OTC-27092-MS

Drillship Moonpool Design to Reduce Added Resistance for Fuel Saving

Peifeng Ma, Keppel Offshore & Marine Technology Centre; Xiuqing Xing, Deguang Yan, Hung-Pin Chien,
Sing-Kwan Lee, and Hai Gu, ABS; Haihua Xu, Ankit Choudhary, Anis Hussain, and Aziz Amirali Merchant,
Keppel Offshore & Marine Technology Centre

Copyright 2016, Offshore Technology Conference

This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, USA, 2–5 May 2016.

This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Offshore Technology Conference, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the
written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.

Abstract
During transit, a drillship moonpool creates added resistance that can be more than 50% of the total
resistance and hence significantly increase the fuel consumption. This paper reports a real case study for
reducing moonpool added resistance by model tests and Computational Fluid Dynamics (CFD) simula-
tions. Model test data show that, unlike conventional ships that experience nearly constant resistance, a
drillship resistance presents largely fluctuating behaviour. In CFD validation, the physics based CFD best
practices are applied for resistance predictions. The largely fluctuating resistance is well captured in
simulations. It is found from flow visualizations of the large amount of CFD simulation results that the
high moonpool induced added resistance is mainly attributed to the vortices shed from the moonpool front
wall, which enter into the moonpool and impinge on the rear wall. The CFD predicted mean resistance
are in good agreement with the model test data, within 3% difference for a wide range of speeds. The well
validated CFD tool is applied to study the effects of moonpool dimensions on the added resistance. The
results of parametric study reveal a design principle that smaller moonpool dimension results in smaller
added resistance. Through comprehensive CFD parametric study, a universal design principle and CFD
best practice are established for industry applications.

Introduction
Water motion in a drillship moonpool during transit is a complex phenomenon. It could cause a drastic
increase in drillship resistance during transit due to the water oscillation in the moonpool and conse-
quently increase the fuel cost. Moonpool oscillations are initiated by vortices that start at the upstream
submerged moonpool edge. The vortices are related to the separation of a shear layer so viscous effects
dominate the initiation of the oscillation as well as the flow pattern in the moonpool. The oscillation has
a dominant frequency and the amplitude increases with forward speed. The resonance oscillation mode
can be piston or sloshing (Fukuda, 1977; Molin, 2001). Recent studies show that both types of oscillation
can increase ship resistance to the same order of magnitude. In general, the sloshing mode dominates in
longer moonpools, while piston mode oscillations are dominant in shorter moonpools. Depending on the
length to draft ratio, vertical moonpool water motion can be excessive at forward speed. Due to the phase
locking phenomenon the violent oscillation can sustain over a large range of speed.
2 OTC-27092-MS

