Design Methodology for Cross-flodding Connects on Naval Vessels

Unclassified

Copyright © USCG Engineering Logistics Center [2004]

Unclassified

This document is supplied by QinetiQ for USCG Engineering Logistics Centerunder Contract No. DTCG40-03-P-40387.

Design methodology for cross-flooding connects on Navalvessels

Mr A Peters; Mr M GallowayQinetiQ/FST/CR033339/1.0January 2004

Requests for wider use or release must be sought from:

Intellectual Property DivisionQinetiQ LtdCody Technology ParkFarnboroughHampshireGU14 0LX

Unclassified

QinetiQ/FST/CR033339/1.0

Unclassified

Administration pageCustomer InformationCustomer InformationCustomer InformationCustomer Information

Customer reference number DTCG40-03-P-40387

Project title Design methodology for cross-floodingconnects on Naval vessels

Customer Organisation USCG - Engineering Services

Customer contact Mr P Minnick

Contract number

Date due January 2004

Principal authorPrincipal authorPrincipal authorPrincipal author

A J Peters BEng CEng MRINA +44 (0) 2392 335217

QinetiQ HaslarHaslar roadGosportHants PO12 2AG

[email protected]

Authorised byAuthorised byAuthorised byAuthorised by

Name Dr M R Renilson

Post Technical Manager, Hydrodynamics

Signature

Date of issue January 2004

Record of changesRecord of changesRecord of changesRecord of changes

Issue Date Detail of Changes

1.0 January 2004

Unclassified


Unclassified

AbstractThe scope of this project was to develop a methodology for the design of cross-connects on Naval vessels. As a demonstration of the state-of-the-art designphilosophy and the use of time-domain simulation for design, an alternative cross-connect arrangement for the USCG’s 270-ft (Corvette-sized) WMEC ‘Famous’ Classship (CG270) was performed. After using the methodology and the time-domainprogram to design the new system the performances of both the current andalternative cross-flooding arrangements were compared in relation to current IMOand USCG criteria. The study was conducted using the FREDYN program in order tocompare the time-domain analysis of the current and alternative cross-flood systems.The existing method to statically calculate, by hand the time to cross-flood isexamined and the merits and shortcomings of the two cross-connect arrangementsare then discussed.

The design and analysis of the cross-flooding ducts were performed using a non-linear time-domain ship motion program called FREDYN. The use of a time-domaincode allows cross-flooding ducts to be modelled to take account of the vessel motionand transient flow after the damage. This allows the effectiveness of the cross-flooding ducts and the time taken to cross-flood to be assessed in a seaway.

The worst damage case under the current USCG criteria was selected as the test caseincorporating the tanks containing a cross-flooding system. For each damagescenario a set of thirty-minute simulations were performed with the vessel in a deepseagoing condition for a matrix of speed, heading sea state conditions. Thesimulations were repeated with and without the existing cross-flooding activatedand with the new system activated.

In the damage cases tested, the cross-flooding system was shown to improve the“after damage” performance of the vessel. In most of the cases tested the damage listangle was reduced by up to 8 degrees. The effectiveness of the new design of cross-flooding ducts was demonstrated.

Unclassified


Unclassified

Blank pageBlank pageBlank pageBlank page

Unclassified


Unclassified

Executive summaryThe scope of this project was to develop a methodology for the design of cross-connects on Naval vessels. As a demonstration of the state-of-the-art designphilosophy and the use of time-domain simulation for design, an alternative cross-connect arrangement for the USCG’s 270-ft (Corvette-sized) WMEC ‘Famous’ Classship (CG270) was performed. After using the methodology and the time-domainprogram to design the new system the performances of both the current andalternative cross-flooding arrangements were compared in relation to current IMOand USCG criteria. The study was conducted using the FREDYN program in order tocompare the time-domain analysis of the current and alternative cross-flood systems.The existing method to statically calculate, by hand, the time to cross-flood isexamined and the merits and shortcomings of the two cross-connect arrangementsare then discussed.

The FREDYN hullform was provided by the customer with a pertinent set of drawingsof the vessel. Compartment definition in FREDYN is limited to orthogonal planedefinitions of the compartment boundaries with the exception of the hull. A 3D solidstatic stability model was created using the PARAMARINE tool, which was used tocalculate hydrostatics and static stability calculations.

The analysis of the cross-flooding ducts was performed using a non-linear time-domain ship motion program called FREDYN. The use of a time-domain code allowsthe ducts to be modelled to take account of the vessel motion and transient flowafter the damage is initiated. This allows the effectiveness of the cross-flooding ductsand the time taken to cross-flood in a seaway to be assessed.

The worst damage case under the current criteria was selected as the test caseincorporating the compartments containing a cross-flooding system. For eachdamage scenario sets of thirty-minute simulations were performed with the vessel ina deep seagoing condition for a matrix of speed, heading and sea state conditions.The simulations were repeated with and without the existing cross-flooding systemactivated and with the new system activated.

In the damage cases tested the cross-flooding was shown to improve the afterdamage performance of the vessel. In most of the cases tested the damage list anglewas reduced by up to 8 degrees. The effectiveness of the new design of cross-floodingducts was demonstrated.

This work was funded by USCG ELC under customer reference DTCG40-02-Q-41363QinetiQ assignment code 300411 0001.

Unclassified


Unclassified

Blank pageBlank pageBlank pageBlank page

Unclassified


Unclassified

List of contentsAdministration page 2

Abstract 3

Executive summary 5

List of contents 7

List of figures 10

List of tables 10

1 Introduction 11

2 Review of cross-flooding systems 132.1 Criteria for cross-flood systems 132.2 Current cross-flooding systems 142.3 Current static based hand calculations 17

3 FREDYN and cross-flooding methodology 193.1 FREDYN and the extreme motions of damaged ship 193.2 Cross-flooding design 213.3 Duct routing and sizing 233.4 Duct positioning 233.5 Effects of through life growth 24

4 USCG 270-ft WMEC cutter 254.1 FREDYN and PARAMARINE model 254.2 Damage extents 254.3 CG270 model generation 264.4 Cross-flooding modelling in FREDYN 304.5 Alternative duct design 314.6 Matrix of tests 314.7 Run selection 324.8 Ship condition 33

5 Simulations 34

6 Discussion 356.1 Run Set 1 - Variation with ship speed and heading 356.2 Run Set 2 - Variation with ship loading condition 366.3 Run Set 3 - Variation with sea state 376.4 Run Set 4 - Variation with position on wave at damage onset 386.5 Results of parametric variation 38

Unclassified


Unclassified

7 Conclusions 40

8 Recommendations 41

9 References 42

10 TablesTable 1: Sea state wave heights and period 43

11 FiguresFigure 1: No duct - sea state 5 - beam seas - 7 kts - 0 kts - deep condition 44Figure 2: USCG duct - sea state 5 - beam seas - 7 kts - 0 kts - deep condition 44Figure 3: QinetiQ duct - sea state 5 - beam seas - 7 kts - 0 kts - deep condition 44Figure 4: No duct - sea state 5 - beam seas - 7 kts - 7 kts - deep condition 45Figure 5: USCG duct - sea state 5 - beam seas - 7 kts - 7 kts - deep condition 45Figure 6: QinetiQ duct - sea state 5 - beam seas - 7 kts - 7 kts - deep condition 45Figure 7: Heading vs transient roll angle at 12 kts before damage 0 kts after

damage 46Figure 8: Heading vs transient roll angle at 7 kts before damage 0 kts after

damage 46Figure 9: Heading vs transient roll angle at 7 kts before damage 7 kts after

damage 46Figure 10: Heading vs RMS roll angle at 12 kts before damage 0 kts after damage 47Figure 11: Heading vs RMS roll angle at 7 kts before damage 0 kts after damage 47Figure 12: Heading vs RMS roll angle at 7 kts before damage 7 kts after damage 47Figure 13: Heading vs transient roll angle - no duct 48Figure 14: Heading vs RMS roll angle - no duct 48Figure 15: Heading vs mean list angle - no duct 48Figure 16: Heading vs transient roll angle - USCG duct 49Figure 17: Heading vs RMS roll angle - USCG duct 49Figure 18: Heading vs time to cross-flood - USCG duct 49Figure 19: Heading vs transient roll angle - QinetiQ duct 50Figure 20: Heading vs RMS roll angle - QinetiQ duct 50Figure 21: Heading vs time to cross-flood- QinetiQ duct 50Figure 22: Heading vs time to cross-flood at 12 kts before damage 0 kts after

damage 51Figure 23: Heading vs time to cross-flood at 7 kts before damage 0 kts after

damage 51Figure 24: Heading vs time to cross-flood at 7 kts before damage 7 kts after

damage 51Figure 25: No duct - sea state 4 - beam seas - 0 kts - light condition 52Figure 26: USCG duct - sea state 4 - beam seas - 0 kts - light condition 52Figure 27: QinetiQ duct - sea state 4 - beam seas - 0 kts - light condition 52Figure 28: No duct - sea state 4 - beam seas - 0 kts - deep condition 53Figure 29: USCG duct - sea state 4 - beam seas - 0 kts - deep condition 53Figure 30: QinetiQ duct - sea state 4 - beam seas - 0 kts - deep condition 53Figure 31: Heading vs transient roll angle in sea state 3 - deep condition 54Figure 32: Heading vs transient roll angle in sea state 4 - deep condition 54Figure 33: Heading vs transient roll angle in sea state 5 - deep condition 54

