At the lock we place our electric motor 1cm from the surface of the fluid and mix a uniform region of desired height (in this case, heights of 5, 10, 15cm are created). We make another traverse with our conductivity probe and then wait for the tank to settle before releasing the gate.

While waiting for the tank to settle, we setup our camera. Each experiment is recorded from just before release of the gate until the tank settles using a digital camera placed 300cm in front of the screen. The recording is digitized and analyzed using the image processing program, DigImage.


Analysis Methods

Most of our analysis is done on DigImage which is specifically designed for analyzing fluid flows. One of its features is it can create horizontal time series (HTS) (see Results to view actual HTS images) of the intrusion as it moves along the tank. Using the HTS, we calculated useful characteristics of the intrusion such as intrusion speed.

Density Profile of a Partially Mixed Stratified Fluid

DigImage also creates false colour processed vertical time series (VTS) images called DNT images (see Results to view actual DNT images) which represents the change of the buoyancy frequency (N) squared. DNT images illustrate the movement of wave packets created by the intrusion with respect to time. They visually enhance the internal waves created by the intrusion and allow us to identify crests and troughs, as well as determining the wavelength. We also made approximations on wave properties such as N2t (range) and vertical displacement (amplitude) using these DNT images.

As well, we took horizontal slices from stored VTS and DNT files, translated them vertically and interpolated the gaps between each slice to come up with an image in x,t coordinates. These xtdnt images allow us to calculate the speed of the wave packets (horizontal phase speed).

Aside from using DigImage, we also investigated the density profiles of our stratified fluid (see diagram at right). We determined a piece-wise function to approximate the density profiles and modified code in the C language that calculated the thickness of the interface and the mean depth of the interface. We also determined the slope of our density profiles which we needed for N and collaspe time of our intrusion.


Results

(rhoBottom ~ 1.10g/L)

Displayed below are 12 images corresponding to three experiments with rhoBottom~1.10g/L. The first experiment is of a 5cm mixed region (no dye), the second a 10cm mixed region (no dye) and the last a 15cm mixed region (red dye). Wave crests and troughs are visualized by bands of red and blue respectively. Red and black corresponds to a range of -0.9 sec-3 while pink and white corresponds to a range of 0.9 sec-3.

Here we have chosen 3 movies illustrating the release of a partially mixed region in a stratified fluid. The fixed parameters of the associated experiments are: rhoBottom~1.10g/L and lock=18.5cm. The movies are split into two screens, the upper part (black & white) is the raw image of the experiment corresponding to the intrusion of the partially mixed region and the lower part (colour) is the processed image data representing the wave packets.

Duration=30sec Duration=30sec Duration=30sec
Length=30.0cm Length=30.0cm Length=30.0cm
Height (B & W)=5.8cm Height (B & W)=9.5cm Height (B & W)=12.0cm
Height (Colour)=21.0cm Height (Colour)=17.5cm Height (Colour)=16.0cm
video of 5cm mixed region video of 10cm mixed region video of 15cm mixed region
image from video_E060609a image from video_E060609b image from video_E060623b

Using horizontal time series images (here in false colour), we calculated the speed and observed that as the height of mixed region increases, the greater its speed.

Time (vert. axis):~34sec Time (vert. axis):~34sec Time (vert. axis):~34sec
Distance (horiz. axis):~33.0cm Distance (horiz. axis):~33.0cm Distance (horiz. axis):~33.7cm
Horiz. time series at z=30.9cm Horiz. time series at z=27.9cm Horiz. time series at z=27.2cm
E060609a_hts96.jpg E060609b_hts131.jpg E060623b_hts144.jpg

These vertical time series are taken from actual images of the experiment.

Time (horiz. axis):~34sec Time (horiz. axis):~34sec Time (horiz. axis):~34sec
Height (vert. axis):~5.0cm Height (vert. axis):~10.0cm Height (vert. axis):~15.0cm
Vert. time series at x=13.6cm Vert. time series at x=13.6cm Vert. time series at x=13.3cm
E060609a_vts149.bmp E060609b_vts149.bmp E060623b_vts146.bmp

Using the DNT images (contours of the time derivative of the perturbation squared buoyancy frequency), we measured the period of the waves and observed that as the height of the mixed region increases, the period also increases.

Time (horiz. axis):~34sec Time (horiz. axis):~34sec Time (horiz. axis):~34sec
Height (vert. axis):~30cm Height (vert. axis):~30cm Height (vert. axis):~30cm
Vert. TS of DNT at 13.6cm Vert. TS of DNT at 13.6cm Vert. TS of DNT at 13.3cm
E060609a_dnt149.bmp E060609b_dnt149.bmp E060623b_dnt146.bmp

Results

(rhoBottom ~ 1.02, 1.04, 1.06, 1.08, 1.10g/L)

The 12 diagrams above highlight only the set of experiments with rhoBottom~1.10g/L. In the second portion of our results sections, we provide an overview of the 5 sets of experiments conducted with varying densities (15 experiments in total) and focus on observed frequency and buoyancy frequency. Below we present our findings in two different forms: a plot and a table.

Plot of Observed Frequency vs. Buoyancy Frequency

Associated values for Plot
RhoBottom (g/L) Height of Intrusion Buoyancy Frequency (N) Observed Frequency

(cm) (1/sec) (1/sec)
1.02 5 0.708 0.442
1.02 10 0.708 0.398
1.02 15 0.738 0.321
1.04 5 1.005 0.439
1.04 10 1.005 0.376
1.04 15 1.019 0.546
1.06 5 1.264 0.668
1.06 10 1.264 0.566
1.06 15 1.264 0.676
1.08 5 1.514 0.806
1.08 10 1.514 0.476
1.08 15 1.514 0.542
1.10 5 1.920 0.885
1.10 10 1.920 0.706
1.10 15 1.633 0.495

From the above plot and corresponding table of values, we make several observations. Clearly, as rhoBottom increases, N increases, which implies the period of the waves descreases. Buoyancy frequency appears to exceed the observed frequency for all the experiments. There is no strong evidence suggesting that a larger mixed region has an effect on either frequency.


References and Related Research


Acknowledgements

First we would like to thank NSERC and CFCAS for their financial support. Thanks to James for designing our motor, Caspar for building it, Josh for the conversation, Joseph for always being so accommodating, Geoff for listening and Bruce Sutherland for everything else. We learnt a lot but most importantly we learnt the power of sharing knowledge. Thank you!


Return to Research Interests page


Bruce's home page

Department of Physics

Department of Earth and Atmospheric Sciences