SimCenter | Tsunami-Induced Turbulent Coherent Structures, February 21, 2018


Welcome. Today is Wednesday, February 21st, 2018 and this is the NHERI Early Career Researcher Forum. This forum is intended
to foster the exchange of ideas and be a platform of best practices for
successful research. It is intended to highlight compelling research, build up
the natural hazards engineering student community, and provide presentation
opportunities. For more information, visit the NHERI website at Designsafe-ci.org
where you can find links to the SimCenter and the NHERI Learning Center.
Today’s webinar is coordinated by the Natural Hazards Engineering Research
Infrastructures Computational Modeling and Simulation Center. This webinar is
supported by the National Science Foundation under
Awards 1612843 and 152817. Any statements in this webinar are those of the presenter and do not necessarily
represent the views of the National Science Foundation. Today’s presentation
is by Nikos Kalligeris. He is a postdoctoral scholar in Civil
Engineering at UCLA. His primary research interests lie in near shore
hydrodynamics during extreme events such as tsunamis and hurricanes. His analysis
combines numerical modeling laboratory experiments and field observations. Dr.
Kalligeris has been a member of several reconnaissance field surveys for
tsunamis around the world and the title of his presentation is “Tsunami-Induced
Turbulent Coherent Structures: Large-scale Experimental Observations
and Interpretations.” And Nikos I invite you to begin. Thank you.
Thank you for the introduction, Matt, and also thank you for giving the
opportunity to present my research on this forum. So,
my talk today as you mentioned is on tsunami-induced turbulent coherent structures
and turbulent coherent structures is the scientific term for eddies that have
been observed to form in the near shore during tsunamis. My talk today — during my PhD
when I was at the University of Southern California, I
participated in an NSF funded project where we studied the generation and
evolution of tsunami-induced turbulent coherent structures, and in today’s presentation, we will
talk about the findings from these laboratory experiments and also about the value for
harbor applications in tsunami prone areas. So, this here is a quick breakdown
of my presentation today. I will first give a brief introduction on tsunamis
and more specifically on tsunami-induced currents and then I will briefly talk
about what we know about tsunami-induced coherent structures both from
observations with field surveys and also from the literature and then in
the main part of my presentation, I will talk about the lab experiments that we
conducted, how we collected the experimental data and also the
interpretation of our findings. And at the end, I will talk about the
value of the experimental results, and how to extend this work that we did in
the future. So, real quick, tsunamis are geophysical scale waves that are predominantly generated by submarine earthquakes. If the earthquake is big
enough, the co-seismic seafloor deformation will set off waves and
these waves can reach wave lengths from tens to hundreds of kilometers. And due
to their very long wave lengths, they can travel vast distances even across oceans.
In the deep water, they have a very small amplitude but as they approached the shore, they grow in amplitude and they have the
potential to flood vast areas. Past research has primarily focused on
flooding for the obvious reason of saving lives and property in coastal
areas. And there has been analytical and numerical models have
been have been developed to predict the coastal flooding. The numerical models
typically employ the nonlinear shallow water equations which capture the
flow at a relatively low computational cost. Given accurate
initial conditions, it has been shown time and time again that these standard
models used for tsunami modeling can accurately reproduce
propagation across an ocean and as seen on the last notes but also predict
coastal flooding in coastal areas. Like an example is shown on the right.
So, tsunami research has advanced to the point that we are now able to do
accurate forecasts and for tsunami-induced flooding even in real time. But when it
comes to tsunami-induced, while tsunamis induce flooding is very well understood,
when it comes to tsunami-induced currents, things start to become a little
bit trickier. These two figures here show on the left the maximum tsunami
amplitude in Crescent City, California during the 2011 Japan tsunami. And on the
right you can see the corresponding maximum current speeds at the
same harbor and same tsunami event. It is immediately apparent from these two
plots, that maximum currents are much more localized as compared to tsunami amplitudes.
High currents can be seen around the port structures
and where they reach up to 12 knots. But in the rest of the domain, they are
relatively, the current speed is relatively low. In this animation here,
which is also for Crescent City, for the same tsunami event, will show you what is
happening during the tsunami. As you can see, while the tsunami currents approach the
near shore, they sort of bend around the coastal structures and they accelerate.
