1

Millennium Class Tanker Structural Design – FromOwner Experience to Shipyard Launching Ways

James Read1, Arne Stenseng2, Rod Hulla3 and Darold Poulin4

ABSTRACT

Three 125,000 DWT double hull tankers are currently under construction for ARCO Marine,Inc. at Litton Avondale Industries in New Orleans, LA. These Millennium Class tankers arebeing built to transport crude oil from Valdez, AK to Cherry Point, WA. The design satisfiesthe requirements of OPA 90 and incorporates a unique structural design philosophy intendedto enhance the structural performance of the vessel. This paper will illustrate how theOwner’s experience with previous vessels in Gulf of Alaska trade is reflected in the structuraldesign of the new ships. The human elements of safety, inspection and maintenance arediscussed and the influence of these factors on the structural arrangement is highlighted. Inconcert with these human factors are structural design improvements that have beenimplemented to specifically address fatigue and stress cracking with the intent of reducingrepair requirements. These topics are presented in a discussion that follows the structuraldesign from concept, through design and analysis and into the construction of the vessels.Lessons learned throughout the process are presented.

1 ARCO Marine, Inc., Long Beach, California2 MCA Engineers, Inc., Costa Mesa, California3 John J. McMullen and Associates, New York, New York4 Litton-Avondale Industries, Avondale, Louisiana

2

FOREWORD

The goal of this paper is to broadly illustrate theprocess by which the structure for the Millennium Classtankers was conceived, designed and constructed. Designand construction of the vessels are addressed from both atechnical and a non-technical viewpoint. The project hasbeen an exercise in philosophy, engineering and projectmanagement, and it is our hope to provide an insight tothe successes and pitfalls on a first-of-class shipconstruction project.

INTRODUCTION

ARCO Marine, Inc. (AMI) is a wholly ownedsubsidiary of the Atlantic Richfield Company (ARCO).AMI is the marine operating company responsible forcarrying ARCO’s crude oil produced on Alaska’s NorthSlope to ARCO’s refineries on the U.S. West Coast.Crude oil is loaded at the terminal in Valdez, Alaska andtransported to refineries in Cherry Point, Washington andLos Angeles, California.

AMI currently operates a fleet of six tankers (fivecrude, one product) ranging in size from 50,000 DWT to265,000 DWT. In 1998 and 1999, the Oil Pollution Actof 1990 (OPA ’90) forced the retirement of three AMI120,000 DWT vessels. These single hulled ships movedcrude oil from Valdez to Cherry Point since the openingof the Trans-Alaska Pipeline Service (TAPS) in 1978.OPA ’90 will also require that AMI retire two 265,000DWT vessels in the year 2000.

In the early 1990’s AMI began to consider optionsthat would replace tonnage lost due to OPA ’90. AMIexplored all options including retrofitting existing tankerswith double hulls, fitting completely new forebodies, andbuilding new vessels. Ultimately the decision was madeto construct new vessels, the Millennium Class Tankers.

The full effort towards preliminary design of theMillennium Class tankers commenced in 1996. In July1997 a contract was signed with Avondale IndustriesShipyard Division, now Litton Avondale Industries.Detailed design commenced immediately and the keelwas laid in May 1998. The first ship, the ARCOENDEAVOUR, is scheduled to deliver in the fourthquarter of 2000. The contract currently provides for atotal of three Millennium Class tankers.

Decision for NewbuildingMaintaining the size of the AMI fleet necessitated

plans for replacement of tonnage phased out by OPA ’90.Initial exploration into possibilities for obtaining double-hulled vessels yielded three options; retrofit of double hullto an existing ship, replacement of a complete forebody,or new building. Retrofit of a double hull into existing

vessels was quickly dismissed. Cargo capacity lost inretrofitting an inner hull to an exiting 120,000 DWTvessel was impractical. It was also understood that theouter hull, while in good structural condition, wouldremain as part of the hull after 20+ years of Alaskan trade.

Forebody replacement presented a more promisingoption than retrofit as it provided completely newstructure in the cargo block. However, while moredesirable from a structural standpoint, the loss in tonnagefrom the previous 120,000 DWT vessels would still beincurred. Lengthening of the vessels was considered butforward visibility limitations and an increased length todepth ratio limited the feasibility of this additionalmodification.

The forebody option would also have required theretention of an inefficient steam propulsion plant. As amajor retrofit, forebody replacement would necessitateinstallation of new cargo control and gauging systems aswell as new deck machinery. Along the same lines, itwould be desirable to upgrade the navigational systems tomore modern bridge equipment. In the end, with safety,efficiency and economic consideration given carefulattention the forebody option was dismissed and thedecision was made to pursue new-build ships.

Design TeamAMI embarked in pursuit of a new design by

bringing together an experienced and highly qualifiedstructural design team. From the early stages AMI beganto work closely with a number of consultants to ensurethat the base of experience was extended as widely aspossible. The main consultants, John J. McMullenAssociates, Inc. (JJMA), MCA Engineers, Inc. (MCA),and Herbert Engineering Corp. served as collaborativedesign consultants throughout the initial phases of theproject.

The intent of this team approach was to draw fromunique skills that each consultant brought to the table.JJMA served as the primary consultant and providedbroad-based naval architecture, marine engineering andcost estimating expertise as well as recent internationalexperience with double-hull tanker design programs.MCA contributed with a strong background in structuraldesign analysis and significant historical experience withvessels in the existing AMI and Alaskan tanker fleet.Herbert Engineering provided expertise in specific issuesrelating to stability, subdivision, and other factors such asballast exchange. This team was responsible for thegeneral development of the Contract Specification andcontinued to work together throughout the design andconstruction phase.

In preparation for contract signing, AMI chose toapproach the final stages of preliminary design as ateaming effort with the consultants and AvondaleIndustries Shipyard Division, now Litton-Avondale

3

Fig. 1 Inboard Profile, Outboard Profile, Bow and Stern Views

Shipyard. At the time of these meetings the concept andpreliminary design had been developed to clearlyrepresent Owner requirements. The goal of teaming withthe Shipyard was to ensure Owner requirements andShipyard capabilities were compatible and fullyincorporated in the detail design.

STRUCTURAL DESIGN PHILOSOPHY

A number of key issues relating to the structuraldesign were identified following the decision to pursueconstruction of new ships. Safety, reliability, provendesign, necessary fatigue life and maintainability were allcited as issues to be addressed in the design process.Strong consideration was given to the fact that the traderoutes served by the AMI crude oil tankers are among themost severe operating environments in the world, andrequire a significant amount of time spent in the waters ofthe Gulf of Alaska. Features specific to AMI were also tobe incorporated into the design based on 20 years ofexperience transporting crude oil in Gulf of Alaskawaters.