Unclassified


Unclassified

Figure 34: Heading vs transient roll angle in sea state 3 - light condition 55Figure 35: Heading vs transient roll angle in sea state 4 - light condition 55Figure 36: Heading vs transient roll angle in sea state 5 - light condition 55Figure 37: Heading vs transient roll angle - no duct 56Figure 38: Heading vs RMS roll angle - no duct 56Figure 39: Heading vs mean list angle - no duct 56Figure 40: Heading vs transient roll angle - USCG duct 57Figure 41: Heading vs RMS roll angle - USCG duct 57Figure 42: Heading vs time to cross-flood - USCG duct 57Figure 43: Heading vs transient roll angle - QinetiQ duct 58Figure 44: Heading vs RMS roll angle - QinetiQ duct 58Figure 45: Heading vs time to cross-flood - QinetiQ duct 58Figure 46: Heading vs RMS roll angle in sea state 3 - deep condition 59Figure 47: Heading vs RMS roll angle in sea state 4 - deep condition 59Figure 48: Heading vs RMS roll angle in sea state 5 - deep condition 59Figure 49: Heading vs RMS roll angle in sea state 3 - light condition 60Figure 50: Heading vs RMS roll angle in sea state 4 - light condition 60Figure 51: Heading vs RMS roll angle in sea state 5 - light condition 60Figure 52: Heading vs time to cross-flood - in sea state 3 61Figure 53: Heading vs time to cross-flood - in sea state 4 61Figure 54: Heading vs time to cross-flood - in sea state 5 61Figure 55: Sea state vs transient roll angle at beam seas - opening towards -

0 kts 62Figure 56: Sea state vs transient roll angle at beam seas - opening towards -

12-0 kts 62Figure 57: Sea state vs transient roll angle at beam seas - opening towards -

7-7 kts 62Figure 58: Sea state vs RMS roll angle at beam seas - opening towards - 0 kts 63Figure 59: Sea state vs RMS roll angle at beam seas - opening towards - 12-0 kts 63figure 60: Sea state vs RMS roll angle at beam seas - opening towards - 7-7 kts 63Figure 61: Sea state vs time to cross-flood - 0 kts 64Figure 62: Sea state vs time to cross-flood - 12-0 kts 64Figure 63: Sea state vs time to cross-flood - 7-7 kts 64Figure 64: Sea state vs transient roll angle at beam seas - no duct 65Figure 65: Sea state vs transient roll angle at beam seas - USCG duct 65

Initial distribution list 66

Report documentation page 67

Unclassified


Unclassified

List of figuresFigure 2-1: Straight duct - cross-section 15Figure 2-2: Inverted U type duct - cross-section 15Figure 2-3: Double bottom tank - cross-section 16Figure 2-4: Double duct - cross-section 16Figure 2-5: Double duct - plan view 16Figure 4-1: Damage zone and cross-flooding duct location 25Figure 4-2: PARAMARINE hullform 27Figure 4-3: Stern of CG270 28Figure 4-4: Damage region 28Figure 4-5: Cross-flooding tanks and existing duct 29Figure 4-6: Alternative QinetiQ cross-flood design (red) 31

List of tablesTable 4-1: Damage region 26Table 4-2: Run list 32Table 5-1: Ship conditions 34Table 6-1: Results from parametric variation 38-39

Unclassified


Unclassified

1 Introduction1.1 In 1990, a Co-operative Research Navies (CRN) committee was established with the

aim of producing intact dynamic stability criteria for naval vessels. The criteria wererequired to ensure that new vessels were safe, while avoiding the high build and lifecycle costs associated with over-engineering. The CRN committee comprisesrepresentatives from the Australian, Canadian, Netherlands, France, UK and USMOD/DoDs, the US Coast Guard, QinetiQ, DRDC, MARIN and NSWCCD. To derive thedynamic stability criteria, the committee needed to test in-service and new shipdesigns, in moderate to extreme seas, to assess their relative safety and probability ofcapsize. This required an extensive sensitivity study of a number of designparameters, including KG, GM, freeboard, dynamic stability and range of positivestability. As this would be impractical at full scale and both costly and timeconsuming in terms of model experiments, a time-domain prediction code, FREDYN,was developed. FREDYN, unlike the currently available frequency-domain programs, isable to take account of the non-linearities associated with the drag forces, excitationforces and rigid-body dynamics. The program was written by MARIN with steerageand guidance from the CRN committee. The latest version of FREDYN can modelvessels with damaged compartments and cross-flooding ducts, and can predict thevessel’s behaviour in waves. This permits investigations into the dynamics ofdamaged vessels in realistic environments, rather than simple pseudo-static analysis,which is the current practice.

1.2 The effectiveness in a seaway of cross-flood (cross-connect) arrangements fitted tonaval ships has not, until recently, been well understood. The scope of this projectwas to develop a methodology using the state-of-the-art design philosophy and theFREDYN program for the design of cross-connects on naval vessels. As an example todemonstrate the methodology an alternative cross-connect arrangement wasdesigned for the USCG’s 270-ft (Corvette-sized) WMEC ‘Famous’ Class ship (CG270).Both the current and alternative cross-flooding arrangements were tested in a largematrix of scenarios and also assessed in relation to current IMO and USCG criteria.The study was conducted using the FREDYN program in order to compare the time-domain analysis of the current and alternative systems to the design guidance givenin the report “Cross-flooding of a Frigate Sized Vessel”, (Peters, March 2001). Themerits and shortcomings of the IMO hand calculation and the proposed method arediscussed as well as the two cross-connect arrangements.

1.3 Static stability based hand calculation using the IMO guidelines for the performanceof the cross-flooding ducts can be performed quite easily using data from standardstatic stability software, however, this does not take account of the vessel motions ortransient flow after damage. The use of the FREDYN program enables theperformance of the cross-flooding ducts to be analysed in a seaway and so allowingthe time taken to cross-flood to be calculated more accurately. The other advantageof the simulation is that the performance of the vessel before, during and afterflooding can be examined.

1.4 To investigate the effect on the vessel of the two cross-flooding duct designs a matrixof runs was completed with FREDYN to assess the performance of the vessel. For

Unclassified


Unclassified

each simulation, statistics of the run parameters were plotted to show the timetaken to cross-flood and the resulting vessel behaviour.

1.5 This work was funded by United States Coast Guard customer reference DTCG40-02-Q-41363 QinetiQ assignment code 350753 0001.

Unclassified


Unclassified

2 Review of cross-flooding systemsLongitudinal subdivision is common practice in ship design. This internalarrangement can introduce asymmetric flooding in damage cases which can beresolved in many ways involving improving general stability with solid ballast or withliquid loading restrictions. Cross-flooding systems are regarded as a possible solutionto this problem. The effectiveness of cross-flood arrangements fitted to naval ships ina seaway has until recently been little known.

Static stability analysis of the effectiveness of the cross-flooding ducts can beperformed using standard static stability software, but this does not take account ofthe vessel motions or transient flow after damage. However, large-amplitude motiondynamics play an important role in the capsize behaviour of a frigate in waves and tothe performance of the cross-flooding arrangement fitted. The use of a time-domainsimulation program enables the performance of the cross-flooding ducts to beanalysed in a seaway, thus allowing the time taken to achieve cross-floodedequilibrium to be assessed.

As with many static-based stability criteria adopted around the world, the originsdate back to data and information gathered over many years. This applies especiallyto the great Pacific Typhoon of December 1944, which struck vessels of USN PacificFleet causing the loss of 790 men and three destroyers (see Calhoun, 1981). Followingthis incident a review of stability assessment was undertaken, which resulted in newstability criteria for US Navy ships (Sarchin and Goldberg, 1962). This covers the intactand damaged stability criteria, which has been adopted by many Navies around theworld including the USCG and UKMOD.

2.12.12.12.1 Criteria for cross-flood systemsCriteria for cross-flood systemsCriteria for cross-flood systemsCriteria for cross-flood systems

The 1960 SOLAS conference first laid out the requirements for cross-flooding systemswhere a maximum time for cross-flooding was defined as 15 minutes. This limit wasprobably based on evidence from the time and the need to set an achievablestandard. This criteria is now included in the current regulations as Regulation 8 (5) inChapter 2 Part B of the International Convention for the Safety of life at sea (SOLAS,2001). This criterion for cross-flooding is stated as follows:

5. Unsymmetrical flooding is to be kept to a minimum consistent with efficientarrangements. Where it is necessary to correct large angles of heel the means adoptedshall, where practical, be self-acting, but where controls to cross-flooding fittings areprovided they shall be operable from above the bulkhead deck. These fittings shall beacceptable to the Administration. The maximum angle of heel after flooding but beforeequalisation shall be less than 15 degrees. Where cross-flooding fitting is required thetime to equalisation shall not exceed 15 min. Suitable information concerning the use ofthe cross-flooding fitting shall be supplied to the master of the ship.