It is this interaction of the currents with the shoreline that creates the
highly localized current field that we saw in the previous transparency. Flow
separation and strong horizontal shearing near the port structures also
introduces vorticity and this vorticity in turn generates eddies. So, if the
conditions are favorable, or unfavorable if you may, these eddies may grow in very
large size and it can even occupy the whole harbor or port base. One such eddy
was spotted in Port Oarai during the 2011 Japan tsunami.
Port Oarai is located south of the Fukushima power plant and where the
nuclear disaster happened right after the 2011 earthquake. In this footage, which was
taken from a helicopter, a gigantic eddy with a diameter much much larger than
the local water depth can be seen spinning inside the harbor for tens
of minutes until it was washed away by the next incoming wave. These large
eddies come with high rotational flow speeds that can exert forces in port
structures but also move a lot of sediment around in the port. They are also
notorious for dragging ships inside these eddies and they can be
carried along with them. One of the best-known examples of a ship caught
inside a tsunami-induced eddy took place in Port Salalah, Oman during the 2004
Indonesian tsunami. Scientists went to Oman and specifically in Port Salalah
to take quantitative measurements nine months approximately after the tsunami.
It is a standard practice to interview eyewitnesses in those reconnaissance
missions and ask information about the tsunami. So, in this case in Port Salalah
the harbor master described how an hour and a half after the arrival of the tsunami and
this 285 meter container ship by the name Maersk Mandraki was caught
inside a tsunami-induced eddy and broke its moorings. It started spinning out of
control and any attempts to free the ship from the eddies via tugboats
was were left in vain. So, the ship was dragged off shore. It almost hit the
break water and eventually was beached in a nearby sandbar. This
sketch here was approximately drawn by the harbormaster who described the
event. Luckily, the ship didn’t collide with any other ship, but you can imagine
many worse scenarios of having a ship inside a port out of control. At
the time, this was a mysterious phenomenon when this was recorded and my
advisor, Costas Synolakis, attributed this to harbor resonance, but now we know that
this was not harbor resonance, but it was one of those turbulent incoherent structures
that form during a tsunami and where the flow separates around a sharp corner
like this one here, seen here. So, apart from observational reports
there is not much that we know about tsunami-induced eddies because fundamental research into
the generation evolution of those tsunami-induced eddies is practically
non-existent. For research into eddies in the context of natural processes
like the ones discussed here, we have to go to literature in the hydraulics
community and specifically studies in shallow flows. So, shallow or 2d flows exhibit weak vertical variability,
similar to what we coastal engineers call shallow water waves where the
length of the wave is much longer than the local water depth. So, one of these
eddies that we study grow in lengths that are much larger than the local
water depth. They are called Turbulent Coherent Structures. The exact
definition of which is given here. From this point on, and excuse me for the
confusion, I will use the term Turbulent Coherent Structure eddy and vortex
interchangeably, but they all mean, all these terms, mean the same thing. So, from
what we know in the literature, the predominant generation mechanism for
these turbulent coherent structures is topographic forcing, as seen in the
figure here. Topographic forcing exerts transverse shear and which
eventually leads to flow separation on the lee side and the generation of an
eddy. Tsunami-induced turbulent coherent structures are not the only types of
geophysical-scale turbulent coherent
structures that we know. One example of another known case is
shown here and it has a name of tidal flushing. So, when the tidal-induced
currents in a tidal inlet are very strong, during ebb tide, the currents at
the tip at the inlet mouth may separate and form this sort of
dipole that can self propel itself to the offshore.There has been significant
research where they’ve studied under what conditions these dipoles form
and under what conditions they separate from the inlet month and migrate
offshore. These sort of dipoles can have important implications for
nutrient exchange between the estuary and the offshore. Other types of geophysical-scale turbulent coherent structures that we know are shown here. On the left, you can see a
mesoscale eddy forming east of Tasmania. And these mesoscale turbulent coherent structures
are very important because they exhibit different properties than the
surrounding ocean and because they can carry heat and carbon and also
oxygen and salt and they can travel very vast distances, and so it has been
studied thoroughly and other geophysical-scale turbulent coherent structures that
we know also exists in the atmosphere, not only in the oceans. For example, hurricanes are a type of turbulent coherent structure structure that exhibits similar
properties like the one shown here. The atmosphere is, like the ocean, is
limited in terms of depth. So, this type of vortex can also be considered as
shallow. Even though these turbulent coherent structures are in a different scale,
and because they’re in a different scale, Coriolis force starts to become
important. There’s still a lot of information that we can extract from the
research that has done on these vortices. So, since
fundamental research is absent for tsunami-induced turbulent coherent structures,
we set up a simplified experiment at Oregon State University’s tsunami basin. When we
recreated the generation of a turbulent coherent structure using appropriate scaling,
as seen in the videos here, we collected all the data necessary to study both the
generation and evolution of those turbulent coherent structures in a well
controlled environment. The results and the analysis can be meaningfully
translated to the prototype scale for harbor applications. So, Oregon State
University’s wave basin is one of the largest in the world and it has a 44 —
it’s 44 meters long and it’s 26 and a half meters wide.
