CHAPTER 11: EXPERIMENTAL WORK ON TRIANGULAR PIPES
11.4 Comparison with Numerical Simulations
11.4.2 Change in the Loading Behavior
We set out to observe experimentally the loading behavior change in the concentration evolution
for the equilateral triangle cross-section. Our experimental results and their comparison with
simulations are new; experiments in triangular pipe as a way to study passive tracers evolution
are not reported in the literature. Nevertheless, we are interested in pushing this study forward to
observe the predicted loading behavior change in the concentration and consequent sign change in
its skewness.
In particular, for a delta initial condition, the skewness evolution in time is reported to change
sign at non-dimensional timeτ
≈1, as shown by the black curve in figure 11.13 from [4, 7]. It should
be noted that the maximum magnitude of the negative skewness is very small (<0.05) compared to
the maximum of the positive skewness (>0.25). Here, the value of the Péclet number isP e= 10
4,
which corresponds to a flow rate of0.3712cm/s, about an order of magnitude larger than those we
have been dealing with so far.
Figure 11.13: Skewness evolution in time from [4, 7]. Results from fifty simulations (black), each with
10
6random walkers,
P e= 10
4, and maximum time-step
10
−4. Initial conditionδ(x)
and spanwise
uniform. Short time (red) and long time (blue) asymptotics.
In order to observe the sign-change, we need to have a flow rate that allows for the switch to
happen before the dye reaches the end of the pipe. A similar image can be created for our flow rates,
to see how far off we are from the sign change.
Figure 11.14 (left) shows results of Monte Carlo simulations with
10
5particles at flow rate
0.0732
cm/s. The blue
+indicate how far an experiment with such flow rate can be considered
reliable; that is, when the rightmost particle of the cloud reaches the right edge of the pipe. While
we are able to see the skewness decreasing in value and heading towards zero, we cannot observe the
sign change. Consequently, we have performed a numerical study to identify a flow rate value that
would allow us to see the sign change without losing any information. Figure 11.14 (right) shows the
skewness evolution in time for five flow rates. In all of these cases, the visible part of the skewness
curve (highlighted by the +’s) reaches below zero. The lower the flow rate, the further the+’s go.
10-3 10-2 10-1 100 101 log( ) -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Cross-Sectionally Averaged Skewness
Cross-Sectionally Averaged Skewness Evolution
experiment simulation 10-3 10-2 10-1 100 101 log( ) -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 Averaged Skewness
Monte Carlo simulation of Averaged Skewness Evolution 10s-20s 10 exp 10 full 15 exp 15 full 20 exp 20 full 25 exp 25 full 30 exp 30 full
Figure 11.14: Skewness evolution in time for pipe of equilateral triangle cross-section. Left: Skewness
evolution in time for flow rate
0.0732cm/s. Simulation performed with10
6random walkers and
maximum time-step10
−3. The red curve is the simulation result, while the blue+’s indicate how
much of the simulation would be visible in experiments with the same flow rate. Right: Skewness
evolution in time for five flow rates; top to bottom:
0.0113
cm/s,
0.0169
cm/s,
0.0225
cm/s,
0.0281cm/s, and
0.0338cm/s. The
+’s indicate how much of each simulation would be visible in
experiments with the same flow rate.
We have been working on experimental runs with these much lower flow rates to be compared
to the corresponding numerical simulations. We choose to focus on the
0.0169cm/s
flow rate and
perform longer experiments to observe the sign-change in the skewness. In particular, this much
lower flow rate requires experiments lasting
90min
(with pictures taken every1s) in order for the
dye to run along the entire length of the pipe.
Figure 11.15 shows results of such experiments compared to Monte Carlo simulations. The red
curve is obtained by merging together the two concentration curves produced by the injection and
reservoir cameras. Note that while att= 850sthe skewness value is positive, for all three of the
later times the skewness value is negative. The spikes that appear at later times are due to the
increasing amount of noise present in the data; as time passes and the solute diffuses more and
more throughout the pipe, its concentration diminishes and the amount of recorded noise increases.
Nevertheless, albeit less pretty, the results are reliable.
-10 0 10 20 30 40 50 60 70 80 90 0 0.1 0.2 t = 0 s, =0 MC Exp -10 0 10 20 30 40 50 60 70 80 90 0 0.05 0.1 t = 850 s, =0.8217, Skew = 0.033764 -10 0 10 20 30 40 50 60 70 80 90 0 0.05 t = 1550 s, =1.4984, Skew = -0.012161 -10 0 10 20 30 40 50 60 70 80 90 x/a 0 0.05 t = 2850 s, =2.7551, Skew = -0.02529 -10 0 10 20 30 40 50 60 70 80 90 x/a 0 0.05 t = 3450 s, =3.3351, Skew = -0.026212