The global dimensions of the moonpool are an important design parameter from an operational
perspective, as well as from a hydrodynamic viewpoint. The length and width of a moonpool are derived
from operational and structural requirements. The moonpool draft equals the draft of the vessel which
follows from operational and stability requirements. As summarized by Veer et al. (2008), the typical
dimensions of existing drillships are about 7-25m in length and 8-15m in breadth. Most of the drillships
were designed to have a designed transit speed in the range of 10 to 12knots, without excessive moonpool
resonance for those moonpool dimensions. Theoretically, the moonpool geometry determines the piston
and sloshing resonant frequency, i.e. the dominant frequencies of the water motion in a moonpool, while
the transit speed determines the oscillation amplitude. The natural frequency of the piston mode of a
moonpool can be predicted based on formulation reported by Fukuda (1977) and Molin (2001), while
natural frequency of the sloshing mode can be predicted based on formulation reported by Newman
(1977), Fukuda (1977) and Molin (2001). According to the formulations which are mostly proposed by
assuming a rectangular moonpool shape, the natural frequency is dependent on the moonpool geometry
ratio such as the length breadth ratio and the draft breadth ratio. The estimated natural period from these
formulas is about 6-8 sec and 4-6 sec for piston mode and sloshing mode, respectively.
Although moonpool natural frequency can be estimated quite reliably and there are numbers of
drillship have been built based on industry experience, how to accurately predict the moonpool added
resistance and water motion resonance in the moonpool still remains challenging for moonpool design.
The resistance prediction in an early design phase relies mostly on scaled model tests in combination with
experiences based on previous designs. This approach provides quite accurate predictions for conventional
ship hulls, but it becomes unreliable for a drillship where a moonpool presents. Furthermore, the variation
in hull and moonpool designs is large and the available test data is often too limited for an accurate trend
prediction. In order to predict a drillship resistance in the early design stage, a reliable numerical method
can be a feasible alternative. Low cost, high efficient numerical prediction methods for resistance and
water motion resonance prediction are of great interest to moonpool designers.
To meet the purpose, strong viscous effects in the moonpool must be well simulated. CFD is one of
such tools which is able to capture all viscous effects in turbulence flows. CFD has been widely used in
predicting resistance and propulsion of conventional ships and proved to be capable in predicting reliable
and accurate results (e.g. Enger et al. 2010; Larsson et al. 2013). It has also been used to study the
moonpool effects on a drillship (e.g. Son et al. 2008, Hammargren and Tornblom, 2012; Wang and Quah,
2013).
One objective of the present study is, through model test data analysis and CFD simulations, to
investigate the impact of moonpool on drillship resistance and the mechanism of moonpool induced
resistance, and to analyse the effects of moonpool dimensions on the drillship resistance. The other
objective is to establish a CFD best practice based on the validation study and a guideline for drillship
moonpool design. First, a drillship model test and the measured data will be introduced and analysed.
Subsequently, CFD simulation setup and results will be presented, followed by further discussions on the
findings from the CFD simulations. Finally the conclusion will be drawn.
Analysis of drillship model test data
In this study, the data from a drillship model test is used for investigating the mechanism of moonpool
added resistance and CFD validation as well. All of tests were conducted in a deep water towing tank. In
the resistance testing the drillship was towed by the carriage and the total longitudinal force acting on the
model was measured at various speeds. Turbulence stimulators were present at the bow of the ship model
to stimulate transition from a laminar into a turbulent boundary layer. During the resistance tests, the
behaviour of the water column inside the moonpool has been measured with two wave height transducers,
which are located at the front end of the moonpool on the centreline and in the middle point of the
moonpool at portside.
OTC-27092-MS 3

An example of the measured time history and the corresponding Fast Fourier Transform (FFT)
analysed frequency of the ship resistance and the water elevation at the centre of moonpool are shown in
Figure 1 and 2, respectively. It is seen in Figure 1a that the resistance fluctuates around its mean value
substantially, ranging from 0.35 to 2.0 time the mean resistance. The fluctuation is mainly caused by the
presence of moonpool. FFT analysis results of resistance as shown in Figure 1b suggests that the dominant
frequency is about 0.66 Hz for the model scale drillship, corresponding to a natural period of 1.52 sec.
This is in between the sloshing mode period of 1.29 sec and the piston mode period of 1.58 sec predicted
by the empirical formula proposed by Molin (2001). Similar to the resistance, the water level as seen in
Figure 2a shows that oscillating water motion occurs inside the moonpool although the drillship is moving
at a constant forward speed. FFT analysed frequency in Figure 2b shows that the dominant frequency of
the water levels is same as the dominant resistance frequency.

Figure 1—Resistance of drillship at a design speed (a) Measured time series and (b) FFT analyzed frequency domain
4 OTC-27092-MS

Figure 2—Water level at the centre of the drillship moonpool (a) Measured time series and (b) FFT analyzed frequency domain

Model test data at various speeds reveals that, unlike the mean resistance that increases monochro-
matically with ship transit speed, the amplitude of the dominant harmonic component increases at lower
speed up to 14knot and then decreases at higher speeds. At higher speeds, the resistance amplitude is more
scattered on the frequency domain plots which is probably due to the more violent water motion inside
the moonpool caused by the surface wave breaking. The behaviour of dominant frequency at different
transit speed regime is also quite different. The dominant frequency at lower speeds are quite close for
both resistance and water levels. For resistance, it reduces at moderate and relatively high speeds, whereas
that for water levels reduces at moderate speed and increases again at a relatively high speed.
CFD validation and parametric study
CFD model setup
The commercial CFD program STAR-CCM⫹ (CD-adapco, 2014) is used in this study to create all grids,
perform all computations and post-processing. The program uses control volumes of arbitrary polyhedral
shape based on a finite-volume type algorithm. A segregated iterative solution method based on SIMPLE-
algorithm is used to solve conservation equations in integral form for mass and momentum, together with
an equation for volume fraction of liquid. A variety of turbulence models are available in the program to
compute turbulence quantities. Details on discretization and solution methods can be found in literature
(e.g. Ferziger and Peric, 2003; Weiss et al, 1999). The 2nd order Euler implicit scheme is employed in the
present study to perform time integration to compute the unsteady flow. The free-surface effects are
modelled using Volume-of-Fluid (VOF) approach. In this approach, the solution domain is assumed to be
filled by a single effective fluid whose properties vary locally according to volume fraction of liquid, with
which the deformation of the interface between liquid and other phases of fluids can be tracked.
OTC-27092-MS 5