The criteria goes on to state that for the unsymmetrical case that the angle of heelafter equalisation has completed should be less than 12 degrees for two or morecompartment damage.

Unclassified


Unclassified

2.1.1 DDS 079-1 criteria

The US Navy stability criteria are documented in the Design Data Sheet (DDS) 079-1(US Navy, 1975), which is divided into criteria for damage stability for both side-protected and non-protected vessels. The non-protected criteria relate to the 270-ftcutter that is the class used in this investigation. The DDS 079-1 states that an angleof less than 15 degrees is required after damage for operational requirements. Thereis no mention of cross-flood systems except for in the side-protected vessels, whichstates that the maximum list shall not exceed 20 degrees and that arrangementsexist for rapidly reducing the list to less than 5 degrees. It does not specify any timeconstraints for this.

2.1.2 USCG criteria

From the current USCG Design and Construction Standard (DCS) SWBS 079 on theuse of "Cross-Connection of Tanks" it states "cross-connection of tanks should only beemployed where other alternatives have been evaluated and are deemedimpracticable”. It then states that where cross-connection of tanks is utilised, thefollowing applies:

• The cross-flooding system shall prevent transference of liquids from onetank to the other during normal rolling of the ship.

• Cross-flooding time shall not exceed five minutes• Prior to cross-flooding the following criteria shall be met:

• Heel shall not exceed 20 degrees.• Area A1/A2 greater than or equal to 1.4.

2.22.22.22.2 Current cross-flooding systemsCurrent cross-flooding systemsCurrent cross-flooding systemsCurrent cross-flooding systems

The following duct types have been seen fitted to both commercial and naval vesselsin recent years. Different types of cross-flooding arrangements are suitable fordifferent tank positions and damage scenarios, and there is no one design suitable forall situations. Even if the tank layout is a similar shape and in a similar position to anexisting design the size of the required ducts is still unknown. A brief description ofsome of the main styles of cross-connection duct designs that are commonly used isgiven below, with some of their advantages and disadvantages.

Unclassified


Unclassified

Straight ductStraight ductStraight ductStraight duct

Figure 2-1: Straight duct - cross-section

This is the simplest of all cross-flooding designs, which consists of a straight pipeconnecting the bottom of two wing tanks together. This creates a totally passivesystem, which is in-line with the IMO guidelines for passenger ships, which statesthat cross-flooding systems should be self-starting where possible. This system isoften used where normally empty compartments are causing the asymmetry. Thissystem is less suitable when it connects two tanks, as this system allows easy transferof liquids. This is not desirable and can cause stability issues, and contamination if thetanks contain different fluids. A variation of this system has a valve in the centre ofthe pipe, which stops the tanks mixing fluids in normal service. The down side to avalve is that it has to be manually opened following damage which means there is anadditional risk and time delay to the cross-flooding operation due to the requiredhuman action.

“Inverted U” type duct“Inverted U” type duct“Inverted U” type duct“Inverted U” type duct

Figure 2-2: Inverted U type duct - cross-section

The “inverted U” type duct is often used to connect tanks together in a similarmanner to the straight duct but with the reduction of the possibility of the fluidsmixing during the ship’s normal motions due to its shape. As long as the top of the UTube remains permanently below the damaged waterline the fluids will cross-flood

Comp1

Comp2

Tank 1Tank 2

Unclassified


Unclassified

successfully after damage. If the top of the pipe emerges from the water then theduct will not function. The design of this type of duct requires careful design toensure that the duct will function after damage.

Figure 2-3: Double bottom tank - cross-section

The pipe arrangement above is a similar system that is often used to connect doublebottom tanks together. This design, because of its low position in the ship, meansthat the tanks are prevented from mixing liquids in the intact state. After the damageevent, the tanks will cross-flood continuously if the duct stays below the damagewaterline, which it is likely to be the case due to its low position in the ship.

Double duct arrangementDouble duct arrangementDouble duct arrangementDouble duct arrangement

Figure 2-4: Double duct - cross-section Figure 2-5: Double duct - plan view

The double duct system is another type of arrangement that has been used in recentyears. The layout of the pipes prevents, or at least minimises, the mixing of the tanks.There are both disadvantages and advantages to this type of system. Thedisadvantage of this system is that double the amount of pipe is required, butdepending on the damage waterline it is possible for both pipes to flood whichdecreases the time to cross-flood and incorporates some element of redundancy inthe system. The downside to this system is that one tank cannot be pressed full whilethe other is empty.

Tank 1Tank 2

Tank 2 Tank 1

Tank 1Tank 2

Unclassified


Unclassified

These previous used arrangements show some of the basic designs for cross-floodingsystems, but actual final positioning and sizing of the duct is critical to the finalperformance and requires careful evaluation.

2.32.32.32.3 Current static based hand calculationsCurrent static based hand calculationsCurrent static based hand calculationsCurrent static based hand calculations

The current practice for a designer assessing the performance of cross-floodingsystems involves using an approximate formula that was derived by Dr Ing Gino Soldain 1961 (Solda, 1961). This formulation takes account of the static water head at thestart and end point of the cross-flooding and the amount of water to cross-flood. Theshape and length of the pipe is also taken into account through the inclusion of atotal pipe friction coefficient. This formulation then provides a simple answer to thetime to cross-flood based on static calculations.

The formula is as follows:

��

��

� −��

�

�

��

�

�−

=

HoHfgHo

HoHf

sfW

To1

1

2

12 (1)

To = Time to cross-flood

W = Total volume of water for equalisation

s = Cross-sectional area of cross-flooding pipe

f = Flow reduction factor for the duct

g = Acceleration due to gravity

Ho = Head of fluid before equalisation

Hf = Final head of water (after complete equalisation)

This formulation is suitable for calculating an initial figure for the time to cross-floodat the early stages of design. Care must be taken in the calculation of the ‘f’ term,which is the flow reduction factor that is based on calculations for flow in pipes. Thisis also true for the time-domain computational method as the calculation of the flowreduction factor is also required. It was shown by Peters (2001) to be possible withcare that suitable values could be produced for the total friction coefficient and usedsuccessfully in the time-domain simulations. This formulation does not take accountof the transient roll or the motion of the vessel during the cross-flooding process. It isalso not always straightforward for the designer to select the water height head touse in these calculations. Using a time-domain program like FREDYN allows theparameters that effect the cross-flooding performance to be assessed bothindependently or together.

Unclassified


Unclassified

In comparison to the tank experiments (Peters, 2001), it was shown possible in calmwater to achieve predictions using this static formula with about 10% error, oftenunder predicting the time to cross-flood. Relatively small changes in the pressureheads were found to change the time to cross-flood significantly. It was easy toachieve 30 to 40% differences in the time to cross-flood in comparison to the resultsfrom the experiments and time-domain simulations due to inaccuracies in theprediction of the pressure heads. An additional problem occurred with double ductarrangements due to predicting the contribution by the second duct which wasdifficult to determine. In the example in this report, using just the damage drafts toapproximate the damage water heads again caused differences of the samemagnitude. This meant that an accurate full hydrostatic computer model of thedamaged ship was required to get accurate water head data to achieve good resultsusing the formulation above in calm water. By exporting the geometry to a programlike FREDYN, with little additional effort a full detailed investigation of the cross-flooding can be made.

Unclassified


Unclassified

3 FREDYN and cross-flooding methodologyThe current practice for any design of a cross-flood system involves basic staticanalysis with the calculations using the assumption of calm water. In reality thevessel would be rolling and pitching about in waves causing the pressure heads tocontinually change thereby effecting how the cross-flooding system operates. Therecent advances in time-domain simulation now allows a damage simulation to beconducted in 6 degrees of freedom with cross-flooding and down-flooding included.This allows the transient behaviour and motions after damage to be taken intoaccount when evaluating the cross-flooding system.

3.13.13.13.1 FREDYN and the extreme motions of damaged shipFREDYN and the extreme motions of damaged shipFREDYN and the extreme motions of damaged shipFREDYN and the extreme motions of damaged ship

The FREDYN program was written by MARIN with steerage and guidance from theCRN committee. FREDYN was designed to enable the simulation of motion of an intactsteered ship in wind and waves (MARIN, 2002). Unlike the currently availablefrequency-domain programs, FREDYN is able to take account of the non-linearitiesassociated with the drag forces, excitation forces and rigid body dynamics. Theapproach is a physical one, where all physical factors are considered. Both the viscousforces and the potential forces are added to complete the physical model. Non-linearities have to be considered as they arise from:

• Effect of large angles on excitation forces.• Rigid body dynamics with large angles.• Drag forces associated with hull motions, wave orbital velocities and

wind; and• Integration of wave induced pressure up to free surface.

The latest version of FREDYN can model vessels with damaged compartments andcross-flooding ducts, and can predict the vessel’s resulting behaviour in waves.

The theory for predicting the large amplitude motions with FREDYN has been describedby McTaggart and De Kat (2000) and by Van ‘t Veer and De Kat (2000). The derivation ofthe equations of motions for a ship subjected to flooding through one or more damageopenings is based on the conservation of linear and angular momentum for six coupleddegrees of freedom. The fluid inside the ship is considered as a free particle withconcentrated mass; using this approach classical rigid-body dynamics can be used toderive the equations of motion.