The design centered on a belief that designing forprevention is the safest and most economical approach inaddressing the long-term operation of tankers. In short,the primary theme of process was “an ounce of preventionis worth a pound of cure.” The overall intent was not tobuild a ship that just met the Rules of Class. The intentwas to construct a vessel that would meet or exceed theRules as required, to a degree that the vessel becameuniquely suited to meet the Owner’s need for extendedservice in a severe environment. Most importantly, it wasfelt that future problems could be avoided by turning topast experience as a learning tool. Past experience withship structure became a key factor in development of thedetail design [1-4].

Learning from ExperienceContributing greatly to AMI’s understanding of

fatigue cracking and monitoring of ship structure was theUnited States Coast Guard’s Critical Area InspectionProgram (CAIP) [5]. Existing AMI vessels had beenoperating under this program since its inception in 1991.As a result, AMI had developed a clear plan for the“management” of structure on existing vessels.

4

Fig. 2 Upper Deck Plan, In-Tank Plan and Plan on Tank Top

Significant amounts of data have been collected and AMIcontinues to support a proactive, rather than reactive,response to fatigue issues in the structure of its existingships.

Historically, AMI’s database of experience withstructural details extends back through 1987. Locations,lengths and types of fractures were documented. Patternsof structural fatigue cracking were clearly identified andproblems were remedied with the development ofstructural fixes through finite element analysis.Participation in this program has undeniably led to greaterunderstanding of designing for fatigue in ship structure.

Also helpful in determining potential problem areaswere the various diverse experiences of the design teammembers. This experience combined with informationalresources published by the industry led to a betterunderstanding of the past performance of a variety ofstructural arrangements. Particularly noteworthy amongpublished data are publications by the Tanker StructuralCooperative Forum (TSCF) [6,7], Oil CompaniesInternational Marine Forum (OCIMF) [8] and Lloyd’sRegister [9].

“Ground Rules”For the Millennium Class tankers, the structural

design was to be developed based on extensiveengineering. AMI required that the ships be designed fora 30-year fatigue life. This included not only compliancewith ABS’ SafeHull A and B programs, but also finiteelement analysis, structural fatigue analysis andcompliance with ABS’ Dynamic Loading Approach(DLA) [10,11]. During concept design preliminarystructural details were studied extensively in areashistorically known to experience structural fatiguecracking. This preliminary analysis was completed priorto contract signing.

Based on AMI’s Alaskan operating experiences, anumber of design criteria were established for thestructure. The vessels were to be built primarily withmild steel, with little or no reduction in scantlings. Wherehigh strength steel was used margins above regulatoryallowable minimums were typically applied. Criticalstructural areas were studied extensively by finite elementanalysis, dynamic load analysis, and spectral fatigueanalysis. This conservative approach led to a number ofdecisions that resulted in a rugged yet functional design.

5

Addressing Fatigue in Structural DesignOne of the most significant conceptual efforts made

in the design was to address fatigue in structural details.Clearly, it is acknowledged that design and analysiscannot yield a zero crack condition. However, duediligence and attention to known problems can yield agreater understanding of those structural details to avoidand those details to incorporate. In some cases, detailsthat performed poorly in previous vessels (e.g. mushroomshaped cutouts at erection joints) were redesigned foracceptable performance in new vessels.

The trade route to be traveled by the MillenniumClass tankers presents a somewhat unique fatigue lifeprofile. Nearly all of the at-sea operation of the vesselswill be in North Pacific waters. Port time spent duringloading and discharge is minimal and each ship willaverage a round trip about every 10.5 days. To illustratethe exclusivity of trade and frequency of voyages, the120,000 DWT vessels recently retired each completedover 600 voyages to Valdez in approximately 20 years ofoperation.

CONCEPT DESIGN

Size of VesselThe decisive commitment to build new vessels turned

the focus towards clear definition of a design philosophyunder which the vessels would be built [12]. Size of thevessel was the first issue to be addressed. The primarytrade route for the proposed Millennium Class tankers isto be between Valdez, Alaska and Cherry Point,Washington. Because this trade extends into Puget Sound,Washington a deadweight limit of 125,000 DWT ismandated. This restriction, imposed by U.S. Coast Guardregulations excludes tankers of greater than 125,000DWT from delivering oil to U.S. ports in Puget Sound,Washington [13].

While USCG regulations will ultimately govern thedeadweight of ships entering into Puget Sound the overallcargo capacity of the Millennium Class tankers was set atone million barrels. This corresponds to roughly 137,000DWT of Alaskan North Slope crude oil. The choice wasmade to proceed with a vessel of this size in an effort toenhance its overall trading versatility during the planned30-year service life. Thus, the vessels have been designedto a scantling deadweight of 137,000 MT but will operateat a deadweight of 125,000 LT when delivering crude oilinto the Puget Sound. Table 1 provides further PrincipalCharacteristic information.

Cargo Block Structural ConfigurationThe overall vessel and tank configurations are shown

in Figures 1 and 2. The cargo tanks are arranged six tankslong by two tanks across with two slop tanks located justaft of the No. 6 cargo tanks. The ballast tank arrangement

includes six pairs of J-tanks with the No. 6 tanksextending aft beneath the slop tanks. It was found thatthis arrangement provided an optimum operationalconfiguration in terms of intact and damaged stability.

The Midship Section is shown in Figure 3.Transverse web frame spacing is 3.96 meters and typicallongitudinal spacing is 862 millimeters. Web framestructure is balanced on either side of the centerlinebulkhead with longitudinal stiffening on the starboardside of the bulkhead. Main longitudinal girders are thecenterline vertical keel and girders port and starboardbeneath the lower hopper radius. Two horizontal stringerssupport the transverse bulkheads in the cargo block andare shown in Figure 4. Five longitudinal stringers areincluded in the ballast tank wing walls and the second andfourth stringers coincide with the cargo tank horizontalstringers.

Of special note is that a continuous, watertight, fullheight, centerline bulkhead runs below the Upper Deck(strength deck) from the collision bulkhead to thetransom. The sole exception to the tightness of thisbulkhead is found at the athwartship access in the cargopump room. Otherwise, the centerline bulkhead dividesthe cargo block into the port and starboard ballast andcargo tanks and the engine room machinery spaces intotwo completely separate and redundant engine rooms.

The after portion of the centerline bulkhead,watertight and insulated to an A-60 standard, is alsocontinuous in the machinery spaces and engine roomcasing above the Upper Deck and continues upward to thetop of the stack. Three sliding, watertight A-60 doors,normally closed, are found in the machinery spaces andare available for athwartship access. The continuity of thecenterline bulkhead continues forward of the collisionbulkhead to the stem but is not watertight in this area.