Computational domain and meshing

In the present CFD computations, only half of ship geometry is considered due to the symmetry. To avoid
boundary reflections, a large domain is defined as shown in Figure 3 which has a length of 5.5Lpp, width
of 2.1Lpp and a height of 2.8Lpp with Lpp the ship length. The ship is located in an upstream position
with a distance of about 1.5Lpp between the ship bow and the upstream boundary of the computational
domain.

Figure 3—Computation domain and overall volume grids

Trimmed hexahedral grids with local refinements and prism layers along walls are used in the
simulations. The grid generation is carried out by specifying a base mesh size, relative to which all spacing
(e.g. prism layer thickness, mesh size in various regions etc.) are defined. In order to avoid using fine grid
everywhere in the domain which will create unnecessarily excessive computing effort, local volumes of
different shapes are specified and particular cell size is defined based on local flow properties, resulting
in mesh structure shown in Figure 4. As shown in Figure 4, local refinement is applied to the free surface
layer such that there are about 20 mesh layers in the range of one wave height. To precisely capture ship
waves, fine grids are used in the Kevin wave zone around the ship hull. In other regions where flow is
more complicated such as around the ship bow and stern area, fine grids are also defined. Moonpool is
the focus of the present study and model tests reveal that the water motion inside moonpool can be very
violent. To simulate such flows, very fine grids in the moonpoon region are created, especially in the free
surface layer inside the moonpool, as shown in Figure 4.
6 OTC-27092-MS

Figure 4 —Mesh refinement at selected region

Boundary layer flow prediction is very important in CFD simulations as it affects the prediction of wall
friction and pressure. In the present study, the high (or all) wall y⫹ treatment is employed. Wall y⫹ value
is an indicator of the distance between the hull surface and the first layer of prism layer. For an all wall
function boundary layer treatment, 30 ⱕ y⫹ ⱕ 200 is a reference range indicating a reasonable prism
layer setup for the turbulence boundary layer flow field. In the present study, the boundary layer mesh is
generated such that 30 ⱕ y⫹ ⱕ 100 can be assured in all of the simulations.
In balancing the accuracy and efficiency, the eventually used grid system in this study has a total
volume mesh number of about 4 million.

CFD validation results

For CFD validation, it is important to include a wide range of speeds as this can test the adaptability of
the CFD model set-up for a wide range change of physics such as wave length, wave height, vortex
shedding frequency and strength. Six forward speeds from 11 to 16 knot have been considered in the
present study in accordance with the model testing.
Based on the CFD set-up described previously, the achieved differences between CFD simulated
resistances and model test measurements are within 3% for all the six speed cases. Figure 5 summarizes
the final validation results. It can be concluded that the CFD set-ups implemented can capture the drillship
flow physics well, and consequently, yield good agreements with the model test results.
OTC-27092-MS 7

Figure 5—Validation results of ship resistance

In order to gain insights from the simulation results, the details of validation results are presented with
13 knot as an example. To evaluate the moonpool induced added resistance, simulations for the drillship
without moonpool are also carried out and presented in the validation study.
Ship waves Ship waves at a forward speed of 13knot are shown in Figure 6a. The transverse and
divergent waves were captured very well. For the drillship without a moonpool case, the simulated waves
are also plotted in Figure 6b. As seen in the Figure 6, the moonpool has insignificant effect on the ship
bow waves. However, due to the existence of the moonpool, the stern wave height of the drillship
increases, which consequently causes the better pressure recovery at the stern part compared to that of the
drillship without a moonpool case.

Figure 6 —CFD simulated drillship waves (a) with a moonpool and (b) without a moonpool

Dynamic pressure Dynamic pressure distributions on the drillship hull surface are shown in Figure 7.
Due to the vortex shedding from the moonpool, the dynamic pressure at the stern of the drillship is in
overall higher than that at the stern of the hull without a moonpool and appears better recovery. It can also
8 OTC-27092-MS

be seen that the pressure on the rear wall of the moonpool is much higher than that on the front wall. This
indicates the moonpool will generates additional negative pressure force in the moving direction, i.e.
producing added resistance to the drillship.