It has a piston type wave maker on the left side and vertical walls on the
other three sides. We built an impermeable breakwater along the width
of the basin leaving a three meter gap between the breakwater tip and the side
wall. And in this way, we created a simplified image of a port. I have
put a label of the offshore basin on the left side, since this is the side of the
wave maker. But based on this geometry, the port side can be either of them,
either left or right. The boundary conditions that created a stable
turbulent coherent structure is shown in the next figure. It consists of a slow push
forward of the main wave maker and a and a sudden retreat which created an
asymmetric N-wave. In terms of the chosen water depth, it translates to
an experimental scale of approximately 1/27 and the experimental
wave period and wave amplitudes both translate to very realistic protoype
scales and this is one of the rare cases where both the time and the length
vectors are left relatively undistorted in this experiment. In terms
of the dimensionless, the important dimensionless numbers, it can be
concluded that the laboratory results are scale independent for small scale
geophysical flows like the ones that we are interested in. The experiment can be
broken down into three phases. In the first phase, the wave maker pushes
forward, and it also pushes the water from the left basin towards the right
basin, creating a left-to-right current in the harbor channel, and this
creates a turbulent coherent structure on the right side of the breakwater tip. As the
wave maker suddenly retreats it creates, it changes the direction
inside the harbor channel and now it’s from right to left and
this current is further reinforced by the reflection of the leading elevation
wave off the right wall. So, this in the beginning generates a starting jet that
forms into a vortex and this vortex eventually detaches from the starting
jet and starts evolving as a free vortex in the offshore basin. And this
constitutes the third and final phase of the experiment. Four overhead cameras
were used to visually capture the water surface and surface tracers were
introduced in the flow to extract a surface velocities through a technique
called a particle tracking velocimetry or PTV, in short. It is a non-invasive
technique to extract surface velocities over big areas. We repeated the same experiment many many times to examine both the
repeatability of the experiment but also to collectively obtain a satisfactory
density of data. This rectified video shows one of the PTV experiments as
recorded by the four overhead cameras. I had to speed up the video for this
presentation. So, you can see how the water pushes the water to the right side
creating initial TCS and then as the flow reverses the starting jet
evolves into a vortex that starts evolving freely after it’s detached from
the trailing jet in the offshore basin and you can see that the tracers
accumulate at the vortex center in what is known as a flow convergent
zone near the center. And we have to keep supplying the flow with with tracers to
extract velocities from this area. The the vortex eventually occupied the
whole offshore basin and circulation was still visible inside the wave basin
even one hour after the experiments. So, in order to — the methodology — to
extract surface velocities from this type of videos consists of four basic steps.
