MassFLOW-3DTM as a simulation tool for turbidity currents: some preliminary results - From Depositional Systems to Sedimentary Successions on the Norwegian Continental Margin - page 590

the bed of was roughened by fixing 160 µm sand to
the flume floor. In a second run (Case Study I-b), a
mixture of coarse quartz silt (d 50 = 50 µm) and
water was utilised on a flume roughened with
144 µm sand. In a third run (Case Study I-c),
another grain size (d 50 = 144 µm) was added to the
previous one (d 50 = 50 µm), not altering the total
sediment volume concentration, in a percentage
of 50% fine sand and 50% coarse silt. Velocity
data were collected by an Ultrasonic Velocity
Profiler (MET-FLOW UVP-DOU MX) at 2.5 m from
the inlet. The probe of the UVP was mounted
0.1 m above the bed, looking down at an angle of
45° (Case Ia) or 60° (Cases Ib-c) against the incom-
ing flow (Fig. 1). At the same location a high-speed
camera (Basler piA640-210 g m) captured the flow
through the glass side-wall of the tank.
To produce a high-density turbidity current
with a well-defined internal density stratification,
the volumetric sediment concentration was set to
21% in Case Ia and 13% in Cases Ib-c. The incli-
nation of the laboratory flume and the flow dis-
charge were then adjusted to achieve the
equilibrium slope profile of the turbidity current
(i.e. slope at which the deposition rate balances
erosion rate and no net-aggradation takes place.
These conditions were achieved by setting the
slope of the flume to 9.1° for Case Ia and 11.3° for
Cases Ib-c. The volume discharge of all flows was
9.3 m 3 /hr.

Results and remarks

Two different regions of the turbidity current for
the performed cases were analysed in detail: the
head of the current for the first run and the body
for the second and third runs.

Case Study I-a: Head of the flow -
mono dispersed mixture

The head of the current, being the first part of the
current to reach the control section (placed 2.5 m
after the inlet, matching the location of the UVP
probes in the physical model) can be recognised
easily and analysed. Fig. 2 shows the velocity and
visual data collected at the measurement location
around the arrival of the head. The velocity probe
location was fixed and the horizontal axis of
Fig. 2 represents time as the turbidity currents
passed by. The vertical axis represents the
distance above the bed. In Fig. 3, some profiles of
the head of the turbidity current, obtained from
numerical simulations with varying settings of
the key parameters, are displayed. All of the snap-
shots of Fig. 3 were taken 4.5 seconds after the
release of the sediments from the mixing tank,
which is the period of time that the flow travelled
2.3 m from the inlet to the control section. The
primary shape of the sediment cloud is largely
similar for all the simulations. However, the
velocity profiles for the various cases are slightly
different. The principal difference is the maxi-
mum velocity (red region in the snapshots)
which was located at different heights within
the turbidity current head, depending on the
simulation. It must be kept in mind that, due to
the highly turbulent and non-steady environ-
ment around the flow front, the maximum veloc-
ity region might shift abruptly from one region
to another within the head of the flow; this might
explain observed differences between simula-
tions. Even so, the velocity maximum in the top
half of the flow in Run d (Fig. 3B) seems unreal-
istic. Fig. 3C and E show a low velocity region
above the flume floor, which might be attributed
to deposition of a thin sediment layer by the
head of the current under the respective settings.
This deposition was not observed in the experi-
ment. Notwithstanding these differences, for
all the cases reported, the velocity of the cloud
is in the range of 0.4 m s −1 to 0.7 m s −1 and the
experimental front velocity was approximately
0.5 m s −1 .

Numerical simulation set-up

The flume geometry was modelled and included in
the simulation domain, which was set up using a
fixed Eulerian rectangular computational mesh. The
computational mesh consisted of 443,900 active
cells. The resolution of the mesh was uniform in all
three directions throughout the whole computa-
tional domain. The size of each cell was set to
0.0065 m × 0.012 m × 0.0036 m (XYZ).
The calibration of the numerical output with
the laboratory experiment consisted of matching
the velocity structure and spatial characteristics
of the turbidity current. The four most influential
parameters in the simulation model were evalu-
ated in order to assess their effect on flow behav-
iour: the pick-up (entrainment or erosion of
particles) coefficient, the bedload coefficient, the
turbulent mixing length and the angle of repose.
See Table 1A for the values of the parameters used
in Case Study I-a and Table 1B for the Case Studies
I-b and c.