3.1.1 Time-varying mass

In time-domain simulations it is necessary to integrate first-order equations of theform (see De Kat and Peters, 2002):

( , , )

x v

v f v x t

==

�

�

(2)

Unclassified


Unclassified

The effect of time varying mass associated with the flooding is treated as follows. Theequations are derived by taking the time derivative of momentum to give:

( )d dv dm

F mv m vdt dt dt

= = + (3)

The second term on the RHS may be moved to the LHS and combined with the forcevector:

( ')dm dv

F v m mdt dt

− = + (4)

The term m’ is the mass that varies with time due to the floodwater, while m is theactual physical mass of the body. After rearranging, this equation can be written:

'

dmF vdv dtv

dt m m

−= =

+� (5)

which is of the form required for numerical integration.

Using the above approach and generalised 6x6 mass matrices yields the followingequations of motion for a damaged vessel in the ship-fixed co-ordinate system:

( )

0 f G Gx

0 f GGy

0 f G Gz0 f G Gf

zz,0 yy ,0x

xx,0 zz ,0y

yy,0 xx ,0z

( + ).( q - r)m m w vF( + ).( r - p )um m wF( + ).( p - q )um m vF[ ] + [a( )] + [ ] . = - - [ ] . + M M x xM( - )q rI IM

( - )p rI IM( - )p qI IM

+

� ��

� � ��

∞ � ��

� ��

��

a d d it io n a l

te rm s

(6)

The matrix [M0] is the generalised 6x6 mass matrix of the intact ship, [a(∞)] is the addedmass matrix that is part of the linear radiation forces (the convolution integrals are partof the force terms in the RHS). [Mf] is the 6x6 matrix containing all ship-accelerationrelated, time-dependent mass and inertia terms associated with the floodwater,including non-zero off-diagonal terms. The summation signs in the RHS represent thesum of all external force contributions, including the effect of damage fluid, potentialflow and viscous fluid forces. The equations of motion are solved using a 4th orderRunge-Kutta scheme.

The “additional terms” in the RHS of the equations of motion stem from crossproducts, which appear when expressing the conservation of momentum in a ship-fixed co-ordinate system, and from the motion of the fluid relative to the ship.

One of the exciting force contributions that is treated "exactly" stems from thehydrostatic and dynamic wave pressure. This represents the Froude-Krylov force,

Unclassified


Unclassified

which is obtained by pressure integration over the instantaneous wetted surface ofthe hull at each time step. This will account for a large part of the non-linearities thataffect the ship response. Linear wave theory is used to describe the sea surface andwave kinematics. In the case of irregular waves, the model makes use of linearsuperposition of sinusoidal components with random phasing.

3.1.2 Water ingress and fluid loading

3.1.2.1 Hydraulic flow

To estimate the flow rates of water entering a compartment, the flooding model isbased on the Bernoulli equation (see De Kat and Peters, 2002, or Van 't Veer and DeKat, 2000). This analysis is applied to each damage opening or holes between twocompartments. It assumes stationary flow conditions and no loss of energy due tofriction or increased turbulence. Based on the difference in pressure head, the velocitythrough a damage opening can be calculated. In addition, airflow and compressioneffects are modelled using the appropriate gas laws.

To obtain the total discharge through an opening, Q, the following empiricalformulation is used:

d 2Q C v A= (7)

where A is the area of the opening and dC is the discharge coefficient. And v2 is fluidvelocity. This coefficient accounts for a combination of several effects (such as frictionlosses). Cross-flooding ducts are modelled in a similar way to this but account is takenof the friction in the pipe.

3.1.2.2 Quasi-dynamic fluid loading

Based on the computed inflow and outflow of fluid through all openings, the fluidmass inside a shipboard compartment is known at each time step. A simple yetpractical approach is to assume that the water level of the floodwater inside anycompartment remains horizontal (earth-fixed) at all times. This implies that thedamage fluid causes a vertical force (due to gravity) to act on the ship and that anysloshing effects are neglected.

Comparison of previous cross-flooding validation between a FREDYN prediction andan experiment of a damaged Leander class frigate operating in a seaway (De Kat andPeters, 2002) demonstrated that the simulation program FREDYN effectivelymodelled the motions of a ship with cross-flooding systems operational.

3.23.23.23.2 Cross-flooding designCross-flooding designCross-flooding designCross-flooding design

Using either the current static hand calculations or using the time-domain programthe following points give some guidance for consideration while designing a cross-flooding system. These points were derived during the “Cross-flooding of Frigate SizedVessels” project conducted in March 2001 for the USCG and the UKMOD (Peters,

Unclassified


Unclassified

2001). The first main point that should be noted is that cross-flooding reduces thereserve of buoyancy on the intact side of the ship. This should be investigated toascertain whether cross-flooding would firstly be beneficial or not. Cross floodingcould, in an extreme case, cause a ship to sink further and reduce waterplane area,substantially reducing stability and lowering downflooding points closer to thewaterline.

During the design of cross-flooding systems the design should be made passive orautomatic where possible, which is the case for the current and alternativearrangements. When a cross-flooding system involves human intervention oradditional machinery or pumps then the time to operate is increased, as is the risk ofthe system not operating effectively. A system requiring activation will not beginoperation during the critical seconds immediately following damage.

The straight duct system is ideal for connecting two empty tanks, void spaces orcofferdams, but if it is used to connect two fuel tanks there is likely to be constantmixing of tanks. This is also undesirable if the ship sustains asymmetric damageelsewhere then the part full fuel tanks can drain freely under gravity into the tank onthe lower side, thus degrading stability. The inverted U duct and double type ductwould not be affected to such an extent due to their design.

If the compartment will not press full when damaged then a system along the lines ofthe straight or inverted U duct is recommended, as the duct openings are likely to bewell below the waterline throughout the damage event. This will result in immediateand continual flooding to the point of equilibrium. The straight duct option will allowtank mixing in the intact state unless a valve is fitted. The inverted duct has theadvantage that its shape resists mixing, to a certain extent, in a totally passivemanner.

Where possible, cross-flooding systems should be fitted in regions where thecompartments will fill completely after damage and, if possible, remain pressed fulleven during rolling. This reduces problems associated with additional free-surfaceeffects and cross-flooding effectiveness, especially with the double type duct.

All parts of the duct route should be below damaged waterlines at all times if theinverted U type duct is used, the highest part of the duct must be formed in a way sothat none of the duct rises above the waterline any time after damage. If the ductrises above the waterline, for example during the transient roll, then cross-floodingwill not initiate until the duct submerges below the waterline. Design studies andmodel experiments at Haslar have shown that even the initial transient roll isfractionally reduced with cross-flooding systems that initiate immediately afterdamage.

When straight duct systems are fitted low in a ship that has a high beam, the positionshould be carefully assessed. It must be fitted to ensure that during the initial rollingafter damage all the openings stay below the water surface. FREDYN simulations(Peters 2001) have highlighted the case where a wider ship, with a duct opening inthe centre of the side tanks, rolled after damage to an angle so as to raise the end ofthe duct above the water, stopping completely any flow into the far side tank.

Unclassified


Unclassified

The double type duct system has shown to be very effective in certain conditions andless effective in others. It is recommended that this type of system should only befitted in compartments that will be pressed full after damage, because when thecompartment does not fill the duct often emerges above the waterline, stopping thecross-flooding. It has been shown that when the compartment is pressed full andbelow the waterline, this system floods through both ducts allowing for a rapid cross-flooding. It is recommended that the tops of the ducts should be positioned so thatthey do not rise above the water surface, particularly in the transient roll, as thesystem then only floods through one duct, increasing the time to cross-flood.

If cross-flooding is to connect two tanks together then the tanks should contain thesame type of contents, as there will definitely be some mixing with passive cross-flooding systems. Ducts should be designed so as to reduce the potential sloshingbetween tanks, so reducing the mixing of tank fluids. Current designs often haveopenings at the top of the tanks or a curved connection to restrain cross-flooding. Thedesign to stop or restrain the mixing of tanks must not reduce the effect of the ductsif damage is sustained.

An option may involve a manual valve to be fitted to stop mixing if the vesseloperates in scenarios that cause large roll angles. This is only recommended in theextreme circumstances to stop the tanks mixing. Any valve should be fitted near thecentre of the duct to reduce the possibility of it not operating after damage. Theremote opening of the valve should be possible from several locations on the ship,including the bridge.

3.33.33.33.3 Duct routing and sizingDuct routing and sizingDuct routing and sizingDuct routing and sizing

It has been shown by Peters (2001) that the effect of increasing the cross-sectionalarea of the duct is proportional to the decrease in roll angle during the first 4 rolloscillations. The duct diameter used in that experiment scaled to a 0.28 m diameterduct at full scale, which is close to the size used in frigate sized vessels. It is suggestedthat the duct diameter should not be lower than 0.25 m, even in small tanks, andpreferably be as large as practically possible for the compartment. Ducts of 0.4 mdiameter are suggested as suitable for frigate type vessels as they showed a rapidcross-flooding in the double type ducts case, reducing the transient peak. Duct sizesshould be physically or computationally modelled to assess any potential free surfaceor stability problems during the cross-flooding stage. Ducting should be the shortestpossible length and contain as few bends and valves as possible to reduce frictionallosses.

