Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT
This paper describes ongoing research to develop a flexible 3D hullform design process and modules with
associated performance models, and integrate these modules into an existing ship synthesis model (SSM) in
a multi-objective optimization approach to perform Naval Ship Concept and Requirements Exploration
(C&RE). Effectiveness is initially based on seakeeping indices and resistance, and then extended to a multi-
objective genetic optimization of an Offshore Patrol Vessel (OPV) total ship design.
KEY WORDS
Ship design; Hullform; Multi-objective optimization, Seakeeping, Operational Effectiveness Models
INTRODUCTION
The goal of the research described in this paper is to develop a flexible 3D hullform design process and modules based on
ORCA3D and Rhino (DRS, 2011) with associated performance models, and integrate these hullform modules into an existing
ship synthesis model (SSM) in a multi-objective optimization to perform Naval Ship Concept and Requirements Exploration
(C&RE) (Brown and Thomas 1998; Brown and Kerns 2010). Objectives include:
Establish Rhino/ORCA3D design variable lanes for displacement and semi-displacement hullforms consistent with small
naval surface combatants.
Build response surface models for estimating hullform hydrostatic, seakeeping and radar cross-section characteristics for
application in C&RE.
Assess the influence of selected ORCA3D design variables on resistance, seakeeping characteristics and RCS.
The Naval Ship Design Concept and Requirements Exploration (C&RE) process used at Virginia Tech, shown in Figure 1, is
based on a Multi-Objective Optimization approach that explores the design space to identify a non-dominated set of ship
design solutions ranked by cost, risk, and effectiveness. Our current method of calculating an Overall Measure of
Effectiveness (OMOE) used in this process is based on expert opinion and pairwise comparison. In the past, it was sufficient
in this method to use a relatively simple and traditional parametric hullform model and design variables (LBP, B, D, T, Cp,
Cx, Crd) with performance based on parametric resistance algorithms (Holtrop 1984) and seakeeping indices (Bales 1980).
Despite the useful results obtained using expert opinion for effectiveness metrics (Stock and Brown 2008; Stepanchick and
Brown 2007), more direct physics-based Operational Effectiveness Models (OEMs) starting with a detailed Design Reference
Mission (DRM) including mission Operational Situations (OpSits), conditions, and measures may provide greater confidence
in the validity of the results and a greater perception that results are unbiased and rational (Kerns et al 2011a; Kerns et al
2011b). Many of these OEMs require analyses that depend on a 3D hullform model. These analyses include resistance,
seakeeping, ship vulnerability and radar cross section. Other performance, synthesis and feasibility analyses also benefit from
having a 3D hullform model early in the design process including space, structural weight, and stability. Having a 3D model
in Concept Development also facilitates transition to preliminary design. Fine tuning and detailed optimization of the
hullform using more sophisticated models and methods can come later, but greatly benefit from a solid foundation based on
early decisions that are consistent with the overall cost/effectiveness/risk of the total ship design.
1
Graduate Student, Aerospace and Ocean Engineering, Virginia Tech (VT), Blacksburg, VA, USA
2
Undergraduate Student, Aerospace and Ocean Engineering, Virginia Tech (VT), Blacksburg, VA, USA
3
Professor, Aerospace and Ocean Engineering, Virginia Tech (VT), Blacksburg, VA, USA
Figure 1: Concept and Requirements Exploration Process (C&RE)
Simulations and optimization algorithms can be combined together early using Simulation-Based Design (SBD) techniques
including response surface models to provide inexpensive approximations of expensive analysis codes. For hullform
modeling, it is helpful to use models that are not overly complex, but that inherently include reasonable physical and
feasibility constraints, and a practical set of design variables that capture important characteristics of the basic hull geometry
to be modeled. It is also advantageous to use as much commercial-of-the-shelf software as possible. Rhino with an ORCA3D
plugin (DRS, 2011) was selected for the hullform modeling tool used in this paper based on this criteria.