Figure 7—Dynamic pressure on the drillship hull surface

Vorticity Vorticity contours for drillship case are plotted in Figure 8. Compared with the drillship
without a moonpool case, the drillship case shows more unsteady patterns due to the unsteady vortex
shedding from the front wall of the moonpool. It is also noted that in front of the moonpool, the flow
maintains steady and basically is not affected by the shedding vortex in the moonpool. The unsteady
vorticity fields as shown in Figure 9 reveals that the unsteady flow inside the moonpool is induced by the
vortices shed from the bottom front edge. This unsteady flow create time varying pressure forces on the
moonpool walls and therefore the added resistance to the ship hull. At the bottom edge of the moonpool
rear wall, vortices are also generated and induces vortex train under the rear part of the drillship hull as
seen in Figure 8.

Figure 8 —Vorticity field at the central plane around the drillship hull
OTC-27092-MS 9

Figure 9 —Vorticity field at the central plane inside the moonpool at different time instants

Drillship resistance Time history of the simulated drillship total resistance is compared with the model
test results in Figure 10. It can be seen that the simulated drillship total resistance at the stable stage
matches the model test time history quite well. With FFT analysis, the mean and dominant harmonic
components of the total resistance can be obtained and compared with model test data. The results indicate
that the predicted mean resistance is in good agreement with the model test data, giving an over-prediction
of 2.6%.

Figure 10 —Resistance time history of drillship

CFD predicted drillship resistance can be readily broken down into various components, e.g. the
friction and pressure resistance on the moonpool and the rest of hull. It is found that the mean resistance
induced by the moonpool is 59% the drillship total resistance, larger than the resistance induced by the
hull of 41%, as shown in Figure 12a. In addition, the friction component dominates the resistance on the
hull (90% of the hull resistance) whereas the pressure drag dominates the resistance in the moonpool
(nearly 100%). In fact, the mean friction force on the moonpool walls is very small and in the opposite
direction of the pressure-induced resistance. This friction force is caused by the complex flow inside the
moonpool where the flow near the side wall has a mean flow in the forward direction.
10 OTC-27092-MS

Figure 11—Resistnace time history of drillship without a moonpool

Figure 12—Resistance component breakdown of (a) drillship with a moonpool, and (b) drillship without a moonpool

In order to estimate the drillship added resistance, resistance of the drillship without moonpool case has
been simulated with the CFD set-ups same as the drillship case. The total resistance and its pressure and
friction components are plotted in Figure 11. It can be observed that the resistance without moonpool
varies very little in time after the ramp up stage. This suggests that the large fluctuations in the drillship
with moonpool is mostly caused by the moonpool. The comparisons of resistances on the two ship hulls
show that the presence of the moonpool increases the total mean resistance remarkably by about 110%.
The results also show that the friction dominates the resistance of the drillship without moonpool case,
which is 81% of the total resistance, as shown in Figure 12b. On the contrary, the pressure drag dominates
the resistance of the drillship, which is 64% of the total resistance. CFD simulations at other transit speeds
show similar results, i.e. the added resistance is about 100% of the resistance without a moonpool.
To further investigate the added resistance from the moonpool, the time histories of the pressure force
on the front and the rear walls are analyzed. As seen in Figure 13, the average pressure force on the
moonpool front wall is a negative value, which means this force is a thrust to the drillship. For the rear
wall average pressure force (which has a positive value), its magnitude is larger than the absolute values
of the average front wall pressure force. This larger rear wall resistance after subtracting the thrust from
the front wall eventually contributes an added resistance to the drillship.
OTC-27092-MS 11

Figure 13—Time History of pressure force on moonpool (a) front wall and (b) real wall

Water elevation inside moonpool Time history of the CFD simulated moonpool water elevations at a
centerline front point (CL) and the portside middle point (PS) are compared with model test result in
Figure 14a and 14b, respectively. From the time history shown in Figure 14, it can be seen that the
amplitude of simulated water elevation is larger than that of the model test at CL and quite close at PS.
FFT analysis results show that the amplitude of dominant harmonic component is quite close (about 1%)
to the model test data at CL whereas the substantial difference (more than 50%) is obtained at PS. This
might be caused by the violent turbulence flow and the free surface waves inside the moonpool, since the
flow visualization shows that the wave breaking occurs at the middle and rear part of moonpool. Unlike
the dominant harmonic amplitudes, the dominant frequency was captured by the CFD simulations very
well, with a difference of about 7% from the model test data.
12 OTC-27092-MS

