Escolar Documentos
Profissional Documentos
Cultura Documentos
2015,27(1):1-23
DOI: 10.1016/S1001-6058(15)60452-8
Abstract: An overview is provided of CFDShip-Iowa modeling, numerical methods and high performance computing (HPC),
including both current V4.5 and V5.5 and next generation V6. Examples for naval architecture highlight capability and needs. High
fidelity V6 simulations for ocean engineering and fundamental physics describe increased resolution for analysis of physics of fluids.
Uncertainty quantification research is overviewed as the first step towards development stochastic optimization.
1. Paradigm for development SBD for ship hydro- k - / k - based isotropic and algebraic stress model
dynamics (ASM) based anisotropic RANS, and DES approaches
Rapid advancements in simulation technology with near-wall models or wall functions. A single-
are revolutionizing engineering practice, as SBD and phase level-set method is used for free-surface captu-
ultimately virtual reality are replacing current reliance ring. Captive, semi-captive, and full 6DOF capabili-
on experimental observations and analytical methods. ties for multi-objects with parent/child hierarchy are
It is expected that a major shift in how scientific me- available. The fully discretized propeller or body-
thod forms its basis of conceptual truth, a shift from force propeller model can be employed for propulsion.
reliance on observations, based on experiments, to re- The water-jet propulsion can be included using actual
liance on logic, based on simulations supported by ex- water-jet with detailed simulation of the duct flow or
periments. SBD covers a broad range from compute- water-jet model with the reaction forces and moments.
rized systems based methods to solutions of physics Incompressibility is enforced by a strong pressure/ve-
based initial boundary value problems (IBVP). Present locity coupling, achieved using either PISO or proje-
interest is in solutions of physics based IBVP for ship ction algorithms. The fluid flow equations are solved
hydrodynamics. SBD for ship hydrodynamics merges in an earth-fixed inertial reference system, while the
traditional fields of resistance and propulsion, seakee- rigid body equations are solved in the ship system.
ping, maneuvering, open-ocean and littoral environ- Other modeling capabilities include semi-coupled two
mental effects, and offers new opportunities for future phase air/water modeling, environmental waves and
ships to meet challenges of the 21st century. Develop- winds, bubbly flow, and fluid-structure interaction.
ment SBD involves new paradigm for hydrodynamics
research in which CFD, experimental fluid dynamics
(EFD), and uncertainty analysis (UA) are conducted
simultaneously for benchmark geometries and condi-
tions using an integrated approach along with optimi-
zation methods, all of which serve as internal engine
guaranteeing simulation fidelity. International collabo-
rations with other research institutions and organiza-
tions include participation in ITTC and NATO AVT
working groups and naval engineering educational
consortium (NEEC), organizing international CFD
workshops and current NICOP projects. Those activi-
ties are mutual-beneficial and magnifying individual
institute capabilities, which has been foundational in
the unprecedented achievements of computational
ship hydrodynamics.
ward difference Euler scheme. Since the solver is de- terms by the second-order explicit Adams-Bashforth
signed for high-Reynolds number flows, the transport method. The pressure Poisson equation is solved to
and re-initialization equations are weakly elliptical enforce the continuity equation. The convective terms
and thus pentadiagonal line solvers in an alternate-di- are discretized using the fifth-order WENO scheme.
rection-implicit (ADI) scheme are used. A MPI-based The other terms are discretized by the second-order
domain decomposition approach is used, where each central difference scheme. The pressure Poisson equa-
decomposed block is mapped to one processor. The tion is solved using a semi-coarsening multi-grid so-
resulting algebraic equation is solved with the PETSc lver from the HYPRE library.
toolkit using block Jacobi incomplete factorization The code is parallelized via a domain decomposi-
(ILU) pre-conditioners and bi-conjugate gradients sta- tion (in three directions) technique using the MPI li-
bilized (BCGSL). All equations of motion are solved brary. All inter-processor communications for ghost
in a sequential form and iterated to achieve converge- cell information exchange are in non-blocking mode.
nce within each time step. Parallel I/O using MPI2 have been implemented such
Extension of CFDShip-Iowa Version 4.5 to that all processors read from and write to one single
Version 5.5 with a fully coupled two-phase flow so- file simultaneously[9]. In order to speed up the compu-
lver using the volume-of-fluid (VOF) method is in tations and improve the accuracy and efficiency for
progress. The approach includes implementing the hi- very large grid simulations (billions of grid points),
ghly accurate geometric VOF interface tracking me- some enhanced technologies have been implemented
thod developed for V6, developing fully-coupled two- such as semi-Lagrangian advection schemes and opti-
phase flow solver, implementing cavitation and mix- mized memory usage. The water/air interface is extra-
ture models for air/water/vapor three-phase interaction, cted as PLY polygon file format for post-processing.
and developing capabilities for the necessary applica- A multi-block grid capability has been recently incur-
tions. The numerical methods, HPC, and SBD functio- porated into CFD Ship-Iowa Version 6.2.
nal areas are similar to Version 4.5. The VOF method Development of the general curvilinear grid so-
has been implemented into V5.5 to replace the level lver, V6.3, is in progress, which is built on the success
set method for the interface tracking and incorporated of V6.1 and V6.2 to achieve all functionalities of V4.5
into the single phase flow solver. The current version and beyond. CFDShip-Iowa V6.3 is aimed at the high-
of the code works with single-phase flow solver, fidelity, high-resolution simulations of fully coupled,
multi-block grids, turbulence, and full 6DOF motions multi-scale, multi-phase, turbulent ship flows with
without overset grids. Figure 1 shows free surface de- fluid-structure interactions utilizing billions of grid
formation around the Numerette Planing Hull for points. The approaches include finite volume method,
fixed static condition. multi-block, body-fitted, general non-orthogonal cur-
vilinear structured grids, overset background
2.2 V6.1, 6.2 and 6.3 modeling, numerical methods, Cartesian grids, and highly modularized, developer-
HPC friendly code structure written in Modern Fortran
The next-generation high-fidelity SBD tools, (2008) and MPI.