HULLFORM DESIGN
One of the most difficult steps in designing a hull is creating the initial 3D shape; modifying and fairing the shape is
relatively straightforward. We use Orca3D Hull Assistant for our hullform design tool (DRS 2011). Orca3D includes a
number of Hull Assistants that allow the user to specify a set of practical design parameters and create a 3D NURBS surface
which can then be modified and faired in Rhino to produce a final hull shape. Orca3D adds the capability in Rhino to define
sections, buttocks, waterlines, cant frames, inclines, and diagonals, and compute intact hydrostatics.
Figure 2 illustrates the somewhat unique design variables used in the Orca3D Hull Assistant to define a 3D NURBS
hullform. The approach takes advantage of the general characteristics of a displacement hullform and shapes the hull using
familiar characteristics, but does not explicitly use the more traditional sectional area curve and area/volume ratios that are
very common. Implicit in these variables and their permissible ranges are reasonable constraints and characteristics for a
displacement hull. This reduces the requirement for additional constraints in the optimization process.
Figure 2: Orca3D Design Variables (DRS 2011)
The Orca3D Hull Assistant uses 33 design variables, but it was hoped that many of these could be predetermined and set to
fixed values for all designs, effectively reducing the design space to a series of designs with some common characteristics
consistent with the mission and good practice. Since the literature provides no established design lanes for these variables, we
first set out to reverse-engineer a collection of recent surface combatant hullforms, and for contrast several other naval
surface ship hullforms, by modeling/matching these in Orca3D. Best-fit values of Orca3D DVs to match these hullforms
were determined by fitting ORCA-generated hullforms to imported IGES models of the target hullforms. First
approximations at matching Orca3D hullforms to the targets were created by hand using the Hull Assistant and visually
comparing the ORCA hulls to the target hulls. These manual matches gave a good starting point for optimized fits that were
performed by running Orca3D from Model Center and using Model Centers gradient optimizer to minimize the difference
between the Orca3D hull and the target IGES hull offsets. Figure 3 shows the hullform match for a DDG hullform.
Table 1 lists the Orca3D DV values for some of the matched hulls. These values were analyzed and interpreted to choose a
subset of the Orca3D hull assistant variables as Design Variables (DVs) with design lanes for use in the hull design process,
and the remainder as Design Parameters (DPs) with fixed values for all designs based the match results. A summary of this
analysis with conclusions is provided in Table 2.
Table 1: Orca3D Design Variable (DV) Values from Match Runs
Variables: Ships:
Num UI ARS50 DDG 51 FFG 7 LPD17 NSC OPC TAKE WHEC 378
Mean Match Difference: 0.054 0.096 0.082 0.021 0.047 0.027 0.077 0.064
1 Length on Deck [m] 77.430 154.480 136.000 208.000 125.550 108.470 210.000 113.080
2 Beam on Deck[m] 15.540 20.210 14.260 30.000 16.080 16.460 32.000 12.600
3 Depth @ Bow [m] 10.370 16.800 9.410 19.000 12.000 10.200 25.650 11.050
4 Depth @ Transom [m] 10.370 9.860 9.410 19.000 12.000 10.200 25.650 8.400
5 Draft [m] 5.000 6.200 4.380 5.000 0.100 5.000 5.000 5.000
6 Transom Height (abv BL) [m] 4.000 5.800 4.000 6.000 3.310 3.625 9.500 3.200
7 Max Area Location 0.500 0.500 0.500 0.484 0.500 0.500 0.550 0.500
8 Long'l Prismatic Control 0.450 0.351 0.050 0.309 0.250 0.250 0.450 0.450
9 Section Tightness Fwd 0.700 0.200 0.500 0.501 0.994 0.700 0.500 0.600
10 Section Tightness Mid 0.800 0.990 0.400 0.401 0.900 0.700 0.400 0.500
11 Section Tightness Aft 0.500 1.000 0.300 0.801 1.000 0.990 0.600 0.300