3.43.43.43.4 Duct positioningDuct positioningDuct positioningDuct positioning

The pipe openings and pipe runs should be positioned so they remain below thewater line during the intact, transient and damage phases to ensure immediate andeffective cross-flooding. Other considerations are to be made to the pipe run toreduce the risk of damage to the cross-flooding arrangement during the damageevent.

Unclassified


Unclassified

For bottom and side tanks, where the risk of collision damage is highest, the ductsshould be as far as possible from the shell plating to ensure the duct itself does notget bent or blocked during damage. Double type ducts, that follow closely the shellplating, should be positioned preferably at each end of the compartment near thebulkheads, so as to offer some protection to the ducts during the damage incident.The double type duct has the disadvantage that it requires the complexity of twopipes to be fitted, although in cases tested often there is flooding through both ductsso increasing the rate of cross-flooding. It is recommended that a minimum of twoducts are fitted to a compartment to incorporate an element of redundancy so thatcross-flooding will still occur (at a slower rate) if one gets blocked through damage.

3.53.53.53.5 Effects of through life growthEffects of through life growthEffects of through life growthEffects of through life growth

The chosen design of cross-flooding system should be analysed at a range of expectedthrough life conditions for the vessel. This is to ensure that as KG and displacementgrow, the ducts would still operate effectively. The condition later in life may result in,for example, part of the duct rising above the waterline, stopping cross-floodingwhere it may not do so in an earlier condition.

Unclassified


Unclassified

4 USCG 270-ft WMEC cutterTo demonstrate the methodology and use of FREDYN in the design and guidance forcross-flooding designs the USCG 270-ft cutter was chosen as an example case for thestudy. The 270-ft cutter has an existing cross-flooding pipe between the Port andStarboard Ballast tanks 4-03-1-W and tank 4-103-1-W.

4.14.14.14.1 FREDYN and PARAMARINE modelFREDYN and PARAMARINE modelFREDYN and PARAMARINE modelFREDYN and PARAMARINE model

The modelling approach taken for this task followed Haslar standard practicedeveloped during previous FREDYN cross-flooding studies. The general approach is toidentify the damage zones and compartments, generate the appropriate computermodels to facilitate the analysis, verify these models against benchmark data andfinally run a matrix of FREDYN simulations. A detail description of the process follows.

4.24.24.24.2 Damage extentsDamage extentsDamage extentsDamage extents

The damage zones and their extents were provided by the USCG. The damage zoneswere selected based on the static stability studies where the damaged list angleswere shown to be the worst. The damage zone extents and associated cross-floodingfor the damage case are shown in Table 4-1 and their location in Figure 4-1.

Figure 4-1: Damage zone and cross-flooding duct location

Unclassified


Unclassified

FramesDamage Space Description Deck

From FromExtent

Clean Ballast 4-103-1-W DB 103 145 Stbd

Seachest - DB 117 120.5 Stbd

Engine Space - DB-1st Platform 103 165 Symm.

CPP Pump Oil 4-162-1-F DB 161.5 165 Stbd

Oily Waste 4-165-1-F DB 165 169 Stbd

Diesel Oil Service 4-165-3-F DB 165 169 Stbd

Clean Ballast Aft 4-169-1-W DB 169 186 Stbd

Void - DB 169 186 Stbd

Dry Provisions 4-169-0-A TT 165 186 Symm.

Engineers Ctrl Room 3-152-0-E 1st Platform 151.5 165 Symm.

Elevator 3-165-1-Q 1st Platform-MainDeck 165 169 Stbd

Crew Accommodation - 1st Platform 165 186 Symm.

Electronic Store 1-103-3-A Main Deck 103 124 Stbd

Uptakes - Main Deck 103 124 Stbd

Galley and Mess - Main Deck 124 165 Symm.

CPO Accommodation - Main Deck 165 186 Symm.

Table 4-1: Damage region

4.34.34.34.3 CG270 model generationCG270 model generationCG270 model generationCG270 model generation

Two computer models of the USCG 270-ft WMEC were required to perform FREDYNsimulations, a basic static stability model and the FREDYN dynamic stability model. Astatic stability model was required to provide the basic hydrostatic inputs for FREDYN;it also served as a benchmark test to validate the FREDYN model.

PARAMARINE was chosen as the software for which the static stability model wouldbe produced. The Graphics Research Corporation (GRC) developed PARAMARINE,which is the static modelling package used by the UK Ministry of Defence (UKMOD).QinetiQ (Haslar) has rigorously tested and validated PARAMARINE against puremathematical models, which gives confidence in the algorithms and equations used.This work was performed on the behalf of the UKMOD.

The hull definition was generated from a surface fit of curve geometry data providedby the USCG ELC in the form of “270wmec.hul” file containing offset data. The surfacefit operation in PARAMARINE automatically provides a good match to the curve data;however, some manual fairing was performed to remove any inflexion points. Thetransom was treated as a separate surface fit due to rapid change in curve direction.The transom surface was later “sewn” to the main hull surface; this surface was then

Unclassified


Unclassified

used to develop a solid hull definition Figure 4-2. The transom is highlighted inFigure 4-3.

The internal arrangement was generated from the general arrangement drawingsand frame sections provided by the USCG ELC. For the purpose of this task only theproposed damage section of the vessel has been fully subdivided. Each section in thedamage zone was subdivided into its watertight compartments and tanks. It shouldbe noted that this PARAMARINE model does not detail every room and passageway ineach watertight compartment, therefore some compartments may have acombination of rooms and corridors. The subdivision breakdown is displayed inFigure 4-4.

Figure 4-2: PARAMARINE hullform

Unclassified


Unclassified

Figure 4-3: Stern of CG270

Figure 4-4: Damage region

The USCG 270-ft WMEC is fitted with a cross-flooding system connecting cleanballast tank 4-103-1-W with clean ballast tank 4-103-2-W. The cross-flooding ductwas modelled using PARAMARINE to help visualise its complex shape, however, theduct was not used in the subsequent static stability analysis. The cross-floodingarrangement is shown in Figure 4-5.

Unclassified


Unclassified

Figure 4-5: Cross-flooding tanks with existing duct

Finally point buoyancies were added to reproduce the additional buoyancy gainedfrom the appendages (e.g. propeller, shaft, boss and hub).

The FREDYN hullform was provided by the USCG ELC in the form of a FREDYN “cda”file. The hull definition is based on twenty-one sections approximately 3.9 m apart.

At each stage in the development of the FREDYN model a number of checks andcomparisons were performed against the PARAMARINE design and the data suppliedby USCG ELC. The model validation was conducted in accordance with currentpractice for UK Navy computer models to Sea Systems Publication Number 24 (SSP24)Stability of Surface Ships (REFERENCE SSP24). The very basic hullform validation wasconducted, as the geometry of the hullform for FREDYN was sent from the USCG.

The FREDYN model does not take into account the additional buoyancy gained by thevessel’s appendages, therefore the PARAMARINE design was required to be modifiedso that the validation parameters would be comparable. This was the only alterationmade to the PARAMARINE design. All subsequent calculations performed used themodified PARAMARINE design.

The internal compartments were also subject to validation to SSP24 standards wheredata existed, e.g., tanks. The standards require the compartment volumes to bewithin 2%, and the vertical centre of gravity to be within 1%. There are no criteria setfor compartment longitudinal and transverse centre of gravity, as these are deemedless important. As this study involves asymmetrical damage the transverse centre ofgravity of the compartments is of great importance, therefore a 1% criteria for thetransverse centre of gravity was introduced.

Unclassified


Unclassified

In order to accurately model the cross-flooding ducts a flow coefficient is required toincorporate the friction effects in the duct. This flow coefficient is required to ensurethat the ducts flow in a realistic manner. The coefficients were calculated based onthe standard flow-in-pipe analysis. This method accounts for the size and length aswell as the shape of the duct. In selecting a coefficient extreme care should be takenas this can greatly affect the flow both in terms of flow rate and time to achieveequalisation. A sensitivity study was completed on the coefficients selected for eachduct to ensure valid calculations with small changes to the coefficients. Thecoefficients calculated for the current duct design were 5.65 and 2.14 as the duct wasmodelled in two parts to account for the height rise of the pipe. The coefficient for thealternative design was 1.74 for each duct. The existing duct coefficients are higherdue to the number of bends in the pipe run.

4.44.44.44.4 Cross-flooding modelling in FREDYNCross-flooding modelling in FREDYNCross-flooding modelling in FREDYNCross-flooding modelling in FREDYN

Defining an accurate cross-flooding system in FREDYN involves careful detailing toensure realistic modelling. The x, y and z position of the openings of the duct aredefined and so is the cross-sectional area of the pipe. The flow coefficient for the pipeis derived from using standard values mentioned above based on dimensions andshape of the duct. This was shown to be sufficiently accurate when used in FREDYNwhen compared with experiment data (Peters, 2001). For straight pipes or pipeswhere the duct openings are at the highest point of the pipe run this isstraightforward. Pipes that have a higher part to stop undesired cross-flooding whenthe tanks are intact require additional modelling. To model the existing cross-floodpipe as currently fitted to the 270-ft cutter involves modelling the duct in three parts.Firstly, at the position of the highest point of the duct a very small ‘virtual’ tank iscreated close to the size of the duct cross-section. The first duct is input with the tankopening defined on the one end and the virtual tank defined on the other end. Thecoefficients in this duct then take into account the length and bends in this part ofthe duct. A second duct then is defined between the virtual tank and the cross-flooding tank. Again the coefficients are defined for that part of the duct. This ensuresthat the vertical path of the water in the pipe is taken account of and that theflooding will occur as the real duct.