CFD Ship-Iowa V6, are already under development The second-order finite volume method with ac-
for milestone achievement in increased capability fo- curate geometric approximations for non-smooth,
cusing on orders of magnitude improvements in accu- non-orthogonal structured grids is used for the discre-
racy, robustness, and exascale HPC capability. tization. A generic transport equation is solved for
In Version 6.1, Cartesian grids are used with im- momentum components and scalars with central diffe-
mersed boundary methods for complicated geome- rence and high-order upwind schemes used for face-
tries[5], and the level set based ghost fluid method is centered value reconstruction. Exact projection me-
used for sharp interface treatment and fully two-phase thod is implemented for machine-accuracy mass con-
coupling with the VOF method for interface tracking. servation where central difference and high-order up-
Extension to orthogonal curvilinear grids was made in wind schemes for contra-variant volume flux reconst-
V6.2[6] with enhanced technologies for the interface ruction at cell face centers. Scalable MPI communica-
modeling[7,8] and similar numerical methods and HPC tion using new MPI-3 features will be implemented
capabilities as V6.1. and MPI sub-array data type is extensively utilized for
A finite-difference method is used to discretize scalable MPI communication and I/O in V6.3.
the governing equations on a non-uniform staggered
grid, in which the velocity components are defined at 3. Naval architecture
the cell face centers. All other variables are defined at
the cell centers. Time advancement is based on the 3.1 Resistance and seakeeping, captive and free ru-
semi-implicit four-step fractional step method. The nning maneuvering, free running course keeping,
diagonal diffusion terms are advanced with the seco- and intact and damaged stability
nd-order Crank–Nicholson method and the other Resistance and seakeeping predictions are inclu-
4
ded in Gothenburg 2010 (G2010) and upcoming vanced turbulence and propeller models for improve-
Tokyo 2015 (T2015) workshops. Prediction of resista- ments in the CFD predictions of static and dynamic
nce is the oldest application of CFD in ship hydrody- PMM maneuvers. Overall, the average error for capti-
namics and its accuracy has been significantly impro- ve maneuvering simulation was 13.6%D . The largest
ved since Gothenburg 1980 (G1980), the first CFD error values were generally observed for pure yaw and
workshop held in 1980. In G2010, 89 submissions of static rudder simulations. For linear derivatives, the
resistance prediction are documented, which is the lar- average error was much larger for yaw moment
gest number in the workshop series[10]. More than (40% D) than sway force (15% D ) . For nonlinear de-
90% of the simulations were conducted using grids rivatives, the average error value was about 40%D .
smaller than 10M points. The resistance prediction si- Free running maneuvering simulations were reported
mulations were carried out for a wide range of appli- for limited cases in the SIMMAN 2008 workshop.
cations and conditions. Other than resistance, sinkage The maneuvering simulation included standard ma-
and trim, local flow fields such as boundary layer and neuver test cases such as turning circle and zigzag.
wake, and wave patterns were also predicted by many The results showed 6%D error for trajectories for tu-
simulations. Different geometries including tankers,
rning maneuver prediction while larger errors
container ships, and surface combatants were studied
(13% D) were obtained for zigzag maneuver. The grid
at a range of very small to large Froude numbers ( Fr ) .
sizes were from 0.4 M to 14.9 M points for these si-
The simulations showed average error of 3.3%D for
mulations. For most SIMMAN 2008 computations,
resistance for both low and high Fr while sinkage the propulsion was implemented as an axisymmetric
and trim showed less errors for high Fr . The average body force distributed in the propeller disk. The body
error for sinkage/trim at low and high Fr was 9.7 / force was specified in a non-iterative manner in which
11%D and 35 / 55%D , respectively. For seakeeping, the ship wake on the body force was neglected. Rece-
several seakeeping test cases were included in G2010 ntly, Wu et al.[12] used Yamasaki propeller model cou-
with numerous contributions for each case. CFD com- pled with the RANS code to give a model that intera-
putations of seakeeping have been rapidly increasing ctively determines propeller-hull interaction without
since Tokyo 2005 CFD workshop (T2005) in which requiring detailed modeling of the propeller geometry.
there was only one forward-speed diffraction case Yamasaki model is based on a potential theory formu-
with no motions. The applications for seakeeping pre- lation, in which the propeller is represented by bound
dictions included a wide range of wave conditions, vortex sheets on the propeller disk and free vortices
Froude numbers, and motion conditions. Similar to the shed from them downstream of the propeller. Wu et
resistance test cases, different geometries including al.[12] showed the Yamasaki propeller model could
tankers, container ships, and surface combatants were predict successfully the asymmetric wake field. In ad-
studied. Grid sizes ranging from 0.4 M to 71 M points dition, the propeller rpm was predicted with less than
were used with a clear trend toward increasing accura- 0.5%D error for Yamasaki compared to 12%D for
cy with grid size. The CFD predictions are assessed
separately for 1st order vs. 2nd order terms. The mean
value of resistance and the amplitude of motions were
considered 1st order terms whereas the amplitude of
resistance and mean value of motions were considered
2nd order terms. The simulations showed large avera-
ge error for the second order terms (44% D) while the
average error was less than 15%D for the first order
terms.