The first step is to detect the center of those tracers and the second step is
to track them in time. Then you have to convert the image coordinates to world
coordinates and then extract surface velocity vectors and physical units. This
animation here corresponds to the rectified video I showed you before. Each of these velocity vectors
corresponds to one of the tracers that was tracked in the flow. The velocity in
the basin reached one metre and a half meters per second which can be
translated to a prototype 15 knots which is very high speeds in these types of
flows that we are studying. The methodology and implementation of
particle tracking velocimetry was validated through measurements of ADV instruments
that were positioned into in the locations shown in the figure here. ADV
stands for acoustic doppler velocimetry and these are instruments
that measure the velocity in all three directions, x and y. And as you can see,
the black here, the black curves in those top plots, correspond to ADV
measurements and the red corresponds to the PTV extracted velocity. So, as you can
see from these plots that they compare well. The velocity extraction was done carefully. Now, we also used a stereo camera
configuration for the area around the the breakwater tip in order to extract
velocity vectors in more detail. What this stereo camera configuration allows is to
extract velocities and the location of the tracers in all three directions, in
what is known as stereo or 3D PTV. Stereo 3D PTV requires an additional step
compared to 2D PTV which is the stereoscopic particle matching between
the tracers and tracked with one camera to match them with the tracers
captured with the other camera. So, since we can extract surface
elevation through 3D PTV, I could compare my results using the wave gauges that
were laid out in the wave basin, the location of which is shown
with red crosses in this figure here. Black again is the measurements and
this time of the wave guages and the red is the 3D PTV extracted surface
elevation. As you can see, the implementation of a 3D PTV worked
really great in this example here. In terms of the tools I used for
the 2D and 3D PTV are shown here. For the camera
intrinsic parameterization which has to do with the distortion introduced by the
lens of the camera, I used the camera calibration toolbox
developed by Dr. Bouguet of Caltech and you can download this toolbox through
this link. In terms of the camera extrinsic parameters, which has to do
with image to world coordinates transformation, I used the Direct Linear Transformation
equations and I’m providing here a reference to
implement these equations. And last in terms of the particle
identification and particle tracking, I used the methodology of Crocker and
Grier, and more specifically, the implementation through the MATLAB
toolbox at the link provided here. For a 3D PTV, I had to develop my own
PTV toolbox which is based on the work by Capart et al. And the
difference here is that for my 3D PTV experiments, the inter-particle spacing
was smaller compared to the flow speed. So, this made it more complicated
for conventional PTV algorithms to follow the tracers. So, we needed an
extra matching criteria and the work by Capart et al uses the topology of
the neighboring tracers and through Voronoi five-stars which worked really
great in the application I wanted to use it for. In terms of the stereoscopic
particle matching, again, I developed my own toolbox which is based on the
work by Douxchamps and others, 2005. Even though I have developed my own toolbox, I
haven’t released those yet. I’m planning to do so in the near future in case
this can be useful for the research of others. Also, if anyone from the audience
wants to know more about this implementation, I can provide more
details at the end of my presentation. Now, after I went through the data, the methodology of collecting data, I
will now move into the analysis and I will start from the second phase of the
experiments where the offshore TCS is being generated. Now for this application, I overlaid all the scattered velocity
vectors from 2Dand 3D PTV, and I averaged them on a regular grid like seen
in this animation, and by obtaining the mean velocity field on a regular grid,
I was able to calculate the spatial derivatives and obtain important metrics
that describe the flow. One of those important metrics is swirl strength.
Swirl strength is a metric that describes the local rotation of the
fluid similar to vorticity. Maps of swirl strength shown here were used to track
the location and the spatial extent of starting-jet vortex and eventually
the turbulent coherent structure. The zero swirl strength contour as shown with the green
polygons, denoted the boundary of the starting-jet and eventually the vortex.
What I was interested in, in this stage, was to see how the starting-jet and
the vortex grow and how quickly it grows and how it scales. So, in other words, I had
to find a dimensionless number that will make the data to collapse
on a straight line. And if you find this dimensionless number, then you can
translate this to prototype scales, and I was inspired by experiments done in
vortex rings and I found this dimensionless number which gave me the
right scaling as shown here. The other thing that I was interested in,
in his experimental stage, was to see at what time the vortex separates
from the trailing jet that keeps supplying it with kinetic energy. So, this
time step corresponds to this time shown here where the
vortex has detached from the jet. This can also be shown through a
plot that shows the growth of circulation and circulation denoted
with capital gamma is a metric that gives you the strength of the vortex.
And you can see that it is at this time that the vortex separates which is also
shown by this rapid decay of circulation. Again, I had to find a
dimensionless number that will tell me what are the important parameters in the
flow with which the vortex detachment time scales with which can again be
translated to the prototype scale for harbor applications. I believe that the
most interesting physics happen after this vortex gets detached from the trailing
jet in what is the third experimental phase. For this experimental stage, the
first thing I had to do was find the path that the experimental
turbulent coherent structure followed in the offshore basin. To achieve that, I used two
different methods to track the center of the vortex. In the beginning,
where I had a nice distribution of tracers, I could compute vorticity
maps and through those vorticity maps I could define my vortex center as the
location of maximal vorticity. But at later stages of TCS development,
because of flow convergence and divergence near the vortex center, in
some instances I was left with very very little tracers to work with and compute
vorticity maps. So, in these cases, I used the tracer conglomerate the vortex center and I detected the
boundary of this tracer conglomerate through image processing and the
center of mud mass of these polygons defined my vortex center. So, after going
through this tedious work of extracting the path of the TCS center, you can get a
plot like this which shows the path of the experimental vortices for all the
22 experiments that we did. So, seeing this plot, I was, I could see that
repeatability is fairly good because the vortices followed
approximately the same path even though this is a highly turbulent
flow which means that there is a lot of there are many stochastic elements involved.