Figure 14 —Time history of water elevation inside moonpool at (a) centerline front point (CL) and (b) portside middle point (PS)

CFD parametric study

Parametric study is carried out in order to understand the moonpool dimension effect on the drillship
resistance, and help find out the optimal moonpool design. For the first step, it is decided to evaluate the
drillship resistance with changing the moonpool length and width based on the base design. CFD
simulations are performed for the moonpool horizontal dimensions shown in Figure 15 by using the
validated model and the developed best practices. For each dimension, the resistance is evaluated at same
4 speeds as those in the model test, i.e. from 11-16 knots.
OTC-27092-MS 13

Figure 15—Sketch of moonpool dimensions for parametric study

The parametric study results are summarized in Figure 16. It is observed that resistance can be reduced
with narrower or shorter moonpool. It can also be seen in Figure 16 that the variation at lower speed is
a bit different from that at relatively high speed. For example, the resistance at 12knot for narrower
moonpool is smaller than that for narrower and shorter moonpool. However, in overall, the optimal
moonpool dimension is based on the narrowest breath and shortest length

Figure 16 —Drillship resistance vs speed with various moonpool dimensions

The parametric study indicates that the smaller moonpool dimension, the lower the moonpool induced
resistance is. In order to understand the physics behind this phenomena, dynamic characteristics of the
drillship total resistance are investigated by FFT analysis. The results show obviously that shorter length
and/or narrower width of moonpool result in higher shedding frequency, as shown in Figure 17. In other
14 OTC-27092-MS

word, the smaller the moonpool dimension, the higher the vortex shedding frequency. With higher
frequency, the moonpool system can accumulate less energy during a shorter period and generate smaller
added resistance. By optimizing the moonpool dimension, 15% resistance reduction can be achieved at
design speed.

Figure 17—Resistance oscillation frequency with various moonpool dimensions

Discussion

Mechanism of water motion inside moonpool during a transit stage

Both model tests and CFD simulations show that strong oscillatory water motions can be agitated in the
moonpool of a drillship on transit at a constant forward speed. Details of flow motion can be visualized
and analyzed from CFD simulations and consequently the mechanism of the oscillatory motion can be
investigated. In principle, the vortices periodically shed from the bottom of moonpool front wall is the
origin of the oscillatory flow in the moonpool. Basically there are three frequencies in the system, i.e. the
vortex shedding frequency f1, the dominant water elevation frequency in the moonpool f2 and the
dominant resistance oscillation frequency f3, as illustrated in Figure 18. When a stable flow state is
established, the frequencies of the three phases should be identical or very close. So this flow phenomenon
can be called phase locking. This was confirmed by both the model test data and the CFD simulated results
in the present study. For example, the identical value of 0.66Hz was obtained for all of the three
frequencies at 13 knot.
OTC-27092-MS 15

Figure 18 —Vortex inside the moonpool and indication of various frequencies

According to the empirical formula (Fukuda, 1977) that estimates the moonpool natural frequency u>n
based on moonpool length L, breadth B and draft D

and also the present CFD simulations, it can be concluded that the shorter length L and/or narrower width
B of moonpool result in higher frequency. At higher frequency, the moonpool system will accumulate less
energy during a shorter period and consequently generate smaller added resistance. For longer moonpool,
the vortex shed from upstream can always enter into the moonpool impinging on the rear wall to create
strong pressure force on the wall. More energy can be accumulated inside a long/wider moonpool,
resulting larger resistance and more violent free surface flow.
CFD model effects on the predicted results
In developing the CFD practice applicable to the drillship problem, a variety of sensitivity tests for
computational domain, grid resolution, far field damping zone, prism layer grid, trim/sinkage motion and
turbulence model are performed based on the judgment and consideration of the flow physics. Some
detailed analysis related to the effects of trim/sinkage motion and turbulence model on the drillship
resistance are briefly discussed in this section. As a consequence, a physics based CFD practice is
developed and summarized.
Although the sinkage and trim motion are set free during the model test, they are not recorded in the
resistance tests but in self-propulsion tests. In the CFD validation study, one of the tasks is to investigate
the sensitivity of the trim and sink motion effect on the resistance to determine if the motion could be
excluded so that the CFD computational cost can be reduced.
In order to investigate the effect of the trim and sinkage, the self-propulsion model test data is analyzed
as no such data for resistance tests. The results suggest that trim and sinkage at the testing speed is
insignificant. These values should be even smaller in resistance tests as in propulsion tests the flow can
be accelerated by the thruster and the hull pressure on the stern will actually be decreased compared to
the bare hull resistance tests and therefore cause larger trim and sink motion. CFD simulations at one
design speed with the trim and sink motion on and off are compared, which show very similar simulation
results. Therefore, it can be concluded that the trim and sinkage effect on the drillship total resistance is
insignificant and they can be neglected in the CFD resistance simulations.
As there are strong turbulence and vortex shedding in the drillship flow field, CFD turbulence model
can be critical to the simulation results. In the present study, most simulations were performed by a
16 OTC-27092-MS