Captive and free running maneuvering simula-
tions are included in the SIMMAN 2008 workshop[11]
and upcoming SIMMAN 2014. The applications for
captive predictions included PMM-type forced motio-
ns such as static rudder, static drift, pure sway, pure
yaw, and yaw and drift conditions for different geo-
metries. For SIMMAN 2008, 16 submissions were re-
ceived for the forced motion simulations, comprising
different CFD-based methods such as RANS, URANS,
and DES. Grid sizes ranging from 2.1 M to 250 M
points were used. It was concluded that finer grids
were needed especially for the rudder and appendages
and in regions of large vortices, as well as more ad- Fig.2 Turning maneuver simulation with water-jet propulsion
5
Fig.3 The grid topology and propeller vortices for KVLCC2 free running simulations with fully discretized propeller
non-iterative axisymmetric body force. Free running including IIHR wave basin[15,16]. For calm water, it
simulations are also conducted with more advanced was shown that the average system based prediction
propulsion system such as water-jet. Sadat-Hosseini et error drops from 16%D to 8%D using the
al.[13] performed maneuvering simulations for a cata- maneuvering coefficients and rudder forces estimated
maran and validated the results against the experime- from CFD free running instead of those from captive
ntal data (See Fig.2). The simulations were conducted experiments. For waves, Araki et al.[14] showed that
either for bare hull with integral force models for the mathematical model with wave loads estimated
water-jet or with actual water-jet with body force im- from CFD outputs provides better prediction for
peller defined by pump curves. Turning maneuver si- maneuvering in moderate following and quartering
mulations showed average error of 9% D - 22.6% D waves, compared to the original mathe-matical model
for CFD simulations with minimum error for the ac- with the wave loads computed from slender body
tual water-jet simulation. Zigzag maneuvers showed theory. However, the improved mathematical model
larger errors. In addition, the extremely large over- was too stable in severe waves and unable to predict
shoot angles in zigzag showed the deficiency of the instabilities such as periodic motion or broaching.
water-jet propulsion system for maneuvering. Since For SIMMAN2014 workshop, Sadat-Hosseini et al.[17]
CFD is computationally expensive for maneuvering in conducted simulations for free running maneuvers of
comparison to system based (SB) methods, some stu- KVLLCC2 in calm water using body-force propeller
dies have focused on improving the SB mathematical model and actual propeller (see Fig.3). The grid size
model by using CFD with system identification me- was 6.8M-8.4M for different cases. The computational
thods. Araki et al.[14] employed CFD free running out- cost was 3-5 times higher for the simulations with the
puts to improve a 4DOF mathematical model develo- actual propeller. The results for turning maneuver
ped for maneuvering in calm water and following showed E = 6.6% D using propeller model and
waves. The CFD predictions were first validated agai- much less error ( E = 2.2% D) using actual propeller.
nst the experimental data from different facilities
6
Fig.5 The predicted transom free surface and vortex structures for turning maneuver simulation
Similarly, zigzag simulations showed better prediction data. The unsteadiness spectra at few points in the
using actual propeller. Sadat-Hosseini and Stern[18] transom wave field showed f 1.5 scaling. The reso-
performed maneuvering simulations for 5415M test lved turbulence kinetic energy was 86% in the tran-
cases of SIMMAN 2014 using twin counter-rotating som region. The simulations showed Karman-like in-
propellers based on body-force propeller model with stability at transom, horseshoe vortices at the juncture
total grid size of 6.7M points (see Fig.4). The results of strut-hull and strut-shaft, and shear layer instability
showed about E = 12% D for turning and zigzag 20/ at the strut-hull intersection. Figure 5 shows the predi-
20 while larger errors were shown for zigzag 10/10. In cted transom wave field and vortex structures. Com-
addition, Sadat-Hosseini and Stern[18] conducted sys- pared to straight-ahead condition, the Karman-like fre-
tem-based simulations for 5415M maneuvering in quencies were 3% higher while others were 8%-35%
calm water. The maneuvering coefficients were found lower for turning. In addition, the predicted frequency
from system identification using CFD outputs. To es- for Karman-like, horseshoe and shear layer vortex
timate the coefficients, parallel processing technique shedding in turning showed 2.4%, 3.7%-7.7% and
was used in which CFD free running data for several 8.6% asymmetry, respectively.
turning and zigzag maneuvers were first combined There are few simulations conducted to investi-
and then used to estimate one set of maneuvering coe- gate free running course keeping and instability. Stern
fficients. The system based predictions showed an and Toxopeus[21] and Sadat-Hosseini et al.[22] perfor-
average error of 5.30%D , 12.64%D and 4.67%D med course keeping simulations in calm water, regular
for trajectories for turning 35, zigzag 10/10 and 20/20, and irregular waves for the fully appended 5415M
respectively. ship hull, in collaboration with NATO AVT 216 se-
Among free running maneuvering simulations, ssion “Evaluation of Prediction Methods for Ship Ma-
there are very limited studies on local flow. Recently, neuvering and Control”. The results were validated
Sadat-Hosseini et al.[19] studied DES predictions of the against the experiments not only for the ship motions
local flow including transom wave field and vortex st- but also for the loads on the appendages. The results
ructures in turning maneuver. Similar study was pre- showed good prediction for the trajectories and loads
viously conducted only for straight-ahead condition[20]. on the appendages ( 10% D) even for very complex
The mean and unsteadiness of transom wave field
geometries with dynamic stabilizer and rudders (see
were predicted with 9%D and 11%D error while the
Fig.6). Comparing the irregular wave results with the
trajectories were predicted with 3%D . The asy- results computed from regular wave simulations at se-
mmetry of mean wave field was significantly under veral discrete wavelength conditions showed that the
predicted due to surprisingly large asymmetry of EFD
7
Fig.7 Overall vortical structures predicted by CFDShip-Iowa URANS (a2, b2) and DES (a3, b3) predictions on adapted 84M grid for
5415 with bilge keels
ship has similar motion in both regular and irregular vortical strength by 10% and over predicted turbulent
waves with same wavelength condition. The course structures by 35%, when compared with the experime-
keeping simulations focusing on intact instability are ntal data. DES predicted unsteady flow with up to
summarized in Stern et al.[1], showing good prediction 95% resolved turbulence. DES mean flow predictions
for different instabilities including parametric roll, were quantitatively comparable to that of URANS, but
broaching and capsize, surf-riding, and periodic mo- were over predictive for both velocity and vortical and
tion. For damaged stability, Sadat-Hosseini et al.[23] turbulent structures. DES showed grid induced separa-
showed good prediction for both ship motions and tion inside the boundary layer and modeled stress de-
water heights inside the compartment for damaged pletion. The former was resolved by using delayed
ship in calm water and waves. DES approach, whereas the latter issue was unreso-
Overall, free running simulations have been in- lved. For 5415, URANS provided reasonably good ag-
creasing in past few years and it is expected that the reement with the experimental data, but under predi-
future challenges and method development efforts for cted the vorticity magnitude and boundary layer bulge,
modelling, numerical methods and HPC will focus on and over predicted turbulent structures at nominal
free running rather than captive simulations. In addi- wake plane. In DES, the resolved TKE levels were
tion, more research will focus on improving the SB less than 3%, thus the results were unacceptable. No-
mathematical model by using CFD since CFD is com- netheless, for the first time it provided plausible de-
putationally expensive in comparison to SB methods. scription of the overall vortex structures, and helped in
understanding the sparse experimental data. Overall
3.2 Turbulence firm conclusions were not possible since grid and tur-
Prediction of turbulent viscous flow for ship hulls bulence modeling errors could not be separated and
is of central importance and focused topic at CFD sparseness of experimental data, especially for turbu-
Workshops since G1980 to most recent G2010. Veri- lence variables and onset and progression of 5415 vor-
fication and validation of CFD predictions have been tices.
