Beruflich Dokumente
Kultur Dokumente
T-Craft Calm Water Resistance and Motions, and Seakeeping in Regular Waves
Shanti Bhushan1, Frederick Stern2 and Lawrence J. Doctors3
1
Center for Advanced Vehicular Systems, Mississippi State University, Starkville, Mississippi, USA
2
IIHR-HydroScience and Engineering, The University of Iowa, Iowa City, Iowa, USA
3
The University of New South Wales, Sydney, Australia
two-equation blended k- /k- turbulence. The location of Substituting Eq. (2) in Eq. (3) yields,
the free surface is given by a „zero‟ value of the level-set (𝑃𝑐0 +𝑃𝑎 )
function, which is advected as a material surface. Negligible ∆𝑃𝑐 𝑡 = 𝑚 ∆𝑡 (4)
𝜌 𝑎 𝑉𝑐0
shear stress is assumed in the air phase which provides the Equation (4) is combined with Eq. (1) to obtain the variable
pressure-jump condition at the free-surface. The governing Pc model with mass flux:
equations are discretized using finite-difference schemes on 𝛾
𝑚 0
body-fitted curvilinear grids. Time marching is performed 𝑃𝑐 𝑡 = (𝑃𝑐0 + 𝑃𝑎 ) 1 + ∆𝑡 − 𝑃𝑎 (5)
𝜌 𝑎 𝑜 𝐴𝑐 𝑐 (𝑡)
using a second-order backward-difference scheme. The
convection and diffusion terms are discretized using a 2.3 Leakage and Fan Flows
second-order scheme. The pressure Poisson equation is The continuity equation for the air flow in the cushion
solved using the PETSc toolkit to satisfy continuity using a control volumes CV as shown in Fig. 1 at time t is:
projection algorithm. 6DoF motions are solved implicitly 𝑑
using multi-block grids with overset grid interpolation using 𝑑𝑉𝑐
𝑑𝑡 𝐶𝑉 𝑐
+ 𝐶𝑆
𝜌𝑎 𝑽 ∙ 𝒏 𝑑𝐴 = 0 (6)
SUGGAR. MPI based domain decomposition is used for 𝑚
multiple processor simulation. The first term on left is the rate of change of mass in the CV
(𝑚). The second term is mass flux through the control
surface CS, which includes leakage from the air gaps
Hinged seals. The bow and stern seals are discretized 4.0 SIMULATION CONDITIONS, DOMAIN,
separately and a small gap is left between the seals and the BOUNDARY CONDITIONS AND GRIDS
catamaran. The seals pitch and heave along with the Table 2 summarizes the simulations performed without Pc,
catamarans and rotate () along the Y-axis about a fixed and prescribed and variable without mass flux Pc models
pivot point on the deck. The rotation is governed by the which includes: (1) resistance and motions using rigid,
moment created by the pressure and frictional force hinged and imaginary without resistance seal models at Re =
distribution (M) on the exterior wetted seal surface and the 2.4 to 7.3×106, Fr = 0.2 to 0.6; (2) impulsive heave, pitch
restoring moment due to Pc acting on the interior seal face: and roll response using imaginary seals without resistance at
𝑑2𝜙 1
Fr = 0; seakeeping response in regular head waves using
𝑘 = 𝑀 − 𝑃𝑐 𝑠 2 𝐵𝑐 (10) imaginary seals without resistance at (3) Fr = 0 for fe = 0.35
𝑑𝑡 2 2
where, k is an arbitrarily chosen moment of inertia and hs is to 0.96, and (4) Fr = 0.6 for fe = 0.8 to 2.0.
the seal length. This is an approximate model and ignores Table 1: T-Craft particulars and natural Frequencies.
important physics, i.e., seal should rotate about water line Parameter (Unit) Model-scale Full-scale Non-dimensional
LPP (m) 2.52 76.35 1 LPP
and moment if inertia should be based on the wetted length. B (m) 0.7366 22.25 0.29 LPP
Lc (m) 2.23 67.14 0.885 LPP
Imaginary seals without resistance. The seals are not Bc (m) 0.546 16.50 0.216 LPP
discretized and their interaction with hydrodynamics are T (m) 0.044 1.33 0.0175 LPP
neglected. T without Pc (m) 0.13 3.93 0.051 LPP
h0 (m) 0.088 2.66 0.0347 LPP
Imaginary seals with resistance. The interaction of the VCG below deck (m) 0.00635 0.19 0.00252 LPP
imaginary bow seal with hydrodynamics is modeled as a 2D Catamaran Aw (m2) w/o cushion 0.3181 290.36 0.0501 LPP2
Catamaran Aw (m2) w cushion 0.4752 433.7 0.0744 LPP2
planing surface (Doctors and McKesson, 2006). The seal Volume of water displacement (m3) 0.055163 1519.5 0.00345 LPP3
resistance is computed from the planing-surface pressure Buoyancy (N) 540.0675 14.875×106
distribution and length. The bow seal interacts with the CB w/o Pc (with Pc estimated) 0.229 (0.256)
Pc0,fwd (N/m2) 220.61 6664.63 Pc/gL= 0.0101
hydrodynamics via a free-surface pressure distribution (PBS) Pc0,aft (N/m2) 296.442 8955.51 Pc/gL= 0.0136
obtained from the linear theory. A similar model is not Weight (N) 856.6 2.354×10 6
available for the stern seal, which is expected to mostly fn,B (Hz) without Pc 0.85 0.154 1/L Scaling
fn,P (Hz) (over rigid surface) 39.04 1.34 1/L Scaling
leak. Thus is modeled as imaginary seal without resistance. fn,P,fs (Hz) (over free-surface) 2.15 0.354 1/L Scaling
fn (Hz) 1.303 0.223 1/L Scaling
5.0 CALM WATER RESISTANCE AND MOTIONS compared to the without Pc case. The wave elevation inside
As shown in Fig. 3, imaginary seal simulations under the cushion shows peaks and troughs as observed for ACV
predict CT by 35-50%D compared to the EFD data (D). simulations (Bhushan et al., 2010). Breaking waves are
Variable Pc simulations do not show significant influence on predicted towards the leading edge of the cushion for Fr =
the CT predictions. Simulations without Pc predicts 50% 0.2, which moves inside the cushion for Fr = 0.3 and 0.4.