Table 1 Principal Characteristics

Length, Overall 272.69 mLength, Between Perpendiculars 258.16 mBeam, Molded 46.20 mDepth, Molded (at Side) 25.30 mDepth, Molded (at Centerline) 26.30 mDisplacement, Design 159,100 MTDraft, Design 16.31 mDraft, Scantling 17.50 mLightship Weight 33,188 MTDisplacement, Design 160,778 MTDisplacement, Scantling 173,310 MTCargo Capacity, Design 899,757 bblsCargo Capacity, Scantling 978,032 bblsGross Tonnage 85,093 MTNet Tonnage 36,299 MTBlock Coefficient, Design Draft 0.830Design Speed 16.5 knots

6

Fig. 3 Typical Midship Section

“Hopper” DesignA significant research effort was made during the

preliminary design to explore possible variations on the“hopper” design. The “hopper” is the area of slopedplating at the outboard lower corner of the cargo tanks. Itwas known that the most critical area of design in thisconfiguration is found in the lower hopper transition.

Options for the connection at the lower hopperincluded use of a cruciform weldment, a cruciformcasting, continuation of the inner bottom plating past thehopper to the side shell, or a radius hopper platesupported by a longitudinal girder. The final decision wasto incorporate a radius plate at the corners of the hopper.Rationale for this decision included finite elementanalysis, past structural experience documented in theTSCF and Lloyd’s publications, ability to holdconstruction tolerances, and overall stiffeningarrangements.

Forebody and Afterbody StructureThe innerbottom plating and the five-stringers in the

ballast tanks transition into the flats in the engine room.These levels also extend forward into the bow thruster

and forepeak areas. The tank top structure carries aft asthe flats of the Pump Room and the lower Engine Room.The major stringer at 9840 mm above baseline (ABL)transitions into the upper Engine Room and PurifierRooms. The major stringer at 17120 mm ABL carries aftand forms the machinery flat for much of the auxiliarymachinery, the Engine Room workshop and a large spareparts storage area. Finally, the ballast tank stringer at20760 mm ABL slopes upward approximately one meterand forms the flats in the port and starboard MachineryControl Rooms.

HUMAN FACTORS – SAFETY, INSPECTION, ANDMAINTENANCE

Special attention was given to the arrangement ofstructure as it pertains to the human element in design.Specific issues that were addressed in the cargo blockincluded emergency removal of personnel, means ofstandard access and egress from tanks, ability toadequately ventilate tanks, and access for tank inspection

7

and maintenance. In the machinery room considerationwas given to equipment removal and maintenance.

Fig. 4 Transverse Bulkhead Stringer at Midship

Safe Access and VentilationFeatures included in the ballast tank design allow for

safe access and proper ventilation during tank entryprocedures. IMO and SOLAS guidelines for access intanks were rigorously applied as were ASTM standardpractices regarding ladder and platform construction [14-16]. Inclined ladders are located at the forward outboardcorners of the ballast tanks and provide primary access.A vertical access trunk with a vertical ladder is located atthe aft inboard corner of the tank. Finally, at the aftoutboard corner of the tank is a second 600 mm x 800 mmclear opening for personnel and stretcher removal. Thisopening extends through each stinger level to the bottomof the tank.

Cross-ventilation is to be provided by locatingportable fans at the vertical trunk or the personnelremoval hatches. An additional provision for ventilationis included with the attachment of a ventilation fan to abranch line on the ballast main that may be used toventilate back through the suction/fill lines into the tanks.Bolted tank cleaning openings are provided in every otherframe bay to facilitate cleaning of sediment from thetanks or, in the event of a leakage from the cargo tanksinto the ballast tanks, the cleaning of oil from the tank viaportable tank washing machines.

Access into cargo tanks is provided via a hatch at theaft end of the tank. A vertical ladder provides accessdown to the first intermediate platform. This ladder ishinged so that it may be moved clear of a 600 mm x 800mm clear drop opening to the bottom of the tank.Inclined ladders extend from the intermediate platformdown to both horizontal stringers and to the bottom of thetank. Handrails are provided on each stringer as requiredby IMO guidance.

Finally, a bolted manhole is located in the forwardand aft tight bulkhead of each cargo and ballast tank inthe cargo block. The manhole consists of a 24 inch pipewith a 150# bolted blind flange that is hinged forhandling. The manholes, which would be opened onlyduring shipyard repair period, provide two additionalmeans of access and egress from each tank.

Cleaning of Cargo TanksCleaning of cargo tanks is benefited by the inclusion

of most major longitudinal structure within the ballasttanks. The result is “clean” bulkheads in the cargo tanks.Longitudinal structure in the tanks is found at the upperdeck and the starboard side of the centerline bulkhead,which carries the centerline longitudinal structure. The aftbulkheads have two horizontal stringers and carry thevertical transverse bulkhead. This arrangementeffectively results in four clean surfaces in the port cargotanks and three clean surfaces in the starboard cargotanks.

It is worthy of note than the IMO standard forcoverage was enhanced in the ship specification tofacilitate safer tank entry and to assist in cleanup forshipyard repairs [17]. This higher standard requireddirect impingement from the crude oil wash machines on95% of the primary and the secondary structure (i.e.bulkhead longitudinals and vertical bulkhead stiffeners).Three top and two bottom machines in the starboard sidetanks and three top and one bottom machine on the portside tanks were used to meet these criteria.

Access and Maintenance in Machinery SpacesSpecial efforts were made to include provisions in the

design that facilitated maintenance and removal ofequipment in the engine room and machinery spaces.Clear vertical accesses to the exterior are providedthrough the decks on the starboard side of the PumpRoom access trunk and in both Engine Rooms. A ten-tonstores crane located on the exterior “B”-Deck services theport side Engine Room access. A five-ton air hoistservices the starboard side Engine Room access.Underdeck bridge cranes service each of the two mainengines and are capable of transporting equipment tocommon lifting points at the vertical accesses.

An extensive monorail system in each Engine Roomis capable of reaching most auxiliary equipment andtransporting it to the port side Engine Room Workshopfor servicing or to the vertical accesses for removal fromthe ship. Bolted equipment removal plates on the 17120mm ABL machinery room flat allow access to the cargopump motors located on the 9840 mm ABL engine roomflat. Again, the monorail system allows for removal ofthese eight-ton motors from the port side of the ship. Awatertight, A-60 insulated and normally-closed boltedequipment removal plate (BERP) is located in the Engine

8

Room Workshop to facilitate the athwartship movementof large equipment across the ship.

PRELIMINARY DESIGN

The overall scantling arrangement and MidshipSection were developed and analyzed using ABSSafeHull and separate coarse meshed finite element

models of the midship section. The separate finiteelement analysis telescoped from global to intermediateand then to general local models.