performed for tanker KVLCC2, container KCS and NATO AVT-183: Reliable Prediction of Separa-
surface combatant 5415 hull forms at straight ahead ted Flow Onset and Progression for Air and Sea Vehi-
conditions. In recent workshop extensive local-flow cles research effort for the sea facet focused on procu-
analysis was performed for KVLCC2 (bluff body) and rement of detailed experimental data using PIV tech-
5415 (slender body) focusing on the effect of turbule- niques, and evaluation and validation of CFD predi-
nce modeling. URANS with anisotropic turbulence ctions using different codes by NATO members[24-26].
model performed better than isotropic model. For The study focused on three ship hulls: KVLCC2 at
KVLCC2, URANS under predicted axial velocity and
8
Fig.10 WAM-V hydrodynamic modeling for CFD simulations in regular head waves compared with EFD sea trials in random seas
Fig.11 WAM-V 2-post testing (a) and 6-post suspension simulation (b) using CFD results as inputs. Comparison of payload accele-
rations is shown (c) for 6-post simulation (white) and 2-post test data (grey)
reported correctly. Therefore, the simulations were re- deep and shallow water. Free-running simulations of
peated with the revised conditions but the errors were T-Craft in turning and zigzag maneuvers in deep and
still large and thus more studies should be conducted shallow water and in calm water and waves are also
to evaluate the experimental setup. In addition, the ac- carried out. Recent analyses showed that the resista-
curacy of the axisymmetric body-force propeller nce and moment due to cushion pressure distribution
model for propulsion in shallow water should be inve- inside the cavity is significant for seakeeping cases
stigated. while not considered in the initial simulations. The
improved results will be published for captive valida-
3.4 Advanced hull forms and fluid-structure intera- tion studies and free-running demonstration simulatio-
ction: ACV/SES, WAM-V, planing hulls ns.
CFD studies of advanced hull forms impose sig- The wave adaptive modular vessel (WAM-V) is
nificant challenges due to complex and multi-discipli- an ultra-light flexible catamaran that conforms to the
nary modeling requirements, very high speeds intro- surface of the water through a collective suspension
ducing different physics than conventional ships, and and is modularly designed enabling a wide variety of
difficulties in validation studies due to limitations in applications. The springs, shock absorbers, and ball
model testing and limited measurements in sea trials. joints articulate the vessel such that the hulls can
Modeling requirements are different for specific hulls, move semi-independently and along with the inflate-
e.g., fluid-structure interactions (FSIs) including ble pontoons adapt to the water surface/waves to miti-
multi-body dynamics (MBD) for suspension systems gate structural stresses and reduce drag. WAM-V ca-
and finite element (FE) modeling for flexible hulls. pabilities are implemented in CFDShip-Iowa inclu-
ACV/SES capabilities are implemented in ding: LS_IBM (level-set immersed boundary) method
CFDShip-Iowa including cushion models, seal models, for treatment of the gap between pontoon and hinged
air-flow over the above water seals and superstructure, pod, a two-body dynamics model for hinged pod mo-
decoupled cushion cavity flow, waterjet propulsion tions, and a jet force model moving with hinged pod
with side forces and yaw moments induced by nozzle for free-running simulations. Captive calm water veri-
rotations and reverse buckets, and air-fan propulsion fication and validation studies are carried out with
model. Validation simulations are carried out for a average error of 5.7% D [29]. Validation against full-
combined SES/ACV ship (T-Craft) for captive tests in scale sea trial data and coupling with MBD modeling
11
Fig.17 Average, min and max of EFD (experimental fluid dynamics) strain with its expected value (EV) and standard deviation (STD)
at peak compared with CFD/FE predicted strain for sea state 3 most probable wave condition and Fr = 2.9
face photo from EFD indicating very close agreement. ( H / ) values for the immediate wave, as well as
CFDShip-Iowa V5.5 simulations with volume of fluid averaged values for the last 2, 3, 4, and 5 waves are
free surface solver showed negligible effects on resi- calculated. In group 1, slam pressures correlate 100%
stance and motions, while the extension of the jet with smaller / L and larger H / for the last 3
spray flow was resolved better than V4.5 level-set so- waves. For groups 2 and 3, strongest correlations are
lver. Regular wave results for USNA experiments (run for larger H / averaged over the last 2 and 3 waves,
43 and 44) and CFD simulations using CFDShip-Iowa respectively. Considering all the slams in all 3 groups,
V4.5 and NFA solvers are compared in Fig.14 for mo- strongest correlation is found for smaller / L from
tions and slamming pressure. The phase of the heave
the last 3 waves and larger H / from the last 2
and pitch is well predicted, while the amplitude of the
waves. Type-2 slams characterized by containing only
numerical simulations is greater than measured expe-
one pressure peak (re-entering pressure) with smaller
rimentally. Pitch motions at twice the lowest freque-
peak values and shorter duration are identified both in
ncy are not evident in simulations performed using ei-
EFD and CFD.