lower CT that those with prescribed Pc, which suggests that
the wave resistance due to Pc accounts for almost 50% of
the drag. Rigid seal simulations over predict CT by 37%D.
Higher CT predictions are due to bow seal plowing
accompanied by breaking bow waves as shown in Fig. 4(a).
The large bow waves are due to the presence of a stagnation
point on the bow seal which causes the water to ride up.
Hinged-seal simulations predict CT within 8.5%D of the
EFD. The improved predictions are due to the decrease in
bow seal plowing as shown in Fig. 4(b). The bow seal
rotation is observed to be 5.2° and 18.7° for Fr = 0.4 and
0.6, respectively, whereas for both the cases the rotation of
the stern seal is < 0.1° as they are mostly in air.
As shown in Fig. 5, imaginary seal simulations without
Pc show diverging wave patterns outboard and inboard of
the catamaran. However, both prescribed and variable Pc
simulations show diverging waves only outboard of the
catamaran which are attached to the catamaran. Further, the Fig. 3. T-Craft resistance, sinkage and trim predictions
diverging wave peaks and troughs are significantly larger in calm-water using V4 are compared with EFD.
predictions are obtained using hinged seal, but trim is over This gives fn,P,r = 39.05 Hz for model-scale and 1.34Hz for
predicted by 20% for Fr = 0.6. Overall, motions do not full-scale T-Craft as shown in Table 1.
show significant influence on the Pc model, and the Pc
Natural frequency due to Pc over a free-surface. Assuming a
variation are <4.5% as motions are small.
case similar to above, but the free-surface depresses due to
Pc. The restoring force is then represented as a series spring
system with spring constants due to free-surface (kfs) and kP
(Milewski et al. 2007) as below:
−1
1 1
𝑘𝑃,𝑓𝑠 = + 𝛾 (𝑃 𝑐0 +𝑃 𝑎 )𝐴 𝑐 (13)
𝜌 𝑤 𝑔𝐴𝑐
∗
𝐹𝑟𝑒𝑒 −𝑠𝑢𝑟𝑓𝑎𝑐𝑒
𝐶𝑢𝑠 𝑖𝑜𝑛 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒
1 𝑘 𝑃,𝑓𝑠
𝑓𝑛,𝑃,𝑓𝑠 = (14)
2𝜋 𝑀+𝑀𝑎
(a) (b)
which gives fn,P,fs = 2.15 Hz for model-scale and 0.0354 Hz
Fig. 4. Wave elevation contours for T-Craft with (a) rigid
for full-scale T-Craft as shown in Table 1.
seals and (b) hinged seal with prescribed Pc at Fr = 0.6.
Natural Frequency due to both buoyancy and Pc over a free-
surface. In this case the restoring force is represented as a
series and parallel spring system with
1 𝑘 𝐵 +(1−𝐴 𝑤 /𝐴𝑐 )𝑘 𝑃 ,𝑓𝑠
𝑓𝑛 = (15)
2𝜋 𝑀+𝑀𝑎
(c)
(d)
Fig. 6. T-Craft impulsive heave and pitch response in calm-
Fig. 7. (a) Pitch and (b) heave transfer functions, and (c)
water at Fr = 0 using V4 are compared with EFD.
pitch and (d) heave phases are compared with EFD data.
7.0 SEAKEEPING V4 predictions show that both the heave and pitch
7.1 Fr = 0 transfer functions grow steadily with the decrease in fe and
EFD data in Fig. 7 shows pitch and heave transfer functions approaches unity for lower fe (larger ). The pitch response
around 0.3 for higher fe and the response overall increases is consistently lower than the EFD, but the heave response
with the decrease in fe. However, large oscillations are compares fairly well with EFD. The variable Pc model
observed for lower fe (larger ), i.e., sharp decline at fe = shows 2 to 3% higher response than the prescribed Pc
0.525Hz (/L=2.2) followed by large transfer functions >1.2 model, which is an opposite trend to the impulsive cases.
at fe = 0.437Hz (/L=3.2). The pitch lags the head wave by Overall, the pitch and heave phases agree very well with the
90º and heave is in phase with the wave for lower fe (large EFD, but the heave phases are consistently 20º lower.
), which is in good agreement with Irvine et al. (2008). As EFD Pc data in Fig. 8 shows good correlation with the
fe increases ( decreases) the pitch phase lag increases to - heave motions for lower fe (larger ), i.e., higher pressure
180º, switches sign to +180º lead and the lead decreases for both aft and forward cushions at heave minima and vice
thereafter. A similar behaviour is observed for the heave versa. For larger fe (smaller ), Pc correlates better with the
phase, but the switch from lag to lead occurs at higher fe. pitch, i.e., minimum forward and maximum aft Pc at pitch
maxima and vice versa. This is expected as heave motions