Nine Critical Details AddressedThe main objective in analyzing critical structural

details was to identify and implement any potentiallylarge structural changes prior to turnover of the design to

Fig. 5 Nine Local Details in Preliminary Analysis (Top row, l to r: Lower Hopper, Upper Hopper,Stringer at Centerline Bulkhead; Middle Row, l to r: Bottom Longitudinal, Transverse WebFrame at Tank Top, Double Bottom at Transverse Bulkhead; Bottom Row, l to r: HorizontalStringer at Transverse Bulkhead, Inner and Outer Shell at Transverse Bulkhead, Inner andOuter Shell at Web Frame)

9

the Shipyard. Further, the owner also wanted to havetimely and complete control in selecting the sensitivestructural details to be used repeatedly throughout thecargo block. This is again a reflection of extensiveexperience with existing ships operating in the harshTAPS trade environment, which has demonstrated that thedesign of the local details ultimately determines theprobability of premature fractures. Concurrent withconcept design, numerous design variations wereanalyzed in common structural details to optimize andimprove their structural behavior.

Nine local details found in the midship section wereselected and analyzed using detailed solid element mesh.The mesh size was on the order of the plate thickness inthe critical regions. The nine details are shown in Figure5 and cover typical problematic construction details. Themodels were analyzed using eight initial design loadconditions. These consisted of four loading conditions:Full Cargo, Ballast and two “Checkerboard” loadingconditions. The Full Cargo and Ballast loadingconditions were analyzed in still water, head-sea designhog and head-sea design sag waves, while the twoCheckerboard conditions were analyzed in still wateronly. The design wave used had a length equal to theship’s LBP and a wave height of 8.05 meters.

Results of Preliminary AnalysisThe preliminary analysis revealed the existence of

high general stresses in the two horizontal stringers whensubjected to the checkerboard loading conditions. Thestresses were too high to be eliminated by localreinforcement, and the need for modest changes to thestringer design was identified. The modification requiredadditional and symmetric transition structure aft of thetransverse bulkhead. Forward of the transverse bulkheadthe stringer transition was softened at the centerline andlongitudinal bulkheads and extended forward by oneadditional frame.

The results also included the identification of twoalternative side shell longitudinal to frame connections.The first involved the use of an offset panel stiffener witha sophisticated web frame collar design. The secondinvolved the use of soft-toed brackets butt-welded to thebulb-section of the side shell longitudinal on both sides ofthe frame. The latter of the two details was chosen foruse on the vessel based on construction preference.

Other recommendations included improvements tothe lower hopper design, including rearrangement of thehopper longitudinal, a reduction of the hopper radius andincreased web frame thickness in way of the hopper. Itwas found that the radius plating in the hopper design,when subjected to a sagging condition and without propersupport, tended to “shrink” similar to the radialcontraction of a pipe subject to stretching. The

deformation in the hopper plating was restricted at theweb frame thereby creating high stresses in the weldedconnection of the web frame to the hopper plating.

Also noteworthy in the findings was the need foradditional support to improve the effectiveness of thecurved sections of the web frame faceplates. Withoutadditional bracket and flatbar support the curved sectionof the web frame faceplates become ineffective, resultingin unacceptable stresses in the web frame and flange atthese locations.

Use of Mild vs. High Tensile SteelThe rationale for the decision to tend towards use of

mild steel was based on historical problems encounteredwith the liberal usage of high tensile steel in ships. In the1970’s and 1980’s the movement of the shipbuildingindustry to look towards high strength steel as a means forreducing the overall weight, and therefore cost, of thevessel created many problems. Credit was taken in hullscantlings as high tensile steel allowed higher allowablestresses. In many cases, particularly for ships operating inless severe trade routes, this was acceptable. However,vessels frequently subjected to a harsh environmentresulting in moderate or high cyclic stresses, developedfatigue cracks in the structural transition details.

In a distinct effort to avoid chronic fatigue problemsit was decided that mild steel (ABS Grade A) would beused for all structure except in areas which specificallyrequired high tensile steel. If high tensile steel were to beused, a reduction in scantling would be allowed,providing there was a minimum ten percent allowanceabove the ABS requirements, including corrosionallowance. For the upper deck, where the inherently lowneutral axis of double hull construction results in mildsteel plate thickness well above 25 mm, the deck isconstructed of ABS Grade DH plate. However, by usingGrade DH plate, ABS rules allow for a plate thickness of19.5 mm, including corrosion allowance, but 22 mm platewas used. A second example of the use of high tensilesteel is in the horizontal bulkhead stringers, which aresubjected to high static stress but relatively low cyclicstress.

Figure 6 illustrates the application of these criteria tothe design. ABS Rule minimum scantlings are shown inparentheses with the as-built plate scantlings shownoutside the parentheses.

Fixed Range for ScantlingsFor structure in the cargo block, in fuel oil tanks, and

in miscellaneous ballast tanks a “rule of thumb” was setfor determination of minimum and maximum scantlings.Structural steel was limited to a minimum of 12 mm inthickness and was to be no greater than 25 mm inthickness. In the case of the minimum scantling,

10

Fig. 6 Transverse Bulkhead Arrangement Showing Scantlings (l), Typical Bulkhead Stiffener (r)

experience showed that steel less than 12 mm provedsensitive to corrosion. For the maximum scantling, steelgreater than 25 mm reduced the benefit of the steel’sductile behavior and thereby reduced the inherentredundancy of the structure [21]. Thick plates create a 3-D stress field at a potential crack tip, making the steelmore brittle. From a maintenance standpoint, therelatively limited availability of steels thicker than 25 mmcan problematic in repair yards.

These decisions were made based on past experiencein vessel operation and with extensive preliminary finiteelement analyses complete. Ultimately, as designprogressed, it was found that in only a very limitednumber of cases would material less than 12 mm havebeen of benefit or met the fatigue life requirements.

Finalization of Preliminary DesignThe preliminary phase in the design process was

closed with JJMA as the lead technical coordinator. Thiseffort included modifying and updating the conceptdesign Midship Section and Scantling Plans.Modifications were based on the findings of the structuralfinite element analyses and included additional changesdesired by the Owner. The finalization of this phase alsoincluded extensive review of the complete concept designwith all consultants present. This combined effort

allowed all participants to benefit from the extensiveexperience of the group as a whole.

Prior to contract a fully developed set of lines, aMidship Section and a General Arrangement were turnedover to Avondale. An overall teaming effort was initiatedat this point in time. The Midship Section and apreliminary Scantling Plan were modified slightly andtailored to meet specific requirements demanded byAvondale’s production scheme. These modificationswere small in nature and generally were the result of platestraking and erection butt location. All modificationswere completed with the approval of AMI and theconsulting team.

SafeHullFollowing contract signing, work quickly advanced

on design issues related to Class approval of the structuralarrangements. These activities included the running ofABS’ SafeHull Phase A and Phase B. Because muchinitial work had been completed on the structural design,SafeHull served almost exclusively as a verification toolrather than as a design tool. The design was based ongood engineering practices and did not rely on SafeHullas a tool to find minimum allowable scantlings.

In Phase A, the rule-based initial scantlings wereverified and shown to satisfy the SafeHull prescribed loadstrength and fatigue requirements. In Phase B, strength

11

assessment of global and local structure was verifiedusing SafeHull imposed loads. Here, global structure wasaddressed as a coarse mesh, three-tank model. Localstructure was addressed as two-dimensional fine meshmodels.