ther CFDShip-Iowa or NFA. Single point pressure
FSI studies[34] are carried out for the Numerette
measurements show good agreement for slam duration
planing hull (slamming load test facility at Lehigh
while the re-entering pressure amplitudes are under-
University) to provide a better understanding of sla-
predicted for both codes. A smaller time step may be
mming using benchmark full-scale validation EFD
needed to capture the peak pressure. The emerging
data. The studies are conducted in collaboration with
peak pressures are missed in NFA simulations while
Dr. Joachim Grenestedt of Lehigh University. Initially
captured in CFDShip with close agreement. Irregular
rigid body CFD simulations are conducted for both
waves simulations are validated with good agreement
bare hull and appended hull with sterndrive unit and
in terms of expected values and standard deviations of
body-force propeller model excluding the superstru-
motions, accelerations, and slamming pressures. Sla-
cture. The predicted motions and loads are used for
mming statistical studies are carried out for both expe-
one way coupling with FE model for composite bo-
rimental data and simulation results and validation re-
ttom panels to evaluate displacement, strain, and stress.
sults are shown in Fig.15 for slamming pressure. Ex-
CFDShip-Iowa is used for CFD simulations and the
treme slamming events are studies both for EFD and
commercial FE code ANSYS is used as structural sol-
CFD by examining the standard score for re-entering
ver. Studies are carried out in calm water ( Fr = 0.7)
pressure ( z P = ( P EVP ) / SDP ) . For EFD, 4 slam
and different regular head waves conditions at Fr =
events with z P 2 and 6 events with 1 z P 2 are 0.7, 2.24 and 2.9. CFD/FE results show good predi-
detected. These events are found to correlate with ship ction for displacement, strain, and stress distribution
motions, namely the vertical velocity of the ship bow for both starboard and bottom panels. Figure 16 shows
at the time of impact. CFD studies are carried out to the panel force and displacement for a regular head
provide further insight by correlating the extreme slam wave simulation with Sea-State 3 most probable wave
events with relative bow/wave motions as well as his- conditions at Fr = 0.7 . Figure 17 shows EFD and
tory of previous zero crossing waves. The CFD extre- CFD-FE strain for a regular head wave simulation
me events are grouped in 3 categories: z P 1.5 (3 with sea state 3 most probable wave condition at
events), 1 z P 1.5 (4 events), and 0 z P 1 (14 Fr = 2.9 . Two-way coupling will be implemented by
events). For each slam event, wavelength over ship first using modal analysis with added mass modeling,
length ( / L) and wave height over wave-length and then fully coupled CFD-FEA. FSI V&V studies
14
Fig.18 Vortical structures with Q - criterion (left) and energy spectra of the streamwise velocity in the shear layer (right)
are also planned for slamming loads on Athena semi- metry for both applications.
planing hull.
4.1 Single- and two-phase vortex shedding
In Koo et al.[37] the two-phase turbulent flow past
4. Ocean engineering an interface-piercing circular cylinder was studied
Simulations of 3-D unsteady separation (vortex using large-eddy simulation with a Lagrangian dyna-
shedding) around offshore structures and wave run-up mic subgrid-scale model. It was shown that the air-
induced by ocean waves are still challenging for ocean water interface makes the separation point more dela-
engineering applications. Recently, the capabilities of yed for all regimes of Re and the air-water interface
state-of-the-art CFD codes for vortex shedding and structures are remarkably changed with different
wave run-up are investigated in ITTC ocean enginee- Froude numbers. However, the deep flow did not dis-
ring workshop held in Nantes, France October 17-18, play the correct single-phase flow behavior due to the
2013. The capabilities of CFDShip-Iowa V4.5 and deficient grid resolution and non-conservative conve-
V6.2 for these applications are reported in Yeon et ction scheme, among other issues, employed with
al.[35] and Yoon et al.[36]. The studies focused on the CFDShip-Iowa V6.2. Yeon et al.[35] conducted a detai-
flow around single/multiple cylinder(s), a typical geo- led study of the single-phase vortex shedding around a
15
Fig.20 EFD and CFD comparison of mean wave field for single cylinder cases
Fig.21 Mean and 1st harmonic amplitude of the wave field cases
Yoon et al.[36]. The simulations were conducted in re- model. Validation studies focused on averaged wave
gular head waves for various wave conditions inclu- height at crest/trough and 0th, 1st and 2nd harmonics for
ding / D = 4.7 and / D = 21.9 with 1/30, 1/16 wave elevation and horizontal force. CFD predictions
and 1/10. Sensitivity studies are conducted for the ef- were assessed separately for 1st and 2nd order variables.
fects of grid distribution, domain size and turbulence The averaged wave height at crest/trough and the 1st
17
Fig.22 Flow charts of three different strong coupling schemes in one time step for fluid-structure
harmonics were considered as the 1st order variables, wave from the total wave field. However, the total
whereas the 0th and 2nd harmonics were considered as wave field could show the diffraction wave at upst-
the 2nd order variables. In addition, the wave field pa- ream which was more dominant for / D = 4.7 and
ttern around cylinder(s), vortex shedding, and intera- steeper waves. Figures 20 and 21 show the wave field
ction among cylinders were analyzed. The grid sensi- for single and four cylinder cases. The studies on vor-
tivity for the 1st and 2nd order variables was 3.17% and tex structures showed more vortex shedding for longer
77.74%, respectively, both less than the facility bias wave conditions as its longer wave period provides
estimated from the provided experimental data from enough time to develop vortices around the cylinder.