And the second observation that intrigued me by looking at this plot was
that the vortices in all the experiments started to converge towards the same
location and inside the offshore basin. So, this was an
observation that I had to explain and I will get back to as to why this happened
in a few slides. Extracting the the vortex center allowed me to convert
all the scattered velocity vectors from a Cartesian coordinate system to a
TCS centered coordinate system, and I could compute the velocity vectors
in the azimuthal, so along the arc of a circle, and the velocity azimuthal
direction, also the radial which is a way or towards the center of a circle.
And from the scattered velocity data, I could obtain the mean velocity profiles
which was a convenient representation of the flow field since my vortex was not
perfectly, the experimental vortex, was not perfectly axisymmetric. So, through
those mean velocity profiles, I could find out what rate the maximum
azimuthal velocity was decaying with time. So, by following the the peaks of each
mean velocity profile in time, you can draw a decay curve like the one shown on
the right. We know from the literature that
the predominant and the key mechanism in shallow turbulent coherent structures is
bottom friction and we were looking for a simple analytical formula that can
explain this decay and we can use this analytical formulation also
for harbor applications, and it would be great to have one to use. A simple
formula can be derived by equating the angular momentum with the
bottom friction with a bottom friction force and if you make the
simplification that you have a perfectly azimuthal flow, so no velocity in the
radial component, and also that the vortex doesn’t grow very quickly in time,
in terms of the spatial growth in the radial direction, then this force balance
comes down into a first order differential equation where you can
compute the decay of max azimuthal velocity with time. If you plug in the
only free parameter which is the bottom shear stress, you can get a green curve
which is the prediction of the analytical decay and it perfectly
matches the experimental decay which is great news because this is a very simple
formula that we can use for harbor applications to see how quickly those turbulent
coherent structures decay in time. As the turbulent coherent structure was slowing down, it was also growing in size by obtaining
more and more ambient fluid from its perimeter in what is called
viscous diffusion. The growth rate predicted by viscous diffusion is shown
here with the blue dashed line. The measured vortex radius is shown with the
black circles and as you can see up to this point it’s following the
theoretical prediction of viscous diffusion. But, as you move further in time, you can see
that the vortex area starts to fluctuate, and it stops following the
theoretical prediction. Again, this was an intriguing finding
that I’ve had to somehow explain. It occurred to me that it
might have something to do with the vertical, with the spatial confinement of
the vortex in the offshore basin. So, once you plot the minimum distance to the
vertical boundaries, you can notice that the vortex radius was confined, but
by those vertical boundaries which was the sidewalls of the basin and the
breakwater. In other words, the vortex, the experimental vortex, was constantly
trying to find more space to grow and the reason why all the TCS centers
ended up in the same location was because this location allowed it to grow
to its maximum size it could go for this
particular geometry that we tried in the wave basin. As a matter of fact,
this, the turbulent coherent structure, that was observed in Port Oarai
also ended up approximately in the center of the port basin. So, again
this is an important observation that we can predict where this, where these large
eddies will end up. And we can also use viscous diffusion as an approximate method to predict the spatial growth of
those turbulent coherent structures in the prototype scale. The last thing I wanted to do is to characterize the turbulent coherent structural flow field and so I went
through the literature to try to find a vortex profile that fits my observations.