Limited modification to the structure was necessarybased on the results of SafeHull. Plate scantlings at thetoes of the cargo tank horizontal stringers required someincrease, as did the local longitudinal bulkhead in way ofthese toes. Panel breakers were necessary in someinstances to meet small panel buckling criteria. Allnecessary modifications were made and the resultantchanges to structure were turned over to MCA forincorporation into the DLA analysis.

DYNAMIC LOAD APPROACH (DLA)

Analytic ApproachConcurrent with the Shipyard effort to run SafeHull,

MCA Engineers commenced work on the ABS’prescribed Dynamic Load Approach (DLA). The DLAanalysis was based upon standard ABS procedure, butusing MCA-developed finite element models andSPLASH-generated wave-induced pressure profiles.SPLASH is a 3-D panel code for computing inviscid,irrotational potential flows [18]. This procedure includedthe following steps:

1. Selection of four worst-case loading conditions.Loading cases selected and shown in Table 2 were:• Alaskan Crude, typical full load condition.• Heavy Ballast, typical ballast condition• Arrival in San Francisco, a partial loaded

condition• Checkerboard #6 SWB, a checkerboard loading

condition per ABS.

Fig. 7 Global FEA Model 2. Calculation of the long-term extreme value of the

Dominant-Load-Parameters (DLPs).This was accomplished using the ABS-specified H-family weather spectra and SCORES II-derivedFrequency Response Functions. SCORES II is a striptheory ship motion program used by ABS for theDLA analysis [19]. Four Dominant-Load-Parameterswere calculated per ABS guidance; hull girderbending moment, roll angle, vertical accelerationforward, and lateral acceleration forward.

3. Calculation of the equivalent waves for each loadcase and Dominant-Load-Parameter combination.These were determined by dividing the long-termDominant-Load-Parameter extreme values by theSPLASH-generated Frequency Response Functions todetermine the equivalent wave heights. TheSPLASH-calculated Frequency Response Functionswere generated based on the equivalent wave systems(headings and frequencies) derived by the SCORES IIresults. The external wave pressure and internal tankpressure profiles were then multiplied by this scalingfactor to obtain the finite element model loadingprofile. The results of this phase are shown in Table3.

Table 2 - Dominant Load Parameters

Case # Load Case

Wave-inducedBending Moment

[MT-m]

Roll[degrees]

Verticalacceleration

[m/s^2]

Lateralacceleration

[m/s^2] 14 125K Alaskan Crude 656350 20.2 5.463 2.054 27 Heavy Ballast 586660 28.3 6.791 2.724 B Arrival S.F. #2 577820 21.1 5.809 2.402 F Checkerboard 573850 23.6 5.880 2.164

12

Table 3 - Equivalent Wave Data

Equivalent WaveCase # Load Case DLP Heading Period (s) Height (m) Length (m)

14 125K Alaskan Crude Moment 180 13.96 8.698 304.28114 125K Alaskan Crude Vert. Acc. 180 13.96 8.212 304.28114 125K Alaskan Crude Lateral Acc. 105 8.38 5.507 109.54114 125K Alaskan Crude Roll 75 15.71 7.718 385.10627 Heavy Ballast Moment 180 12.57 11.402 246.46827 Heavy Ballast Vert. Acc. 120 11.42 6.961 203.69327 Heavy Ballast Lateral Acc. 105 9.67 7.253 145.83927 Heavy Ballast Roll 90 13.96 7.069 304.281B Arrival S.F. #2 Moment 180 12.57 7.721 246.468B Arrival S.F. #2 Vert. Acc. 120 11.42 7.271 203.693B Arrival S.F. #2 Lateral Acc. 105 8.38 6.279 109.514B Arrival S.F. #2 Roll 75 15.71 8.215 385.106F Checkerboard Moment 180 13.96 6.648 304.281F Checkerboard Vert. Acc. 180 13.96 7.565 304.281F Checkerboard Lateral Acc. 105 9.67 6.490 145.839F Checkerboard Roll 105 15.71 8.851 385.106

4. Development of numerous finite element models forthe DLA analysis.The models developed included:

• A 3-D stem-to-stern global model for overallDLA stress assessment (Figure 7).

• A 3-D intermediate model of the amidshipsstructure.

• 3-D detailed general local model of the typicalmidship structure at frames.

• 3-D detailed general local model of the upperportion of a transverse bulkhead structureincluding the two stringers (Figure 9 and 11).

• 3-D detailed general local model of the lowerportion of a transverse bulkhead (Figure 8).

• A 2-D refined mesh model of the aft centerlinebulkhead between Frame 33 and 73 to examinehigh shear stresses observed in the 3-D globalmodel (Figure 12).

Scantlings were revised as necessary until allstructure met the maximum allowable ABS Von Misesstress limit of 95% of yield. Note that since this analysisincluded detailed 3-D models of the complete midshipstructures, the general stress field was included in alltypical structural components. For example, in additionto seeing the stresses in a 2-D web frame, the generalstress levels in the frame brackets and the brackets tolongitudinal connections were included. Anotherexample would be the deck stresses; instead of looking

only at the nominal deck plating stress the more detailedlocal models captured the additional stress increasecreated at the longitudinal-to-bracket connections atframes and bulkheads (Figure 9).

Fig. 8 General Local Model at Transverse Bulkhead

Appropriate judgment was used when evaluating thedifferent stress levels. Judgement factors includedconsideration of element size, element shape and themodeling of structural discontinuities. Higher peakstresses are expected in a more detailed and finely meshedmodel. Hence stress results larger than 95% of yield werein some cases considered acceptable in finely meshedmodels. The structure was also analyzed for bucklingwithout resulting in any significant changes. In all cases,the results were submitted to ABS for specific approval.

13

Fig. 9 Upper Shear Strake (Plating and BulkheadRemoved)

Fig. 10 Initial Centerline Bulkhead Design, Sag Wave

DLA ResultsThe following modifications were shown in the

analysis to be effective and were implemented in thedesign:

• The centerline bulkhead thickness was increased atseveral locations. Initially large pump-room cutoutsin the centerline bulkhead were reduced in size andmodified in shape (Figure 10).

• The depth and thickness of forepeak frames and oneforepeak stringer were increased.

• The thickness of the lower transverse bulkheadstringers was increased locally at the connection ofthe toes to the longitudinal wing bulkhead. Thismodification took place both forward and aft of thebulkheads. The panel stiffener arrangements weremodified at both the upper and lower stringers.

• Two additional brackets were included between thevertical bulkhead stiffeners and the sloped hopperplating. The thickness was increased for all fourbrackets at this attachment.