two facilities. Nonetheless, the 2nd order variable sen- For both wavelength conditions, the vortex shedding
sitivity was large indicting the need for finer grids to is more at instants the wave crest is located near the
resolve 2nd order terms. The domain size sensitivity cylinder as the flow field velocity is larger. Lastly, the
was also very small, 1.14% and 2.34% for 1st and 2nd comparison of four and single cylinder cases shows
order variables. The turbulence model sensitivity was that the interaction of cylinders increases wave trough
conducted using URANS and DES and the sensitivity for 4%-10% while the wave crest increases about 9%-
for 1st and 2nd order variables was 1.64% and 6.55%, 25%. The largest interaction effect is found for the
respectively, suggesting that the URANS turbulence shoulder side of the cylinders.
model is sufficient for the validation studies. The vali-
dation studies showed 10%D error for wave crest/
trough, 7%D error for the 1st harmonic of wave ele- 5. Fundamental physics
vation, and 70%D error for the 2nd order variables in-
cluding the 0th and 2nd harmonics of wave elevation. 5.1 IBM for idealized and practical geometries
The horizontal forces also showed 9%D error for the Immersed boundary methods are simple and effi-
1st harmonic amplitude while larger errors are predi- cient approaches for many problems with complex
cted for the mean and 2nd harmonics. The detailed geometries and moving boundaries, thanks to the rela-
study of the wave field showed that the mean wave xation of the requirement of generating boundary-fi-
field elevations are similar to the free surface eleva- tting grids in numerical simulations. CFDShip-Iowa
tions for a cylinder in a steady flow due to the large V6.1 is a Cartesian grid solver utilizing a direct for-
wave induced current (up to 15% of the orbital velo- cing immersed boundary method. The research focus
city for the steepest wave). The results showed larger is on efficient strong coupling schemes for FSIs and
effects of the wave steepness on the wave mean and the extension to wave-body interaction problems in
2nd amplitude than on the 1st harmonic amplitude. The naval architecture and ocean engineering. In Yang and
wave steepness effect was also more prominent for Stern[38] an efficient strong coupling scheme for 6DOF
/ D = 4.7 than / D = 21.9 . The studies were also motion prediction was developed. The predictor-co-
conducted on the total wave field to evaluate the di- rrector loop in each time step includes the adjustments
ffracted wave pattern. The nonlinearities in the incide- of the structure displacements and velocities, but the
nt wave caused difficulties extracting the diffraction fluid flow solver was excluded. Then in Yang and
18
Stern[39] an efficient and robust immersed boundary te step with a non-inertial reference frame (NIRF) at-
setup procedure was developed for further accelera- tached to a solid body and no iterative loop is invo-
ting the strongly coupled simulations of FSIs. This lved. The improved efficiency and reduced algorithm
approach can be a viable choice for particulate flows complexity is evident. The next step of development
as shown in Yang and Stern[40]. Currently a non-itera- will be combination of this new scheme with an effi-
tive strong coupled scheme has been developed. cient two-phase flow solver for ultra-scale simulations
Figure 22 shows the flow charts of three different st- of 6DOF motions in naval architecture and ocean en-
rong coupling schemes in one time step for FSI pro- gineering. It should be pointed out that the develop-
blems. Compared with the scheme in Yang et al.[41] ment of wall models in immersed boundary methods
with a complete iterative loop including multiple is necessary if high Reynolds number flows are the
Poisson solves and the scheme in Yang and Stern[38] target application and a reasonable approximation of
with one Poisson solve but multiple local reconstru- the turbulent boundary layers is required.
ction steps, the present scheme utilizes an intermedia-
19
24(b) shows the bubble growth, merging, and adve- MC with dynamic metamodels was found the most
ction process in the simulation of the same foil in 3-D promising method overall. The high computational ef-
shortly after cavitation inception. This type of high fi- ficiency of dynamic metamodels, by auto-tuning and
delity simulation offers the opportunity for deeper in- adaptive sampling, makes the approach also recomme-
sight into the physics of cavitating flows. nded for stochastic optimization. Metamodel-based
UQ has been applied to stochastic design optimization
of DC in real ocean environment and operations, as
6. Uncertainty quantification shown in Diez et al.[54] and Tahara et al.[57]. Future ex-
Initial research focused on development and app- tensions include the application of metamodel-based
lication of deterministic verification and validation UQ and optimization to multi-disciplinary analysis
(V&V) methodologies and procedures for high-fideli- and optimization (MDA, MDO) of FSI problems.
ty CFD simulations. Initial studies for validation me-
thodologies[47] were subsequently extended to verifi-
cation procedures for deterministic uncertainties ste-
mming from iterative, grid and time-step converge-
nce[48,49]. V&V methodologies and examples were
presented at the AVT-147 Symposium on Computa-
tional Uncertainty[50]. Recently, the research focus
moved to stochastic uncertainty quantification (UQ)
methods as an essential part of stochastic design opti-
mization for real ocean environment and operations,
such as robust design optimization (RDO) and reliabi-
lity based design optimization (RBDO). UQ research
was undertaken within NATO AVT 191 “Application
of Sensitivity Analysis and UQ to Military Vehicle
Design”. The objective was the development and vali-
dation of efficient UQ methods for application to rea-
listic ship hydrodynamic problems. Non-intrusive UQ
methods were addressed with high-fidelity physics-
based CFD solvers. Evaluation metrics for efficient
UQ methods were developed, based on deterministic
and stochastic convergence criteria and validation ver-
sus numerical benchmark[51,52], and efficiency of over-
all UQ procedure by assessing the number of CFD si-
mulations required to achieved prescribed error thre-
sholds[53]. Numerical benchmarks were provided by
statistically convergent MC simulation with direct use
of CFD computations. UQ methods included metamo-
del-based Monte Carlo (MC) simulation, quadrature
formulas, and polynomial chaos methods. Applicatio-
ns covered unit studies and advanced industrial pro-
blems. Specifically, a unit problem for a NACA 0012
hydrofoil with variable Reynolds number was prese-
nted and assessed in Mousaviraad et al.[51]. The high-
speed Delft catamaran (DC) advancing in calm water
with variable Froude number and geometry was prese-
nted and studied in Diez et al.[54]. DC in stochastic ir-
regular and regular head waves (see Fig.25) with va-
riable speed and geometry was assessed in He et al.[55].