Unclassified


Unclassified

4.54.54.54.5 Alternative duct designAlternative duct designAlternative duct designAlternative duct design

Figure 4-6: Alternative QinetiQ cross-flood design (red)

As an example, an alternative cross-flooding design was created for the 270-ft cutterusing the FREDYN program and the guidelines listed in chapter 3. This was thenexamined during a large matrix of runs alongside the current cross-floodingarrangement. The chosen design was based on the double duct design, as this wasdeemed suitable for this tank layout, and is shown in Figure 4-6, which indicates theposition relative to the current duct. This pipe system joins the top of each tank to thebottom of the other. The diameter of the two ducts was smaller than that of thecurrent duct as it was expected that both ducts would cross-flood, as would bedemonstrated in the tests. Due to the position inside the tank, the ducts have a slightcurvature to them so that restrictions to the flow within each pipe is kept to aminimum. The ducts are also shorter than the current duct design. The performancebefore fine-tuning was demonstrated in the matrix of runs. The results form thetime-domain analysis highlight where improvements to both designs could be made.

4.64.64.64.6 Matrix of testsMatrix of testsMatrix of testsMatrix of tests

To fully investigate the performance of the cross-flooding arrangements a number ofsimulations were required. A list was compiled of parameters, which could affect theperformance of the ship following damage and hence how it may affect the cross-flooding. These were to be included in a matrix of runs to assess the current andalternative duct performance. Ship speed was also included to investigate if thisimproved the situation for the ship after damage. The matrix was selected to not onlythoroughly investigate the ducts but, to provide some guidance to the operator onheading and speed selection after damage, if available, and how it may affect thevessel’s behaviour.

Unclassified


Unclassified

Parameter/MatrixParameter/MatrixParameter/MatrixParameter/Matrix Run Set 1Run Set 1Run Set 1Run Set 1 Run Set 2Run Set 2Run Set 2Run Set 2 Run Set 3Run Set 3Run Set 3Run Set 3 Run Set 4Run Set 4Run Set 4Run Set 4

Ship ConditionShip ConditionShip ConditionShip Condition 1 2 1 1

Cross-flood systemsCross-flood systemsCross-flood systemsCross-flood systems 3 3 3 2

SpeedsSpeedsSpeedsSpeeds 3 1 3 1

HeadingsHeadingsHeadingsHeadings 8 3 2 1

Sea ConditionsSea ConditionsSea ConditionsSea Conditions 1 3 5 1

Damage OccurrenceDamage OccurrenceDamage OccurrenceDamage Occurrence 1 1 1 3

RepeatsRepeatsRepeatsRepeats 1 1 1 3

TOTALTOTALTOTALTOTAL 63636363 54545454 90909090 18181818

Table 4-2: Run list

4.74.74.74.7 RunRunRunRun selectionselectionselectionselection

To conduct the entire run combinations for all the initial variables that were selectedwould have resulted in a requirement for over 8000 simulations, which was agreedexcessive to meet the aims of this project.

Each set of runs in the above table concentrates on a particular part of the matrixwith the number of runs set so data trends can be deduced.

The first set of runs in the table, Run Set 1, aimed to assess the performance withoutcross-flooding, with the current design and with an alternative design with the shipat different speeds and orientation to the waves while the other variables were keptconstant. This allowed an assessment of the effect of heading and speed on theperformance of each of the cross-flooding systems to be made. The zero speed/nocross-flooding case has been used as a baseline case to identify where the situation isimproved.

The second set, Run Set 2, was selected to assess the effect of ship loading conditionon the performance of three cross-flooding systems. The selection tested twodifferent ship conditions at three headings and in three sea conditions to allow theperformance to be assessed.

The third set, Run Set 3, was selected to assess the effect of sea state on theperformance of the cross-flooding systems. In this set, the non-cross-floodingsituation, the current design and an alternative design were tested in a range of waveconditions at three speeds in beam seas (damage opening towards and away fromthe waves). This demonstrated how the cross-flooding performance is affected indifferent sea conditions as the ship motions increased. It also identified the issuesthat occur at slow forward speed following damage.

The aim of Run Set 4 was to evaluate the effect of the ship’s position on the wave toestablish how orientation effects the initial damage transient responses for the non-cross-flooding and cross-flooding cases. Repeat runs were also conducted at different

Unclassified


Unclassified

points in the wave realisation (damage initiating on top of a wave crest, in a trough,or in a quiescent location between larger wave groups) to investigate how thateffected the ship behaviour.

4.84.84.84.8 Ship conditionShip conditionShip conditionShip condition

To select a ship condition to test, the main condition for the 270-ft cutter class B deepcondition was initially selected. The current loading conditions of this class did notprovide an interesting case as all of the damage criteria were met and the static listangles were less than 15 degrees. Consequently, minor modifications were made sothat the final list angle after damage was increased to just over the current15 degrees criteria limit (USCG and SOLAS). Standard permeabilities were used exceptthe stores were lowered in permeability to 60% representative of a full store. Theengine room was also reduced slightly in permeability to 75%. The diesel oil servicetanks were also lowered to 25% full. The KG was then raised by 1.8% to give a listangle of 17.5 degrees. This gave a more suitable condition in which to test cross-flooding designs. For Run Set 3 a second condition was required which was basically aminimum operating condition. In this condition, the two ballast tanks that cross-flood are both pressed full. To create a suitable condition for the tests the two ballasttanks were emptied, which caused a list angle greater than required. The KG was thenlowered by 3 inches to give a list angle close to 19 degrees. This condition issufficiently different to the deep condition to investigate the effect that shipcondition has on the performance of the cross-flooding systems.

Unclassified


Unclassified

5 Simulations5.1 The simulations were conducted mostly in the deep modified condition described in

section 4.8. To define the ship condition in FREDYN a full loading case is not required,although due to the vessel size the main tanks were all modelled and loadedaccordingly. In FREDYN the vessel displacement is calculated from the draft marksinput from data provided by PARAMARINE, with the tanks loaded separately. For thisthe overall ship conditions used are outlined in 5-1. The tanks that contain fluid priorto being damaged are entered into the FREDYN load case so that the correct floodingwill occur after damage.

Ship ConditionShip ConditionShip ConditionShip Condition DisplacementDisplacementDisplacementDisplacement(Tonnes)(Tonnes)(Tonnes)(Tonnes)

KgKgKgKgfluidfluidfluidfluid(m)(m)(m)(m)

GmGmGmGmfluidfluidfluidfluid(m)(m)(m)(m)

Draft APDraft APDraft APDraft AP(m)(m)(m)(m)

Draft FPDraft FPDraft FPDraft FP(m)(m)(m)(m)

Deep (1) 1875 5.41 0.64 4.34 4.20

Light (2) 1716 5.42 0.72 4.38 3.64

Table 5-1: Ship conditions

The simulations were performed as defined in the test matrix above. The zero speedruns had yaw fixed in the simulation so that they could not change heading so thetrue effect of heading on the vessel performance could be realised. At the forwardspeed cases the vessel was started at the correct speed and control made by theautopilot. At the point of damage the RPM was set to zero (in the required runs)manually, as this currently cannot be done automatically with FREDYN. The vessel inthese simulations then drifted to take up whichever heading it naturally wanted tothe waves. This makes these runs more realistic but more difficult to directly compareafter the initial performance. All of the compared runs are conducted in the samewave time history and with damage occurring at the same point of time so that theperformance of the ducts can be directly compared.

Unclassified


Unclassified

6 DiscussionThe dynamic stability simulations were performed as described in Chapter 5. Eachsimulation run produced large data files containing the vessel’s motion and thedamage water levels in each of the flooding compartments. The main statistics weregathered from each run and collated in a spreadsheet. This allowed plotting to beconducted to show trends in each of the run sets of data. These plots are shown inFigures 1 through 65. For the purpose of this report not all of the simulations arediscussed in detail. A selection of time traces of certain runs are also included todemonstrate the performance of the vessel after damage that can be created usingthis method.

6.16.16.16.1 Run Set 1 - Variation with ship speed and headingRun Set 1 - Variation with ship speed and headingRun Set 1 - Variation with ship speed and headingRun Set 1 - Variation with ship speed and heading

Run Set 1 includes runs at 8 headings and 3 speed variations to identify the effect onthe vessel and cross-flooding after damage.

Figures 1 through 3 show an example of the roll angle traces for the 3 duct options(no duct, USCG duct and QinetiQ Duct designs) in a sea state 5 at 7 kts prior todamage and 0 kts after. Figures 4 through 6 show the same plots but with the vesselat 7 kts after damage. These six plots show the how the speed variation effects theroll motion after damage.