It can be shown that, in theory, any monopolar polar vortex will converge to the
so-called stirring vortex. The stirring vortex has an idealized flow
profile as shown in this plot here and it has a free parameter alpha that
controls the steepness of the velocity profile. The higher the alpha is, the more
unstable the vortex becomes and by unstable I mean that the single vortex
can break down in a multi-polar vortex or break down into more individual
vortices. The name of stirring vortex is derived by the way it can be recreated
in a laboratory. In the lab, you spin, you submerge, a cylinder inside your
fluid, start spinning the fluid inside the cylinder and then lift the
cylinder, and let the rotating fluid interact with the ambient fluid
and your flow profile will start converging to this a-profile and certain
profiles have have been used and obviously in the literature describes
cyclonic vortices like the recent hurricane Harvey. So, with that in mind, I
started fitting my scattered experimental data to this theoretical stirring profile.
And you can see, that this free parameter alpha is approximately 0.35 based on the
experimental results. This is much lower than alpha equals 2 that has been found
to fit the atmospheric vortices and this vertical profile has been derived
for using the root mean square error as a goodness of fit parameter. I was able
to determine that the scattered velocity data started to converge towards
the a-profile at around 400 seconds which is very meaningful because
this shows you that there is some time after which we can use this theoretical profile
to describe a tsunami-induced turbulent coherent structure. The only question
that remains is what happens before this key time here. Why didn’t the scaled
velocity data follow this theoretical prediction at previous times? And the
clue lies, the clue to answer this question lies in the radial velocity
components. In shallow flows, the radial velocity and the vertical
velocity components are described as secondary flow components because
they are much weaker compared to the azimuthal. But as you can see from the
plots here in earlier times, the radial velocity component was significant and
at around 400 seconds, it started to become insignificant. So, in other words,
the deviation from the perfect azimuthal velocity profile was the reason why
the experimental vortex didn’t follow this theoretical stirring profile at
earlier times. By computing the ratio of the kinetic energy in the
radial direction over the kinetic energy in the azimuthal direction, I was further
able to demonstrate that at around 400 seconds was what is called the
transition to quasi 2-dimensional flow. This has important implications because
when it comes to mixing and sediment transport, but also when it comes to the
use of numerical models. Because the standard numerical models used for
tsunami modeling are depth-averaged which means that everything that happens in
the water column is averaged out, it means that in the beginning, while
the flow is highly three-dimensional, those models will not capture the
physics accurately, but there is a time after which they will start reproducing
the physics more more accurately. So, in summary, in the context of near shore
currents induced by tsunamis, we started, we created the
generation and evolution of a tsunami induced current in the laboratory. And we
we collected all the data necessary to study both its generation and evolution.
And the results that were yield have important implications for applications
for ports and harbors where the timescales of developments of TCS
development can be applied for. And the forces on port structures can be
computed as well as the sediment transport potential. The laboratory data can be further used to validate numerical models used
for tsunami forecasting. And if better numerical models are derived for
forecasting, means that decision-makers can make, can manage, ship evacuation
during a tsunami more efficiently. My last aspiration is that my research will
have an effect on future port and harbor design in tsunami-prone
areas. We are currently in the process of modeling the Oregon
experiments using a large eddy simulation model which is a fully 3D
model with very small, with very high resolution. We are testing different
turbulence schemes including the Standard Smagorinsky and Dynamic
Lagrangian, and the results will be validated through the
laboratory data. We have already made a lot of progress already and the results are promising. And once
we fully validate the methodology, we can examine the 3D flow structure of the
flow and the turbulent properties of the flow as well. Also in the future,
using this implementation, we can test different boundary conditions and
different geometries in the harbor. This is an ongoing collaboration between
the University of Southern California and the University of Delaware. So, while
this study has already yielded very interesting results, there are still
things that can be improved. For example, because I only collected surface
velocity information on the water surface, I couldn’t
examine the full 3D structure of the flow. Also, I wasn’t able to study the
turbulent properties of the flow through in-situ measurements. Hopefully, the large
eddy simulations will offer insights in these two areas.
But it would be very good if these can be studied also in the laboratory in
the future. Last, since I only studied one boundary condition one harbor configuration, I
wasn’t able to fully examine under which conditions wave induced TCSs form, and
examine also how the time causation to quasi two-dimensional scales with time.
The last thing I want to show is a video captured in the lab. We put some model
ships at the tip of the breakwater to simulate how ships can be entrained
in those turbulent coherent structures. Just like the unfortunate ship Maersk Mandraki was captured in an eddy in Portal Salalah in Oman. You can see how
this video demonstrates how difficult it would be to control the ship
under those circumstances. I would like to acknowledge all the
sources of funding and this project was funded by an NEES-NSF grant awarded to
my advisor Professor Lynett. And I was also supported by two fellowships at
USC. As I said, it was Professor Lynett’s project and I was very fortunate to be
involved in. And I would also like to thank my colleagues, Aycut Ayca and Adam
Ryan, for their help during the the experiments.