A combination of UQ problems for the DC was sele- Fig.25 Comparison of time history distributions from irregular
cted from Diez et al.[54] and He et al.[55] and used for wave (benchmark) and regular wave UQ for the Delft
further investigation in He et al.[56], focusing on the Catamaran in head waves, at sea state 6 and Fr = 0.5 .
polynomial chaos method, and Volpi et al.[53], focu- Empirical and Normal density functions are shown
sing on dynamic metamodels.
In conclusion, stochastic UQ methods were
found mature for application to realistic stochastic op- 7. Future research
timization problems. Based on the evaluation metrics, The oncoming exascale HPC era is to change our
21
Trajectories for ONR Tumblehome maneuvering in tion of wave adaptive modular vessels (WAM-V) and
calm water and waves[J]. Ocean Engineering, 2013, validation against sea trials[J]. Naval Engineers Jour-
72: 45-65. nal, 2014, Accepted for Publication, special edition: the
[16] SANADA, Y., ELSHIEKH, H., TODA Y. et al. Effects current fleet, the next class and the new prototypes.
of waves on course keeping and maneuvering for surfa- [31] MOUSAVIRAAD S. M., WANG Z. and STERN F.
ce combatant ONR tumblehome[C]. Proceedings of URANS studies of hydrodynamic performance and sla-
30th Symposium on Naval Hydrodynamics. Hobart, mming loads on high-speed planing hulls in calm water
Tasmania, Australia, 2014. and waves for deep and shallow conditions[C]. Procee-
[17] SADAT-HOSSEINI H., WU P. C. and CARRICA P. M. dings of 3rd International Conference on Ship Ma-
et al. CFD simulations of KVLCC2 maneuvering with neuvering in Shallow and Confined Water. Ghent,
different propeller modeling[C]. Proceedings of Belgium, 2013.
SIMMAN2014 Workshop. Copenhagen, Denmark, [32] MOUSAVIRAAD S. M., WANG Z. and STERN F.
2014. URANS studies of hydrodynamic performance and sla-
[18] SADAT-HOSSEINI H., STERN F. System based and mming loads on high-speed planing hulls in calm water
CFD simulations of 5415M maneuvering[C]. Procee- and waves for deep and shallow conditions[J]. Applied
dings of SIMMAN2014 Workshop. Copenhagen, Ocean Research, 2014, Submitted.
Denmark, 2014. [33] FU T. C., BRUCKER K. A. and MOUSAVIRAAD S.
[19] SADAT-HOSSEINI H., KIM D. H. and TAYLOR G. L. M. et al. A computational fluid dynamics study of the
et al. Vortical structures and instability analysis for hydrodynamics of high-speed planing craft in calm
Athena in turning maneuver with full-scale valida- water and waves[C]. Proceedings of 30th Symposium
tion[C]. Proceedings of 30th Symposium on Naval on Naval Hydrodynamics. Hobart, Tasmania,
Hydrodynamics. Hobart, Tasmania, Australia, 2014. Australia, 2014.
[20] BHUSHAN S., XING T. and STERN F. Vortical stru- [34] VOLPI S., SADAT-HOSSEINI H. and KIM D. H. et al.
ctures and instability analysis for Athena wetted tran- Validation high-fidelity CFD/FE FSI for full-scale high-
som flow with full-scale validation[J]. Journal of Flui- speed planing hull with composite bottom panels sla-
ds Engineering, 2012, 134(3): 031201. mming[C]. Proceedings of International Conference
[21] STERN F., TOXOPEUS S. Chapter 1–Experimental on Coupled Problems in Science and Engineering.
and computational studies of course keeping in waves San Servolo Island, Venice, Italy, 2015.
for naval surface combatant[R]. Technical Report 161, [35] YEON S., YANG J. and STERN F. Large eddy simula-
NATO AVT, 2013. tion of drag crisis in turbulent flow past a circular cyli-
[22] SADAT-HOSSEINI H., STERN F. and TOXOPEUS S. nder[C]. Proceedings of ITTC Workshop on Wave
CFD simulations of course keeping in irregular waves Run-Up and Vortex Shedding. Nantes, France, 2013.
for 5415M[J]. Ocean Engineering, 2014, in Prepara- [36] YOON S. H., KIM D. H. and SADAT-HOSSEINI H. et
tion. al. High-fidelity CFD simulation of wave run-up around
[23] SADAT-HOSSEINI H., KIM D. H. and LEE S. K. et al. vertical cylinders in monochromatic waves[C]. Procee-
CFD and EFD study of damaged ship stability in regu- dings of ITTC Workshop on Wave Run-Up and
lar waves[J]. Ocean Engineering, 2014, in Preparation. Vortex Shedding. Nantes, France, 2013.
[24] YOON H., GUI L. and BHUSHAN S. et al. Tomogra- [37] KOO B., YANG J. and YEON S. et al. Reynolds and
phic PIV measurements for a surface combatant at st- Froude number effect on the flow past an interface-pie-
raight ahead and static drift conditions[C]. Proceedings rcing circular cylinder[J]. International Journal of
of 30th Symposium on Naval Hydrodynamics. Naval Architecture and Ocean Engineering, 2014,
Hobart, Tasmania, Australia, 2014. 6(3): 529-561.
[25] ABDEL-MAKSOUD M., MÜLLER V. and XING T. et [38] YANG J., STERN F. A simple and efficient direct for-
al. Chapter 7Experimental and numerical investigatio- cing immersed boundary framework for fluid structure
ns on flow characteristics of the KVLCC2 at 30o drift interactions[J]. Journal of Computational Physics,
angle[R]. Technical Report 183, NATO AVT, 2015. 2012, 231(15): 5029-5061.