Reynolds Averaged Navier-Stokes (RANS) solver with k-␻ turbulence model due to its good performance
in capturing the strong flow separation (Wilcox, 2006). In the CFD validation study, it appears that the
simulated water elevation based on k-␻ turbulence model always shows some discrepancies from the
model test results at a low speed, even applying very fine mesh in the moonpool region. The reason might
be related to the weakness of RANS model at simulating the complex vortex structure in strong wake flow
such as the case in the present study. To improve the CFD simulation results of the water elevation, a more
advanced model DES (Detached Eddy Simulation) is applied. To evaluate the results, the simulated water
elevations at CL obtained from RANS with k-ro turbulence model and DES model are compared with the
model test results, as shown in Figure 19. It can be seen that the DES computation can improve the
predicted water elevation a lot, showing better agreement with the model test results than that of RANS
with k-␻ model simulation. However, in order to perform the DES computation, much finer mesh size and
smaller time step are required. This eventually causes several times computing time compared to the
RANS model. This make the DES based simulation impractical for the present study where added
resistance is the primary concern.

Figure 19 —Turbulence model effect on predicted water elevation (a) k-␻ model, and (b) DES model
OTC-27092-MS 17

Physics based CFD practices for drillship resistance prediction

The CFD practices for drillship resistance simulation in calm water are developed based on ship and
moonpool flow field physics. Based on the practices developed, the CFD simulations for drillship
resistances can be controlled within 3% difference against model test data for a wide range of speeds. The
practices consists of the choice of domain size, mesh resolution, time step, turbulence model and
numerical scheme.
The computational domain should be large enough to avoid the influence of wave reflection from
upstream, downstream and side of the domain on the flow field around hull and ship Kelvin waves. For
most CFD simulations, finer mesh would lead to better results, which however is very costly in term of
computing time if using very fine mesh in the entire computational domain. Practically, appropriate mesh
refinements should be specified in specific regions based on local flow properties to avoid too coarse mesh
that results in inaccurate result or too fine mesh that cause excessive computing time. For the drillship
flow problem, the critical zones include bow, Kelvin waves, moonpool, hull surface, prism layer and hull
bottom.
Due to the complexity of the flow field inside the moonpool and its significance on the drillship
resistances, mesh resolution inside the moonpool should be carefully controlled to ensure the accuracy of
CFD simulations. As known, violent free surface and strong vortex shedding from the moonpool front
wall are the two dominant features inside the moonpool. Therefore the set-ups of the mesh control
volumes are more focused on the mesh size to capture free surface waves and the circulating flow inside
moonpool.
In an accurate CFD simulation, the surface mesh on the hull and in the wall boundary layer are very
important for the friction prediction. Also, to maintain the accuracy of the hull shape, sufficient mesh
refinement should be applied, especially in the areas with high curvatures. To use a wall function
approach, the first cell should be appropriately located in the fully developed turbulence boundary layer.
The corresponding wall y⫹ value is 30 ⱕ yⴙ < 200.
In addition to the hull surface mesh and the prism layer mesh, the zone beneath the hull bottom also
requires special consideration since the vortex shedding from the moonpool will affect the flow field
underneath the hull bottom.
Conclusions
In this study, the mechanism of added resistance caused by a drillship moonpool has been investigated
through both the analysis of model test data and CFD simulations. Model test data show that, unlike
conventional ships that experience nearly constant resistance during a transit, a drillship resistance
presents large fluctuations. In addition, violent water motion in the moonpool and significant added
resistance can be generated on a drillship. Flow details from CFD simulations reveal that, while a drillship
is in transit state with a constant forward speed, vortexes shed from the moonpool front wall evolve into
the moonpool and finally impinging on the real wall to create high pressure and therefore generate
additional pressure force on the drillship. In the meantime, the water inside the moonpool is agitated by
the vortices and oscillatory water motion is induced on the surface inside the moonpool.
Through the investigation, the CFD tool was benchmarked using drillship model test data. The largely
fluctuating resistance was well captured in the simulations. The predicted mean resistances are in good
agreement with the model test data, within 3% difference for a wide range of speeds. In the CFD
validation, a physics based CFD best practice was established in terms of domain set-up, mesh refinement,
turbulence model selection, etcѠ The validated CFD tool and set-ups were then implemented to carry out
a parametric study in order to investigate the effects of moonpool dimension on the drillship resistance.
In the parametric study, various moonpool dimensions were investigated. The results reveal a design
principle that smaller moonpool dimension results in smaller added resistance. In addition, from visual-
izations of the large amount of CFD simulation results, it was found the large moonpool induced added
18 OTC-27092-MS