• The vertical bulkhead stiffeners above the topstringer were modified, as was the attachment to thedeck longitudinals. The depth of each stiffenerslopes from 693 mm at the upper stringer to 280 mmat the deck longitudinal. The soft-toed bracketattaching the stiffener and the deck longitudinal wasmade smaller than the original design (Figure 6).

All DLA models were also rerun using elementstresses instead of unaveraged nodal stresses, per ABSrequest. The results showed that all elements in the finalglobal and intermediate models passed the 95%-of-yieldcriteria. The 3-D general local models showed a limitednumber of locations (elements) which did exceed the95%-of-yield criteria. The upper most side shell girderand the double bottom girder exceeded the yield criteriain way of the access openings included in this model(Figure 9). Hull girder bending combined with themodeled cutouts stress concentration drives the stresses inthese girders. The nominal stresses in the girders are infact significantly lower than the nominal stress in thedeck, which itself is significantly lower stressed than whatis required by the ABS rules.

The Checkerboard load case combined with itscorresponding maximum roll conditions created amaximum stress of 105% of yield in the triangularbrackets at the base of the transverse bulkhead verticalstiffeners. The checkerboard loading pattern is not atypical condition and combining this in port leak-testcondition with a long term extreme roll condition isconservative. The final design of this bracket had passeda detailed finite element analyses using solid elements andwas subjected to both a still water checkerboard loadingand an extensive fatigue analysis as described below.

Fig. 11 Horizontal Stringers, Checkerboard Loading,Bulkheads Removed

The same combination of checkerboard loading andlong-term extreme roll created an above yield stress in

14

one element at the upper horizontal stringer (Figure 11).If such a combination of in port loading condition andextreme wave load were to occur it could lead to yieldingof one localized region of the stringer. However, theremainder of the structure can absorb any load that maybe shed. Since the normal loading conditions and cyclicloads create only low stresses at this location the designwas deemed acceptable without further changes.

Fig. 12 Initial (r) and Final Pumproom (l) OpeningModels

The 2-D model of the centerline bulkhead at thepump room showed above yield stresses at the finemeshed regions around the cutouts. These stress levelswere deemed acceptable since they were induced by thefine mesh around cutouts rather than by the nominalstresses (Figure 12). Also, beam elements that were usedin the 2-D model do not properly represent the beneficialeffect of the flange coaming that is used as reinforcementaround these cutouts.

No additional changes to the design wereimplemented nor required due to the evaluation of theelement stresses. This is primarily a result of having usedthe more conservative unaveraged nodal stresses for theinitial DLA submittal. Note that using the average nodalstresses, which is the default of many finite elementpackages and typically used by ABS for illustrationpurposes, is not a conservative approach. It will in manyinstances significantly reduce or hide high stress regions.

Importance of Model Test ProgramA rigorous model test program was undertaken at

SSPA in Gothenburg, Sweden prior to contract signing.As a result the development of a hull form was effectivelycomplete when negotiations with the shipyard began.This proved advantageous as it allowed AMI and theShipyard to immediately concentrate on structural designand resulting machinery arrangements. Foreseeing thisbenefit, during model tests, the seakeeping models wereinstrumented to later aid designers in correlating the DLAmotions model to actual measurements. Comparison ofthe motions and extensive side shell pressuremeasurements has been documented in other publications[20].

Of note is that there was intent to determine ifstructural modifications were necessary based on resultsof a hydrodynamic model test program. The emphasisplaced on redundancy in the overall design (e.g. the shiphas two rudders, two propellers, two engine rooms etc.)demanded that relatively uncommon features were presentin the hull form. For example, modifications were foundnecessary in the twin skeg arrangement that was chosenfor the afterbody configuration. The area between theskegs was modified from a flat bottom to a V-shapedbottom to reduce aft slamming in following seas.Likewise, forepeak structure was later modified toaccommodate resultant pressures found in the relativelyflared areas of the bow structure.

FATIGUE ANALYSIS

Analytic ApproachThe fatigue procedure used on the Millennium Class

tankers has been developed by MCA in a continuallyevolving process that combines theory, finite elementanalysis techniques, and dynamic 3-dimensional loadgeneration. The strength of the analytical process hasbeen its consistent verification through actual fieldexperience. Identifying, analyzing, and developingeffective repairs for structures on a number of AMI’sexisting ships has successfully proved the fatigueprocedure. The analytical process includes the followingkey steps:

1. Construction of global, intermediate, general localand local detail models capable of capturing andapplying hull girder responses into small details suchas rat-holes and bracket toes

2. Generation of dynamic 3-dimensional loads for amatrix of wavelengths, angles, and ship operatingconditions using the SPLASH 3-dimensional CFDcode

3. Analysis of the finite element models for the matrixof SPLASH loads to develop structural element RAOcurves

4. Calculation of the fatigue crack initiation life ofdetail structures using the stress RAO’s, anticipatedship voyage/weather data, and published S-N fatiguecurves

Telescoping Finite Element AnalysisThe Millennium Class tanker design was analyzed

using a telescoping process that captures global andregional structural performance and applies it to thestructural detail of interest. The process allows for

15

examination of details as small as rat-holes, lap-weldedcollars, and bracket toes.

The same Global, Intermediate and General Localfinite element models used in the DLA analysis were alsoused in the fatigue analysis. In addition to the modifiedversions of six of the nine local models used in the initialdesign, three new local models were analyzed. Theseincluded bottom longitudinal to frame connections inhopper, deck longitudinals at frames and decklongitudinals at bulkheads.

These models are the final step in the telescopingfinite element analysis process. They are constructedwith general-local plate mesh along the boundaries, whichis then transitioned to a mesh of greater density in the areaof interest, using 8-noded brick (solid) elements with edgelengths as small as the plate thickness. Weld contourshave been simulated using either 6-noded bricks or 8-noded bricks meshed along the weld contour to accountfor their geometric effect.

SPLASH-Generated LoadsThe fatigue analysis used for this project is a fully

three-dimensional dynamic spectral analysis. The loadcases are generated using the three dimensional free-surface computational fluid dynamics code SPLASH.

Unsteady flows are treated in the frequency domain,as a linear small-disturbance, harmonically oscillatingperturbation to the steady basis flow. First-order unsteadyflow predictions yield unsteady ship forces and motions.Flow solutions are fully 3-D, with six degree-of-freedomship motions, and arbitrary incident wave-heading angle.The unsteady flow is computed using a steady flow panelmodel with the hull and free-surface panels fixed, whilemovement is simulated via transfer of unsteady boundaryconditions to the steady time-averaged panel location.Forces and moments are computed by integration of panelpressures over the vehicle surface, plus panel-basedwaterline integral contributions. The latter account forthe oscillating area at the waterline due to ship and free-surface motions.