[26] FALCHI M., FELLI M. and GRIZZI S. et al. SPIV [39] YANG J., STERN F. Robust and efficient setup proce-
measurements around the Delft 372 catamaran in steady dure for complex triangulations in immersed boundary
drift[J]. Experiments in Fluids, 2014, 55(11): 1844. simulations[J]. Journal of Fluids Engineering, 2013,
[27] MOUSAVIRAAD S. M., SADAT-HOSSEINI S. H. and 135(10): 101107.
CARRICA P. M. et al. URANS studies and validation [40] YANG J., STERN F. A sharp interface direct forcing
of ship-ship interactions in calm water and waves for re- immersed boundary approach for fully resolved simula-
plenishment and overtaking conditions[J]. Journal of tions of particulate flows[J]. Journal of Fluids Engi-
Ocean Engineering, 2014, Submitted. neering, 2014, 136(4): 040904.
[28] SADAT-HOSSEINI H., WU P. C. and TODA Y. et al. [41] YANG J., PREIDIKMAN S. and BALARAS E. A stro-
URANS studies of ship-ship interactions in shallow- ngly coupled, embedded-boundary method for fluid st-
water[C]. Proceedings of 2nd International Confere- ructure interactions of elastically mounted rigid bo-
nce on Ship Manoeuvring in Shallow and Confined dies[J]. Journal of Fluids and Structures, 2008, 24(2):
Water. Trondheim, Norway, 2011. 167-182.
[29] MOUSAVIRAAD S. M., BHUSHAN S. and STERN F. [42] WANG Z., YANG J. and STERN F. High-fidelity si-
URANS studies of WAM-V multi-body dynamics in mulations of bubble, droplet, and spray formation in
calm water and waves[C]. Proceedings of 3rd Interna- breaking waves[R]. HPC Insights, 2012, Fall Issue: 5-
tional Conference on Ship Maneuvering in Shallow 7.
and Confined Water. Ghent, Belgium, 2013. [43] WANG Z., YANG J. and STERN F. High-fidelity si-
[30] CONGER M., MOUSAVIRAAD S. M. and STERN F. mulations of bubble, droplet, and spray formation in
et al. URANS CFD for two-body hydrodynamic simula- breaking waves[C]. Proceedings of 30th Symposium
23
on Naval Hydrodynamics. Hobart, Tasmania, [52] DIEZ M., CHEN X. and CAMPANA E. F. et al. Relia-
Australia, 2014. bility-based robust design optimization for ships in real
[44] GUI L., YOON H. and STERN F. Experimental and ocean environment[C]. Proceedings of 12th Interna-
theoretical investigation of instabilities for flow over a tional Conference on Fast Sea Transportation,
bump in a shallow water flume with steady downstream FAST2013. Amsterdam, The Netherlands, 2013.
wave train[R]. Technical Report 487, IIHR, University [53] VOLPI S., DIEZ M. and GAUL N. J. et al. Deve-
of Iowa, 2014. lopment and validation of a dynamic metamodel based
[45] GUI L., YOON H. and STERN F. Techniques for mea- on stochastic radial basis functions and uncertainty qua-
suring bulge-scar pattern of free surface deformation ntification[J]. Structural Multidisciplinary Optimiza-
and related velocity distribution in shallow water flow tion, 2014, DOI 10.1007/s00158-014-1128-5, in Press.
over a bump[J]. Experiments in Fluids, 2014, 55(4): [54] DIEZ M., HE W. and CAMPANA E. F. et al. Uncertai-
1721. nty quantification of Delft catamaran resistance, sinkage
[46] MICHAEL T., YANG J. and STERN F. Modeling cavi- and trim for variable Froude number and geometry
tation with a sharp interface[C]. Proceedings of 30th using metamodels, quadrature and Karhunen-Loève ex-
Symposium on Naval Hydrodynamics. Hobart, pansion[J]. Journal of Marine Science and Technolo-
Tasmania, Australia, 2014. gy, 2014, 19(2): 143-169.
[47] COLEMAN, H. W., STERN F. Uncertainties and CFD [55] HE W., DIEZ M. and ZOU Z. et al. URANS study of
code validation[J]. Journal of Fluids Engineering, Delft catamaran total/added resistance, motions and sla-
1997, 119(4): 795-803. mming loads in head sea including irregular wave and
[48] STERN F., WILSON R. and SHAO J. Quantitative uncertainty quantification for variable regular wave and
V&V of CFD simulations and certification of CFD geometry[J]. Ocean Engineering, 2013, 74: 189-217.
codes[J]. International Journal for Numerical Me- [56] HE Wei, DIEZ Matteo and CAMPANA Emilio
thods in Fluids, 2006, 50(11): 1335-1355. Fortunato et al. A polynomial chaos method in CFD-
[49] XING T., STERN F. Factors of safety for Richardson based uncertainty quantification study for ship hydrody-
extrapolation[J]. Journal of Fluids Enginerring, 2010, namic performance[J]. Journal of Hydrodynamics,
132(6): 061403. 2013, 25(5): 189-217.
[50] STERN F. Quantitative V&V of CFD solutions and cer- [57] TAHARA Y., DIEZ M. and VOLPI S. et al. CFD-based
tification of CFD codes with examples for ship hydro- multiobjective stochastic optimization of a water-jet
dynamics[C]. Proceedings of Symposium on Compu- propelled high speed ship[C]. Proceedings of 30th
tational Uncertainty, AVT-147. Athens, Greece, 2007. Symposium on Naval Hydrodynamics. Hobart,
[51] MOUSAVIRAAD S. M., HE W. and DIEZ M. et al. Tasmania, Australia, 2014.
Framework for convergence and validation of stochastic
uncertainty quantification and relationship to determini-
stic verification and validation[C]. International Jour-
nal for Uncertainty Quantification, 2013, 3(5): 371-
395.