12 Deadrise Fwd 1.000 0.836 1.000 0.800 0.300 1.000 1.000 0.900
13 Deadrise Mid 0.009 0.200 0.800 0.028 0.300 0.100 0.100 0.300
14 Deadrise Aft 0.100 0.002 0.800 0.292 0.011 0.100 0.200 0.300
15 Side Slope Fwd 1.000 0.999 0.800 1.000 1.000 1.000 0.900 0.900
16 Side Slope Mid 0.012 0.198 0.100 0.077 0.200 0.300 0.000 0.200
17 Side Slope Aft 0.100 0.199 0.300 0.000 0.200 0.200 0.000 0.200
18 Flare Fwd 1.000 0.999 0.500 0.936 1.000 0.997 1.000 1.000
19 Flare Mid 0.002 0.001 0.000 0.000 0.000 0.000 0.000 0.100
20 Flare Aft 0.000 0.002 0.000 0.000 0.000 0.000 0.500 0.000
21 Sheer Height 0.800 1.000 1.000 1.000 1.000 0.700 1.000 1.000
22 Sheer Height Position 0.500 0.500 0.500 0.500 0.500 0.200 0.500 0.500
23 Fullenss Fwd 0.300 0.600 0.300 0.199 0.400 0.400 0.000 0.500
24 Fullness Aft 0.000 0.956 0.100 0.400 0.400 0.400 0.100 0.500
25 Stem Rake [deg] 29.004 37.154 47.000 27.168 44.555 48.022 23.000 40.000
26 Stem Curvature 0.200 0.300 -0.100 0.016 0.007 -0.500 0.000 0.100
27 Bow Rounding 0.100 0.100 0.000 0.199 0.200 0.100 0.100 0.000
28 Forefoot Shape 0.500 0.600 0.950 0.500 0.200 0.600 0.500 0.000
29 Transom Rake [deg] -11.003 -16.460 -40.000 -10.349 6.000 10.004 1.000 0.000
30 Transom Deck Width 0.500 0.611 0.700 1.000 1.000 1.000 1.000 0.780
31 Keel Rise Point 0.600 0.741 0.550 0.589 0.699 0.500 0.700 0.500
32 Vel0 0.296 0.300 0.400 0.394 0.300 0.400 0.100 0.300
33 Vel1 0.005 0.229 0.050 0.106 0.200 0.000 0.100 0.100
34 Number of Net Columns 12.000 12.000 12.000 12.000 12.000 12.000 12.000 12.000
35 Number of Net Rows 9.000 9.000 9.000 9.000 9.000 9.000 9.000 9.000
Table 2: Orca3D Hullform Match Summary and Conclusions
We run DOEs, build RSMs and run optimizations, including our MOGO, using Model Center software (Phoenix Integration
2011). We typically merge data from a full-factorial DOE with a Latin-Hypercube DOE to obtain a good coverage of the
design space including the boundaries. We use polynomial or Kriging RSMs depending on their quality of fit for a particular
problem. Figure 4 shows four modules integrated in Model Center for running our hullform DOE. The Orca3D module
receives DV inputs and interfaces externally with Rhino/Orca3D to build a hullform as described, perform hydrostatic
analysis, perform a resistance calculation using a HydroComp, Inc. Holtrop/Mennen algorithm supplied with Orca3D,
generate a set offsets and calculate specified response location coordinates for input into our PDStrip seakeeping module. The
PDStrip module interfaces externally with a public-domain strip theory program, PDStrip (Source: Forge 2006) to perform
our initial seakeeping analysis. PDStrip computes the seakeeping characteristics of ships and other floating bodies using
Sodings method (Soding 1969) to calculate motions.
Seastate 4-5 is used as the limiting seastate for this paper. A significant wave height of 3 meters, modal wave period of 10
seconds and long-crested waves in head seas are used for calculating significant motions. Head seas and motions in the
vertical plane are considered worse-case in this analysis with the assumption that roll can be addressed in the design synthesis
by considering the design GM/B ratio when more is known about the ships weight distribution and KG. Significant heave,
pitch, vertical displacement at Station 15 and the approximate center of a helicopter deck (s3S15), vertical acceleration at
Station 10 (midships) and the approximate location of the bridge (a3S10), relative vertical displacement at Station 20 and the
approximate location of a stern boat ramp (r3S20), and relative vertical displacement at Station 0 and the keel (r3S0) are
computed. These motions are considered separately and combined into a simple seakeeping index similar to Bales, 1980 and
McCreight, 1984, and as discussed by Sarioz 2006.