The SPLASH calculations were performed by SouthBay Simulations, resulting in motions and pressureprofiles for a matrix of waves. The hull was analyzedwith nonlinear roll damping derived from Ship MotionProgram (SMP) runs, at an average speed of 15 knots[24]. The analytical matrix (384 spectral fatigue loadcases) included the following:

• Eight wavelengths, from 1/4 to 2 times the ship’slength

• Seven wave-angles from ahead to astern in 300

increments. Since “windward” and “leeward” effectsare not symmetric, off-centerline models required 12wave angles (full 3600)

• Two loading conditions - Full Load and Ballast

• The real and imaginary cyclic stress components (twoinstances in time are used to determine the completecyclic variation, assuming sinusoidal loads andresponses)

Spectral Fatigue AnalysisFatigue can be divided into two stages: an initiation

phase (calculated here using Miner’s Rule) and a growthphase (calculated using fracture mechanics), until thecrack reaches a critical length and propagates in a brittlemode. Crack propagation analysis was not completed forthis effort, as the objective was to minimize or eliminatecrack initiation [21]. The crack initiation phase iscalculated in the following steps:

1. The finite element models are analyzed for everySPLASH run, starting with the global model andtelescoping down to the detail local models. Thehydrodynamic pressure fields are mapped onto thefinite element models using a bi-linear interpolationprogram. The 6-DOF calculated accelerations areapplied to the light ship weight, and are used toderive the fluctuating internal tank pressures. Onceparent model analysis is complete, the boundaryreaction forces are transferred to the telescopedmodel, and pressures and accelerations applied to theapplicable contained elements.

2. The real and imaginary stress results are vector-added(square root of the sum of the squares) to calculatethe cyclic stress component for a given loadcondition, wavelength, and heading. By analyzingthe finite element models for all loading cases andnormalizing to wave height, stress RAO’s aredeveloped for all free surfaces of every solid elementin the detailed models.

3. The weather spectra, taken from a hindcast studybased on historical hindcast weather data (NavyGSOWM) along zones in the TAPS trade route arecombined with the stress RAO’s to produce responsespectra. This directional weather data combined withan assumed trip profile of travel between Valdez,Alaska and Cherry Point, Washington three times permonth determined the weather spectra.

4. The response spectra are compared with published S-N curves to calculate Cumulative Damage Ratios(CDR’s), an integrated ratio of the actual fatiguecycles at a given stress divided by the allowablenumber (defined by the S-N curves).

5. The CDR’s are integrated across the ship voyageprofiles to obtain a fatigue crack initiation prediction.

16

SN-CurvesA number of SN curves have been experimentally

determined by the United Kingdom Department ofEnergy, and are used as an industry standard [25]. Curvessuch as F and F2 apply to calculations based on nominalstress fields. Since the models are built and analyzed tothe “hot-spot” detailed level, the appropriate curves arethe ‘C” and “D” curves. The “C” curve is appropriate forparent plate. The “D” curve is more conservative and isused for elements adjacent to welds. It is possible toimprove the fatigue performance at the weld toes, byreducing or putting in compression the intrusions or otherimperfections, which is an inherent feature of weld toes insteel.

Edison Welding Institute’s experience based onvarious test data, indicates that the fatigue life at the weldtoes can be extended to approximately a “C” curve if burrgrinding or hammer peening is used to form a smoothcurved area which removes the toe and all undercut [26].Hence, the fatigue life for these details can be assumed toincrease by a factor of ~2.5, making “C” curve fatigue lifeprediction more appropriate than “D” curve prediction.

Fatigue ResultsModifications were necessary in most of the

structural details in order to achieve the desired 30-yearfatigue life. Most modifications were minor and includedadjustments such as softening of bracket terminations,changing shape and location of erection joint cutouts,changing plate thickness and the use of full collars at afew locations.

More extensive improvements were needed in orderto reach the desired fatigue initiation life in the lowerhopper and at the deck longitudinal to vertical bulkheadstiffener connection. Even with the improvementsimplemented from the preliminary design analysis, furtherimprovement were needed to extend the fatigue life in theradius tank top plating at web frames, and in the adjacentweb frame cutouts.

In the hopper connection, the desired fatigue life wasreached by using a full collar in way of the adjacentlongitudinals. A partial longitudinal stiffener was addedbetween the last tank top longitudinal and the girderbelow the radius hopper. Finally, to ensure that thedesired fatigue life was achieved, peening was used on thewelds between the curved portion of the tank top platingand web frame (Figure 13 and 14).

A significant stress riser was found in the decklongitudinals where they intersected the vertical bulkheadstiffener. The initial design used a large vertical stiffenerin combination with a large bracket at this connection; thefinal design, which gave us the desired fatigue life, used atapered vertical bulkhead stiffener in combination with asmall soft-toed bracket (Figure 15). The tapered verticalstiffener reduced the longitudinal stiffness transition at the

connection with the deck longitudinal while still havingthe needed stiffness at the upper horizontal stringer.

Fig. 13 Lower Hopper, Tanktop Plating Removed

Fig. 14 Lower Hopper, Zoom of Radius, PlatingRemoved

Fig. 15 Initial and Final Bulkhead Stiffener to DeckLongitudinal Connection

17

VIBRATIONS

The shipyard performed preliminary vibratory naturalfrequency studies on a subcontract basis with Det NorskeVeritas (DNV). The studies were based upon a two-dimensional variable beam model for the hull-girder and atwo-dimensional beam model with springs for thedeckhouse. Simple beam and plating theories were usedfor the local structural elements.

The studies were performed to determine if therewere any resonant frequencies with the 4th and 8th order ofthe propeller and/or 7th order of the main engine. Thestudies revealed that, in the after body, several localmembers and panels were deficient due to the reducedfrequencies in the submerged mode (fluid one side or bothsides). The corresponding inertia of these members andpanels was increased to ensure that fatigue failures wouldnot result from their vibratory response.

The preliminary studies also determined that therewas a resonant condition of the deckhouse bridge wingswith the propeller 4th order excitation frequency. Becausethe studies determined that there was a resonant conditionwith the bridge wings, additional studies were warrantedto determine the forced response of the bridge wings withthe interaction of the hull.

The final vibratory (natural and forced) responseswere determined by a three-dimensional model of theentire vessel including the main engines, with the modelconsisting of 6,570 elements with 20,190 degrees offreedom. The calculated pressure impulse forces inducedfrom the propeller as well as the imposed main engineforces were applied to the model, for the full load andballast conditions.

The forced response study determined that theresulting vibratory response for the bridge wing was notin compliance with ISO standard velocity limit of 4.0mm/sec for a frequency of 8.47 Hz, related to thesusceptibility of human exposure. Hence, the vertical-truss support configuration for the wings was amended aswell as the support structure between the bridge windows.

In the sloshing analysis, the first mode naturalsloshing resonance period (with the vessel’s pitchingperiod) was determined to be at only 18% of the cargotank filling height. This would result in the correspondinginduced pressures being less than the normal requireddesign pressures.