Analysis of the time traces allows an easier understanding of how the variables affectthe performance of the vessel. The transient roll angle in this report refers to the firstlarge roll excursion that occurs as the vessel floods. Figures 7 through 9 show thetransient roll angle at the different headings for the 3 duct options (no duct, USCGduct and QinetiQ Duct designs). The three plots show the speed variations, 12 ktsprior to damage and 0 kts after, 7 kts prior to damage 0 kts after and 7 kts before andafter. The first point that can be seen in these three plots is that the no-duct casegenerally has a higher transient at most headings and all speeds. This suggests thatboth the current USCG duct and the QinetiQ duct are operating quickly enough toaffect the initial transient roll. In all but head seas the USCG duct reduces thetransient roll by 5 to 7 degrees and the QinetiQ duct by 1 to 2 degrees more. FromFigures 7 through 9 it is generally shown that the lowest transients occur in head seaand increase with the worst between beam and stern quartering seas.

The roll motion once the vessel has cross-flooded was compared using RMS motionabout the mean heel angle. These runs were not conducted with fixed headings sothe vessel was free to take up whatever heading after the revs were stopped whenthe damage occurred. The vessel often tended towards beam seas after forwardspeed was lost. Plots 10 through 12 show the RMS motion for the three ductarrangements at the 3 speeds tested. All three plots show a general reduction in theRMS roll motion after cross-flooding has completed in comparison with the no-ductcase. The lowest motions for all ducts was at the 7 kts before and 7 kts afterwardsdue to the stabilisers working efficiently to reduce the roll motions. At the forwardspeed both the cross-flooded cases have lower RMS roll after damage compared tothe non-cross-flooding case and are at a lower mean heel angle.

Unclassified


Unclassified

Figures 13 through 21 also show the variation for each of the duct cases plotted foreach speed to show how the speed affects the performance. Figure 18 shows how thetime to cross-flood with the USCG Duct is effected by the heading and speed. Thetime to cross-flood can be seen to vary between 200 to 220 seconds (10%) due toheading and speed change in the same waves. Figure 21 shows that QinetiQ duct alsovaries by 14 seconds but cross-flooding completes in 100 to 114 seconds, which isclose to twice the rate of the current USCG duct. The diameter of the QinetiQ duct is0.25 m, which is smaller than the 0.32 m diameter of the current duct. The differencesoccur due to two factors. The pipe run in the QinetiQ system only contains one slightbend and the pipe run is also shorter. This ensures a much less restricted flow thanthe current design, which is both long and has many tight bends. The second reasonis due to the double duct system flooding through both ducts when all of the ductsare below the water surface, which appears to be the case with this vessel as theducts are low in the vessel. Figures 22 through 24 also highlights this with plots of thetime to cross-flood of the two ducts at the three speeds and the different headings.

6.26.26.26.2 Run Set 2 - Variation with ship loading conditionRun Set 2 - Variation with ship loading conditionRun Set 2 - Variation with ship loading conditionRun Set 2 - Variation with ship loading condition

In Run Set 2 the objective was to identify how the ship condition may change theperformance of the ducts in typical wave conditions. Figures 25 through 27 show theplots of roll against time for the vessel in a light condition in sea state 4. Figures 28through 30 show the same run condition but with the vessel in the deep condition.This shows an example of time history behaviour of the vessel with different shipconditions.

The two ship conditions were tested in sea states 3, 4 and 5 at beam seas (damageopening towards and away from the waves) as well as in head seas at zero speed.Figures 31 through 33 show the transient roll response for the three duct variations inthe deep condition in the three sea states. Similarly Figures 34 through 36 show thesame plots but for a light condition. Comparing the plots of the transients in the seastate 3 shows transient rolls angles of between 22 and 25 degrees for both shipconditions while the ducts appear to make little difference initially on this heel angle.In the sea state 4 and 5 it shows that the ducts do start to reduce the transient roll,with the QinetiQ duct producing the lowest transient rolls in the sea state 5 at bothconditions and at all the three headings. The condition 2 results indicate greatertransient rolls in the three sea states tested when compared with the deep conditiondata. At all the headings both of the ducts show reduced transient rolls due to therapid flooding performance of the ducts. The effect of the opening towards or awayfrom the showed similar motions and the time to cross-flood was the same.

Figures 37 through 39 show the transient roll, RMS roll and mean list anglerespectively for the no-duct case in the deep condition. Each plot shows theperformance of the vessel in sea states 3, 4 and 5. As expected the transient roll andRMS roll (after flooding completed) increases with sea state. Figures 40 through 42show the same plots for the current USCG duct. The same pattern can be seen withthe transient roll and the RMS roll increased with the higher sea states. In comparisonto the no-duct case, Figures 37 through 39, the pattern is similar with the duct casebut with lower transient roll angles.

Unclassified


Unclassified

With the QinetiQ duct, Figures 43 through 45, a similar pattern can be seen as seen tothat from the USCG duct. The transient heel angle does not vary as much as in theother two cases with the angle consistently below 25 degrees.

The RMS roll for the three cases are shown in Figures 46 through 48 for the three seasstates in the deep condition. The light condition is shown in Figures 49 through 51.These plots show that in both conditions the duct cases reduce the RMS roll afterflooding as well as reducing the mean heel angle. The RMS roll of the QinetiQ duct isvery similar to that of the USCG duct case, the QinetiQ duct cross-floods quicker, soproducing a slightly different RMS value after cross-flooding.

Figures 52 through 54 show the time to cross-flood of the USCG and QinetiQ ducts, atboth conditions and the 3 sea states. The time to cross-flood for the USCG duct can beseen to increase slightly as the sea state increases (seen clearer in Figure 30), withhead seas been the worst in each sea state for the time taken to cross-flood.

Like the USCG duct the time to cross-flood is effected by sea condition with the timeto cross-flood increased by 10 seconds in the sea state 5 conditions. The sea state 3and 4 are very similar with a time to cross-flood of 101 seconds to cross-flood. Theheading does not show much effect on the time to cross-flood of the QinetiQ duct asmuch as seen with the USCG duct. The two conditions clearly show only a smalldifference in the time to cross-flood, with the least difference shown in the higher seastates.

6.36.36.36.3 Run Set 3 - Variation with sea stateRun Set 3 - Variation with sea stateRun Set 3 - Variation with sea stateRun Set 3 - Variation with sea state

Run Set 3 was to extend the number of sea states and speeds in beam seas to identifyperformance trends. Figures 55 through 57 show the transient roll for the three casesin a number of sea states up to an extreme sea state 9 that causes capsize in the no-duct case. The zero-speed cases show the increase in transient roll with sea state.Both the duct cases show that the transient roll does not increase as much with theduct operating. The QinetiQ duct produces the least difference in transient roll withthe angle consistently below 25 degrees in all the sea states. At 12 kts before thedamage and 0 kts after damage little difference is shown with the cross-floodingcases but the no-duct case is improved. For the 7 kts before and the 7 kts afterwardsthere is also a reduction in roll in the no-duct case that closes on the performance ofthe two duct cases. This is due to the active fins, which operate effectively at theforward speed and improve the roll performance after damage. Figures 58 through 60show the comparison in the RMS roll response for the different duct arrangements inthe sea states and at the three speeds. The duct cases have consistent lower RMS rollafter damage. The 7 kts before and after damage again show the lowest RMS for all ofthe three cases, showing that the stabiliser fins are operating effectively.

Figures 61 through 63 show the time to cross-flood as a function of sea state for thedifferent speeds and headings. The three figures show that there is some variation incross-flooding time at the different sea conditions, as previously discussed, withvariation less than 20 seconds. This shows that the ducts continue to operate in even

Unclassified


Unclassified

the most severe sea states. The different speeds made very little difference to thetime both the cross-flooding designs took to complete cross-flooding.

6.46.46.46.4 Run Set 4 - Variation with position on wave at damage onsetRun Set 4 - Variation with position on wave at damage onsetRun Set 4 - Variation with position on wave at damage onsetRun Set 4 - Variation with position on wave at damage onset

Run Set 4 was completed to investigate the effect of ship position and point ofdamage on a wave to the transient behaviour. The results are shown in Figures 64and 65. With the point of damage occurring with the ship on top of the wave thelowest transient rolls were produced, but in most cases it made little or no difference.The transient is more greatly affected by the motion at the point of damage ratherthan the point that it occurs on the wave. There was also only 2 to 3 degreesdifference between seed numbers, which randomly changes the waves, showing thatthe transient roll was affected only slightly effected by the size and steepness of theirregular wave where damage occurred. The runs in sets 1 through 3 were all run anddamaged at the same point in the same wave realisation so that better directcomparisons could be made between the runs.