Design data is extracted from Orca3D and PDStrip using a Design of Experiments (DOE) in Model Center over a pre-defined
design space as shown in Figure 4 and Figure 5. This data is used to build Response Surface Models (RSMs) for application
in variable screening, analysis, the total ship synthesis model, and optimization. Rhino/ORCA3D are kept open in the
background as the DOE is run in Model Center. Hullforms are generated as shown in Figure 6, hydrostatics, resistance and
seakeeping are analyzed, and the data is returned and collected in Model Center. POFACETS is a program for analyzing
radar cross section. We have interfaced POFACETS with our hullform geometry and we are in the process of adding a
deckhouse to this automated process at which time RCS will be addressed. It is not discussed further in this paper. Influence
plots generated from the DOE data are shown in Figure 7, and representative cuts of the Response Surface Models (RSMs)
built from the data are shown in Figure 8.
Influence diagrams are used to screen DVs for their relative effect on model responses, in this case hydrostatic characteristics,
resistance and sustained speed, and ship motions, represented in Figure 7 by sustained speed for a specified propulsion
effective power, hull form displacement, and pitch significant amplitude in the specified seastate.
Figure 10: Longitudinal Prismatic Control Figure 14: Section Tightness Forward
Figure 17 through Figure 22 assess the consistency of individual seakeeping parameters with respect to the seakeeping index.
Figure 17 shows that the lowest significant heave for these designs is consistent with a high seakeeping index. Pitch shown in
Figure 19 is remarkably low for all of the non-dominated designs because of its importance to a number of the other
seakeeping parameters (s3S15, r3S20, r3S0). The lowest r3S20, s3S15, a3S10, and r3S0 also all occur consistent with a high
seakeeping index. This correlation indicates that the seakeeping index was a reasonable choice for this optimization.
Finally, Figure 23 shows that maximum sustained speed is also consistent with low endurance speed SHP as would generally
be expected.
Figure 17: Heave Significant Amplitude (meters) Figure 19: Pitch Significant Amplitude (degrees)
Figure 18: r3S20, Relative Vertical Displacement, Sta 20, DWL (m) Figure 20: s3S15, Vertical Displacement, Sta 15 (m)
Figure 21: a3S10, Vertical Acceleration, Sta 10, Bridge (m/s2) Figure 22: Relative r3S0, Vertical Displacement, Sta 0, BL (m)
As a final analysis of this non-dominated set, three designs were selected for further discussion. These designs are shown in
Figure 24. Design 21379 is the non-dominated design with maximum seakeeping index. Design 26734 is the non-dominated
design with minimum endurance speed resistance. Design 8428 is a non-dominated design at the knee-of-the-curve with a
high seakeeping index and good resistance. Table 3 lists the characteristic for these designs and Figure 25 shows the hullform
geometry.
Design 8428
Design 21379
Design 26734
Figure 25: Selected Designs - Hullforms
Design 21379 has the best SKI. It has the lowest length to beam ratio, a prismatic control equal to the maximum of 0.4, the
highest section tightness forward, the highest deadrise midships, the lowest fullness forward, the highest stem rake and the
highest positive stem curvature. All of these are consistent with the non-dominated results.
Design 26734 has the lowest endurance resistance and highest sustained speed. It has the highest LtoB ratio, lowest prismatic
control, the lowest section tightness forward, lowest deadrise midships, the highest fullness forward, and a low negative stem
curvature. All of these are consistent with the non-dominated results. Design 8428 DV values are between Designs 21379 and
26734 in most cases.
Pitch RAOs for these designs are shown in Figure 26. Design 21379 shows the lowest response although Design 8428 is very
close. Again, Design 26734 shows the largest pitch response.