resistance is mainly attributed to the vortices shed from the moonpool front wall. FFT analysis of the CFD
results shows that the oscillation frequency of the moonpool induced resistance is same as the vortex
shedding frequency, which is controlled by the moonpool dimension. Larger moonpool dimension induces
longer vortex shedding period. With longer period, the shedding vortex can accumulate more energy and
consequently generate higher added resistance.
For the studied drillship, parametric study results indicate that moonpool dimension has significant
effects on the drillship resistance, especially at relatively high transit speed. The increase of the resistance
with the increasing moonpool size becomes more significant at high transit speed. An effective design
procedure is demonstrated through this real case study. By optimizing the moonpool dimension, 15%
resistance reduction can be achieved. A physics based CFD best practices have been established for
industry applications through this study. Based on the parametric study, a design principle is obtained to
guide drillship moonpool design, which suggests to minimize the moonpool dimension while meets the
operation requirements.

References
CD-adapco, 2014. User guide - Star-CCM⫹ version 9.04.
Enger, S., Peric, M., and Peric, R., 2010. Simulation of flow around KCS-hull. The Gothenburg 2010 Workshop on
Numerical Ship Hydrodynamics, Gothenburg, Dec 8-10, 2010.
Ferziger, J.H. and Peric, M. 2003. Computational methods for fluid dynamics. 3rd Ed., Springer Verlag, Berlin,
Heidelberg.
Fukuda, K., 1977. Behavior of water in vertical well with bottom opening of ship and its effects on ship motion. Journal
of the Society of Naval Architects of Japan, Vol.141, pp 107–122.
Hammargren, E. and Tornblom, J., 2012. Effect of the moonpool on the total resistance of a drillship. Master of Science
Thesis, Chalmers University of Technology, Gothenburg, Sweden.
Larsson, L., Stern, F. and Visonneau, M., 2013. Numerical ship hydrodynamics, an assessment of the Gothenburg 2010
workshop. Springer Dordrecht Heidelberg London New York.
Molin, B., 2001. On the piston and sloshing modes in moonpools. Journal of Fluid Mechanics, Vol.430, pp27–50.
Newman, N., 1977. Marine hydrodynamics, The MIT Press, Cambridge.
Son, H., Choi, S., and Kim, M., 2008. Drag reduction of recess type moonpool under vessel’s forward speed. Proceedings
of ASME 27th International Conference on Offshore Mechanics and Arctic Engineering. Estoril, Portugal.
Veer, R.v. and Tholen, H. J., 2008. Added Resistance of Moonpools in Calm Water. Proceedings of ASME 27th
International Conference on Offshore Mechanics and Arctic Engineering. Estoril, Portugal.
Wang, S., and Quah, M., 2013. CFD simulation of water oscillation in the moonpool of a drillship. SNAME 34th Annual
Journal 2012-2013, pp50 –69.
Weiss, J., Maruszewski, J.P., Smith, W.A., 1999. Implicit solution of preconditioned Navier- Stokes equations using
algebraic multigrid. AIAA, J., 37, 29 –36.
Wilcox, D. C., 2006. Turbulence modeling for CFD, 3rd edition. DCW Industries, Inc., La Canada CA, 2006.