Thank you very much for your attention. Thank you Nikos very much for the
presentation. At this time, we’ll start the question and answer session.
Attendees are reminded that questions should be submitted through the chat
panel and directed to the moderator and Nikos, the first question starts out
with a comment which is: A great discussion, a very interesting topic. How
could the insights provided by the research be useful to emergency managers
to improve ship evacuation before tsunamis, and what would your
takeaway advice be to harbor masters and emergency planners in coastal areas? Okay.
So, the first question was how this can be used — this information – can be used for ship evacuation. So, what we what this fundamental research
does is to show the first order problem. So, we have a very
simplified geometry that doesn’t necessarily correspond to any harbor
geometry that you can find out in the field, in the prototype scale. But it
shows what the important physics are in the flow. So, once we figure this out then
we can improve the numerical models to capture the right physics. So, if we
improve the numerical models, then we can improve the forecast, right? And
by improving the forecast, then emergency managers can make
better decisions of which parts of the port have to be first evaluated.
But there’s an answer, there’s another takeaway message that whatever —
which is a very simple one — that wherever you have very sharp corners inside the
harbor, then you can expect these eddies to form, right.
This is very simple, and if the conditions are favorable then you can
use some of those simplified analytical equations that came out of my
research to predict how quickly those eddies will grow in size and what the
flow structure looks like in terms of the velocity profile, and so you can
derive some simple models to calculate both forces on structures, see what the sediment transport potential is, etc. Okay.
Oh thank you. Are there any field measurements of tsunami induced eddies
that would help corroborate this research? We don’t have quantitative
measurements per se of those eddies. There are some, there are very limited
data out there in the literature that measure tsunami-induced currents in harbors.
We published a paper in 2016 where we measured GRL where we measured
tsunami induced currents in a harbor in Southern California,
in Ventura during the 2015 Chilean tsunami but the measurements that we
took mostly corresponded to the starting jet and not in a fully developed eddy
like the one I studied in my experiment. Now, as part of my post doc
at UCLA, we are laying out a methodology to be ready to deploy instrumentation,
and in a future tsunami, where we can capture the flow field of a fully developed TCS.
We don’t know if we’re ever gonna be successful in
such an attempt, but we will try. Okay Another question is: can you remind me
how the radius of a vortex was estimated? Okay. So, the radius of the vortex — and I
have included this. It was extracted from vorticity maps. So, I was setting a
threshold to maximum vorticity. So, at each time I had a maximum vorticity that
correspond to the center of the vortex. If you can define your vortex
boundary, as for example, ten percent of the maximum vorticity at each time, and
this contour of ten percent of maximum vorticity defined the vortex boundary.
That is how I defined it in my study. Okay. And that’s how those black
points were extracted. I think we have time for one more question and it is: from your
experience working at OSU’s wave basin, what suggestions do you have for those
starting their PhDs who plan on conducting experimental research in
fluid mechanics. The most important aspect when it comes to experiments in
fluid mechanics is the experimental scaling. So, if you don’t get the
experimental scaling correctly, your results are meaningless. And they cannot
be meaningfully extrapolated in the full-scale. So, that’s the most
important message. Another couple of tips is that it was always
important to do some quality assessment of the data as you collect them.
For example, each day or each week, make sure that your data is good.
Not go back home without anything meaningful to work with. Also,
from my research, I found out that there’s always room to develop your
own ideas when it comes to data collection. It’s always good to read the
literature in what tools are already available out there, but don’t be
afraid to to develop your own that suit your application better. Okay,
Well in fact, we’re at the conclusion of today’s Early Career Researcher Forum.
On behalf of the attendees, thank you Nikos for taking the time to share your
research. To the attendees, thank you for your participation and questions. Please
check the SimCenter’s website at simenter.designsafe-ci.org and check
your inbox for emails from [email protected] that will have
registration links for additional upcoming webinars in the Early Career
Researcher Forum and the Natural Hazards Engineering 101 webinar series.
Thank you for attending today’s forum.

Leave a Reply

Your email address will not be published. Required fields are marked *