Figure 27 plots the pitch significant response for Design 21379 in a polar plot as a function of ship speed and heading. This
shows that head sea vertical motions at 14 knots were a reasonable choice for the seakeeping index and for seakeeping
optimization.
Figure 27: Design 21379 Sea State 5 Significant Pitch Response (deg)
SHIP SYNTHESIS MODEL AND MULTI-OBJECTIVE OPTIMIZATION
As discussed earlier in the paper, the Naval Ship Design Concept and Requirements Exploration (C&RE) process used at
Virginia Tech, Figure 1, includes a Multi-Objective Genetic Optimization (MOGO) that explores the design space to identify
a non-dominated set of ship design solutions ranked by cost, risk, and effectiveness. Integral to this optimization is a ship
synthesis model (SSM) that assembles the designs and assesses their balance, feasibility, performance, effectiveness, cost and
risk. Response surface model modules for hullform hydrostatics, resistance and seakeeping were developed as described in
the previous sections and used to represent the 3D hullforms developed in Rhino/ORCA3D.
The SSM and MOGO will be presented here in the context of an Offshore Patrol Vessel (OPV) design case study. OPV is
intended to bridge the gap between the USCGs smaller FRCs and larger NSCs. It will be deployed primarily in the
outskirts of littoral regions of the Americas with the capability to stay at sea for extended periods of time. The primary
responsibilities will be Port, Waterway and Coastal Security (PWCS) and Search and Rescue (SAR). PWCS includes, but is
not limited to, the enforcement of exclusion zones and the performance of tactical reconnaissance in and around US maritime
ports. SAR will require the vessel to be capable of rescuing multiple individuals and towing any ship up to the OPVs weight
that is incapacitated. To aid in search and rescue OPV will have aerial support for helicopters and VAUVs. The OPV will
also be heavily involved in Drug Interdiction (DRUG), in which the vessel will conduct maritime interception of trafficking
operations as well the search and seizure of suspected vessels. Migrant Interdiction (AMIO) will require OPV to be capable at
all times to provide humanitarian support to any parties in need as well being fully capable to manage any number of refugees
that are intercepted. The vessel will be also be responsible for the Protection of Living Marine Resources (LMR) which
entails the enforcement of maritime fishing and wildlife regulations and the apprehension of non-cooperative vessels and
those on board. Finally, OPV is charged with Defense Readiness (DR) in which the vessel must engage in the collection of
tactical intelligence and the provision of Harbor Defense and Port Security.
Following the C&RE process shown in Figure 1, a clear mission definition was developed including a Mission Essential Task
List (NMETL), Operational Situations, a Design Reference Mission (DRM) and Required Operational Capabilities (ROCs).
Measures of Performance (MOPs) were developed from these which were assembled in an Overall Measure of Effectiveness
(OMOE) using the Analytical Hierarchy Process (AHP) and expert opinion. The resulting OMOE hierarchy is shown in
Figure 28 with the resulting MOP weights in Figure 29. A technology risk register and Overall Measure of Risk (OMOR)
were also developed. An enhanced weight-based cost model was used to estimate acquisition and total ownership costs. The
design space for hullform, power and propulsion, mission/combat systems including boats, helos and UAVs, Table 4, was
developed to consider a broad range of available and required technologies and systems.
Figure 33 shows the distribution in seakeeping index for the non-dominated set with a mean of 0.575, lower-midrange for the
designs optimized only for seakeeping and resistance. Figure 36 shows the non-dominated set distribution for sustained speed
with three speed ranges corresponding to the three main engine options.
Of the design variables discussed above which seemed to max (or min) out in the optimization for seakeeping and resistance,
we find good distributions for Length to Beam ratio (Figure 34), Section Tightness Midships (Figure 35), and Stem Rake
(Figure 38). Only Beam to Draft ratio continued to push its upper limit and could benefit from being extended up to 3.4 or
3.5. Overall the hullform, seakeeping and resistance modules functioned very well in the MOGO with consistent results
throughout.