6.56.56.56.5 Results of parametric variationResults of parametric variationResults of parametric variationResults of parametric variation

Overview of the results from the parametric variation studyOverview of the results from the parametric variation studyOverview of the results from the parametric variation studyOverview of the results from the parametric variation studyRunRunRunRunSetSetSetSet ParameterParameterParameterParameter VariationVariationVariationVariation No DuctNo DuctNo DuctNo Duct USCG DuctUSCG DuctUSCG DuctUSCG Duct

AlternativeAlternativeAlternativeAlternativeQinetiQQinetiQQinetiQQinetiQ

DuctDuctDuctDuct

Comment on VariationComment on VariationComment on VariationComment on Variationto Parameterto Parameterto Parameterto Parameter

1 Ship Speed 7 kts Before-0 kts After12 kts Before-0 kts After7 kts Before-7 kts After

Highesttransient roll

The fin stabilisers areeffective in reducingmotions if the ship canmaintain speed afterdamage as compared todead in the water(assuming they are stilloperational).

1 Ship Heading Every 45 degrees Time tocross-floodabout220 sec.

Time tocross-floodabout110 sec.

Time to cross-flood variesby up to 10%.

1 Ship Heading Every 45 degrees Highesttransient roll

Lowesttransient rolland RMS roll

90 and 270 largestvariation in transient roll -180 lowest transientgenerally - for No duct the90 degree seas are worsetransient and RMS thanthe 270 seas.

1 Ship Heading Every 45 degrees Highest RMSroll

Lower RMSroll

Lower RMSroll

Cross-flooded producesbetter RMS Roll at allheadings - Lowest headingfor RMS roll is 180 degrees.

2 Ship LoadingCondition

Deep ConditionLight Condition

Transient roll, RMS roll andtime to cross-flood allincrease slightly as theloading condition movestowards light ship.For all intents andpurposes, however, theeffect is minimal.

Unclassified


Unclassified

RunRunRunRunSetSetSetSet ParameterParameterParameterParameter VariationVariationVariationVariation No DuctNo DuctNo DuctNo Duct USCG DuctUSCG DuctUSCG DuctUSCG Duct

AlternativeAlternativeAlternativeAlternativeQinetiQQinetiQQinetiQQinetiQ

DuctDuctDuctDuct

Comment on VariationComment on VariationComment on VariationComment on Variationto Parameterto Parameterto Parameterto Parameter

2 Ship LoadingCondition

Damage Openingtowards the SeaDamage Opening awayfrom the Sea

Time to cross-flood varieslittle with openingtowards and away fromthe waves.

3 Sea State SS3 to SS9 Highesttransient rolland RMS roll

Lowesttransient rolland RMS roll

Transient roll and RMS rollincrease with increasingSS and the spreadbetween systemsincreases with SS.Time to cross-flood isbasically constant w.r.t. SS.

3 Sea State SS3 to SS10 Very little variation in timeto cross-flood at differentspeeds and sea states.

4 Sea State Damage Openingtowards the SeaDamage Opening awayfrom the Sea

Time to cross-flood varieslittle with openingtowards and away fromthe waves.

4 Wave Position TroughWave SlopeWave Crest

No runs Very little variation in rollresponse betweendamage initiation intrough, on the wave slopeor on the wave crest.

Table 6-1: Results from parametric variation

Unclassified


Unclassified

7 ConclusionsDynamic stability cross-flooding simulations allowed the effectiveness of the cross-flooding ducts to be thoroughly investigated. This time-domain analysis allowed theperformance of the cross-flooding to be assessed and the time taken to cross-flood tobe calculated with the vessel motion taken into account. Previously, the method todetermine time to cross-flood utilised purely static calculations, which assume calmwater for the entire cross-flooding process. These static calculations are suitable forinitial sizing of ducts and give an initial insight into the performance of the cross-flooding but do not provide the complete picture of the performance. The time-domain simulations give a better insight into how the cross-flooding ducts operate atsea and their effect on the vessel during and after damage.

Both the cross-flooding system developed using the guidance in this report and theexisting system fitted on the 270-ft cutter pass the current criteria examined in thisreport. From IMO, the time to cross-flood is less than 15 minutes, where the USCGdefines 5 minutes to cross-flood. The heel angle in the conditions tested also reducedthe mean heel after damage to the order of 10 degrees, which also passes the currentcriteria.

The duct design using the new methodology and time-domain simulation showed atime for cross-flooding almost half that of the current design (see Figure 6), thoughthe pipe diameters were less than that of the current design. Two ducts were used inthis system and in nearly every case both ducts contributed to the counter flooding,which decreased the time to complete cross-flooding. Due to the rapid flooding, theinitial large transient rolls were also reduced in comparison to the no-duct case. Thepipe run of the redesigned duct has only a sight curvature to it thereby allowing asfree flow as possible, unlike the current design that incorporates multiple tight bends.

In the situation of the ship with speed prior to the damage and zero speedafterwards, there appeared to be little or no difference as compared to the zero-speedcase, as the speed was quickly lost and the control of heading was lost. The cases withthe ship continuing on at 7 kts after damage showed an improvement in thetransient and RMS motions after damage. This is due to the anti-roll stabilisersremaining effective and reducing the roll even after damage. The transient roll wasoften seen to be worse in head seas than in beam seas, probably due to the positionof the wave trough at the point of damage. Once past the transient roll, the 7 kts intohead seas case resulted in the lowest RMS roll motion after damage, for the duct andno-duct cases.

The transient roll after damage depended more on the vessel’s response to the waveitself rather than the point on the wave where the damage occurred. An explorationof different damage initiation times within the same seaway showed variations ofonly 4 degrees in the transient roll angle.

The run plan as presented above has shown to provide a suitable test matrix in whichto evaluate the performance of existing and new designs. This ensures that theperformance meets the requirements in wide selection of scenarios.

Unclassified


Unclassified

8 RecommendationsIt has been shown that following this methodology and using a suitable time-domaincode that an effective cross-flooding arrangement can be designed to ensure aneffective operation in all conditions and including all transient effects from the onsetof damage through the point of equilibrium. It is therefore recommended that theguidelines and test plan described in this report should be adopted when consideringfitting or replacing cross-flooding systems in the future.

Even though the existing ducts were not as effective as the new duct design createdusing the described methodology, the existing duct system was seen to improve thevessel performance after damage over the no-duct case. Due to modifications to theloading conditions for the 270-ft WMEC class, this analysis was conducted for moreunstable conditions than the vessel currently operates at. It is recommended thatthese ducts be kept operational as they have been shown to improve the meandamage list angle of the vessel where the initial mean roll angle without cross-flooding is less than 20 degrees. The current duct cross-flooded between 200 and 220seconds in all of the runs tested, which is within the current guidance.

Unclassified


Unclassified

9 References[1] CALHOUN, Capt C. Raymond, USN (ret.). “Typhoon: The Other Enemy - The Third

Fleet and the Pacific Storm of December, 1944”, Naval Institute Press, Annapolis,Maryland, 1981.

[2] DE KAT, J.O. and PETERS, A.J. “Model Experiments an Simulations of a damagedFrigate”, Proceedings of the IMAM 2001 Congress, Crete, 2002.

[3] MARIN. “FREDYN User’s Manual Version 9.0”, MARIN, 2002.

[4] McTAGGART, K. and DE KAT, J.O. “Capsize Risk of Intact Frigates in Irregular Seas”,Transactions SNAME, 2000.

[5] MOD Defence Standard 02-109 (NES 109), UK Ministry of Defence. StabilityStandards for Surface Ships, Part 1, Conventional Ships, 2000.

[6] MOD. “Stability of Surface Ships Part 1 - Conventional Ships”, Sea SystemsPublication No 24, UK Ministry of Defence, Stability Standards for Surface Ships,Part 1, Conventional Ships, 2000.

[7] PETERS, A.J. “Cross-flooding of Frigate Sized Vessels”, March 2001 - Commercial-in-Confidence - USCG and the UKMOD.

[8] SARCHIN, T.H. and GOLDBERG, L.L. “Stability and Buoyancy Criteria for US NavalSurface Ships”, Transactions SNAME, 1962.

[9] SOLAS Consolidated Edition, 2001, International Maritime Organisation,London, 2001.

[10] SOLDA, G.S. “Equalisation of Unsymmetrical Flooding” - Transactions of theRoyal Institution of Naval Architects, 1961.

[11] US Navy, Naval Ship Engineering Center, Design Data Sheet - Stability andBuoyancy of US Naval Surface Ships, DDS 079-1, US Navy, currently Naval SeaSystems Command, Washington, DC, 1 August 1975.

[12] VAN ’t VEER, R. and DE KAT, J.O. “Experimental and Numerical Investigation onProgressive Flooding and Sloshing in Complex Compartment Geometries”,Proceedings of the 7th International Conference on Stability for Ships and OceanVehicles, STAB 2000, Vol. A, Launceston, Tasmania, Feb. 2000, pp. 305-321.

Unclassified


Unclassified

10 TablesSea State Wave Height and PeriodSea State Wave Height and PeriodSea State Wave Height and PeriodSea State Wave Height and Period

Sea State Wave Height (m) Modal WavePeriod (s)

Sea State Three (SS3) 0.88 7.5Sea State Four (SS4) 1.88 8.8Sea State Five (SS5) 3.25 9.7Sea State Six (SS6) 5.00 12.4Sea State Nine(SS9) 20.1 20

Table 1: Sea state wave heights and period

Design Methodology for Cross-flodding Connects on Naval Vessels

Documents

Transcript of Design Methodology for Cross-flodding Connects on Naval Vessels