Figure 33: Non-Dominated Set Seakeeping Index Figure 36: Non-Dominated Set Sustained Speed
Figure 34: Non-Dominated Set Length to Beam Ratio Figure 37: Non-Dominated Set Beam to Draft Ratio
Figure 35: Non-Dominated Set Section Tightness Midship Figure 38: Non-Dominated Set Stem Rake (deg)
SUMMARY AND CONCLUSIONS
In this paper, we have described the application of Rhino/ORCA3D and PDStrip to hullform generation and analysis.
Rhino/ORCA3D design lanes were developed for a medium-sized surface combatant displacement hull, and a working
understanding of the relationship between Rhino/ORCA3D design variables, seakeeping and resistance was developed and
presented. Response surface models for hullform hydrostatics, resistance and seakeeping were derived from data using
Rhino/ORCA3D and PDStrip. These models were applied as modules in a hullform-only optimization and in a multi-
objective genetic optimization of an OPV ship design. These modules worked very effectively in both optimizations
providing consistent and effective results in both.
Future work includes adding a bulbous bow and a simple deckhouse to the hullform design, radar cross-section analysis,
basic subdivision, an automated interface for hull structural design and optimization, and an early stage vulnerability analysis.
Similar work has also been completed for a semi-planing hullform which will be the subject of a future paper. We are also
exploring the application of physics-based operational effectiveness models (OEMs) in place of our expert opinion-based
effectiveness models. These OEMs will require more quantitative ship motions analysis and integration in operations like
helo launch and recovery and boat launch and recovery. These will also be subjects of our future work.
ACKNOWLEDGEMENTS
The authors wish to thank our gracious sponsor, Ms. Kelly Cooper, ONR 33, for her support of this project.
REFERENCES
BALES, N.K., Optimizing the Seakeeping Performance of Destroyer-type Hulls, 13th Symposium on Naval
Hydrodynamics, Shipbuilding Research Association of Japan, Tokyo, 1980
BROWN, A.J., THOMAS, M., "Reengineering the Naval Ship Concept Design Process," From Research to Reality in Ship
Systems Engineering Symposium, ASNE, 1998
Brown, A.J., KERNS, C., Multi-Objective Optimization in Naval Ship Concept Design, Marine Systems and Technology
(MAST) 2010 Conference, Rome, Italy, 9-11 November 2010
KERNS, C., BROWN, A.J., WOODWARD, D., Application of a DoDAF Total-Ship System Architecture in Building
Naval Ship Operational Effectiveness Models, MAST Americas 2011, Washington DC, 1416 November, 2011
KERNS, C., BROWN, A,J., WOODWARD, D., Application of a DoDAF Total-Ship System Architecture in Building a
Design Reference Mission for Assessing Naval Ship Operational Effectiveness, ASNE Global Deterrence and Defense
Symposium, Bloomington, IN, September 13-14, 2011
HOLTROP, J., "A Statistical Re-Analysis of Resistance and Propulsion Data", International Shipbuilding Progress, Vol. 31,
No. 363 November, 1984
MCCREIGHT, W.R., Estimating the Seakeeping Qualities of Destroyer-Type Hulls, DTNSRDC Report/ SPD-1074-01,
January, 1984
SARIOZ, K. and SARIOZ, E., Practical Seakeeping Performance Measures for High Speed Displacement Vessels, Naval
Engineers Journal, No. 4, 2006
STROCK, J., BROWN, A.J., Methods for Naval Ship Concept and Propulsion System Technology Exploration in a CGX
Case Study, Naval Engineers Journal, Vol. 120, No. 4, pp. 95-122, 2008
STEPANCHICK, J., BROWN, A.J., Revisiting DDGX/DDG-51 Concept Exploration, Naval Engineers Journal, Vol. 119,
No. 3, 67-88, 2007
SODING, H., Eine Modifikation der Streifenmethode, Schiffstechnik, 16, pp.15-18, 1969
Source Forge, Open Source Software, PDStrip Public Domain Strip Method, http://pdstrip.sourceforge.net/, 2006.