OVERVIEW OF SHIPYARD ACTIVITIES

Plan Review ProcessThe design review process resulted in a particularly

interesting relationship between the Shipyard, AMI, andJJMA. AMI contracted with JJMA to support designreview. Drawings were issued from the Shipyard to bothAMI and JJMA. JJMA performed initial review of the

drawings for compliance with the specification andregulatory body rule. Comments were provided to AMIin the form of detailed comments on the design. AMI inturn reviewed the comments, made appropriate changes asnecessary based on Owner preference or experience andsubmitted the final review comments to the Shipyard. Inorder to ensure all issues were addressed, weeklymeetings were held between AMI and the Shipyard toreview design issues. The end result of this processyielded minimal on-site staff for AMI but included theoverall expertise of a design firm with extensive designreview experience.

Construction MilestonesKeel laying for the ARCO ENDEAVOUR occurred on

May 5, 1998 and the ship was launched on December 17,1999. Keel laying for Hull 2498, the ARCORESOLUTION, took place on July 12, 1999. Keel layingfor the ARCO DISCOVERY, Hull 2499, is scheduled forNovember 2000.

Analysis in Support of ProductionDuring the detailed design phase and into the

construction period, several miscellaneous structuralissues were also addressed through the use of finiteelement analysis during the design and building phase ofthe project:

• Rudder and Associated Castings - A rudder analysiswas performed to verify the adequacy of the rudderdesign when subjected to hydrodynamic loading dueto both ocean waves and ship’s steering. Pastexperience with cracking at sharp radii at the lowergudgeon led to an expressed emphasis on the rudderdesign by AMI. Prudent design work by theShipyard resulted in an acceptable design and utilizeda continuous casting between the upper and lowergudgeon.

• Cargo Riser and Drop Pipes - An extensive analysisof the cargo riser and drop pipes in the cargo tankswas performed. The piping system was analyzed forinternal thrust loads, cargo sloshing and loadsinduced by thermal expansion. In the desire to keepthe exposed Upper Deck clear of oil carrying pipes tothe greatest extent possible all cargo pipes were runthrough the cargo tanks. At the manifold the cargofill and discharge lines run vertically from theinnerbottom tank top to the Upper Deck and passthrough the horizontal stringers. Initially, the riserand drop pipes were to be welded at theirpenetrations through the stringers. With the ambientsteel temperature as low as 40 degrees F (4 degreesC) and the cargo loaded at as much as 105 degrees F(41 degrees C) there proved to be a significantthermal expansion in the pipes. The solution was to

18

Fig. 16 Cargo Riser and Drop Pipe Model

leave the cargo pipes loose through the stringers andanchor them to restrict horizontal motion, allowingvertical expansion. This eliminated “locking” thestringers together as well as pushing up the UpperDeck (Figure 16).

• Plug Unit Stresses - An analysis was also performedto identify the amount of locked in stresses created byfitting of the last of the construction block in eachtransverse band across the ship. These Upper Deckunits were plugs placed between the centerlinebulkhead and the longitudinal bulkhead structure. Inmatching of this plug unit with existing structure atthe upper web frame erection joint it was found thatthe sequencing of weld-out on panel stiffeners wascritical to maintaining the in-plane alignment of theweb frame. The analysis demonstrated how certain

panel stiffeners could be freed-up to avoid anyadverse effects of the weld-out process.

• Erection Joint Cutout Tolerance - The early finiteelement analyses proved to be a useful tool later inthe construction process. This became apparent inthe fabrication of “mushroom”-shaped cutouts thatspanned erection joints. Because of minor variationsin the amount of neat material on some units or theoccasional shift in placement of longitudinals thedimensions of the mushroom cutouts varied.Because of prior knowledge that these mushroom cutdetails were sensitive to fatigue a design study wasundertaken. The desire was not to analyze eachdetail but to provide a range of dimensionaltolerances for the mushroom cutouts. Knowledge ofthe relative effects of variations in the mushroom

19

cutouts gave both Shipyard production personnel andAMI’s steel inspectors a hard and fast guide forcutouts in question during the fabrication process(Figure 17).

Fig. 17 Construction Tolerance for “Mushroom” Cutouts

• Weld Procedure for Bulb Plates - The desire byARCO to use bulb plates provided another uniquechallenge for the Shipyard. With no previousconstruction experience using bulb plates there was aneed to develop weld procedures and disseminate theinformation throughout production. Little was foundin the way of published guidance relating to the endpreparation for butted connection of bulb plates. TheARCO steel inspectors had significant internationalexperience and worked with the Avondale weldingengineers to perfect end preparation details suitablefor use in production. Samples were fabricated in theShipyard Welding Laboratory and an approvedprocess was established. Buy-in was achievedbetween the Owner, the Shipyard and theClassification Society.

Stages of Construction/ Block Breakdown PhotosAppendix A provides photos taken during various

construction phases of the Millennium Class tankers. Thephotos provide various perspectives on the structure andthe structural details that are included on the vessels.

LESSONS LEARNED

The significant scope and unique approach on theMillennium Class Tanker project has offered manylessons to be learned. With the extensive analytical workthat was performed and the feedback into the design it issafe to say that few, if any, aspects of the overall design

would be changed if the project were repeated. Thegreatest application of the lessons learned on the projectcan be used in the processes undertaken to meet designrequirements. Important caveats were illustrated in thefinding of unexpected analytical results and in the uniqueproduction challenges that this ship presents.

Encourage Extensive Development of Scantling PlanOne of the most important aspects of the design effort

was the translation of the Scantling Plans into the designdetails. The Midship Section was clearly defined and thedetailed design of the cargo block was fairlystraightforward. However, when the detail design movedto the forward and aft transition structure, the challenge toachieve continuity was significantly increased. Thisoccurred for a number of reasons. Scantling Plans weredeveloped only to a preliminary extent. Instead, effortwas made to quickly support the detailed structural needsof the DLA and fatigue analysis rather than to develop thedetailed Scantling Plans. As detailed design progressed itbecame clear that additional development of the ScantlingPlans at the early stages of design would have beenbeneficial. Most importantly, a full definition of the foreand aft structural transition areas should be clear at thescantling level.

Explore Local Details CarefullyNeither SafeHull nor DLA initially identified some

problematic structural areas. It was particularlynoteworthy that no stress or fatigue problem showed up atthe lower hopper in either the SafeHull or the DLAanalysis. No problems were identified at the side shell,bottom or deck longitudinal connections. Fatigue wasaddressed in SafeHull but SafeHull empirical tablespredicted much longer life than what the detailed fatigueanalysis revealed.

In other cases, findings from the DLA that were notaddressed in SafeHull were also of significant interest. Aprime example is the strengthening of centerline bulkheadaway from midship, particularly in way of cutouts in thepump room. SafeHull and DLA are useful tools in designbut it is also necessary to take care of local fatigue-sensitive details.

Know Your Design SoftwareThe sensitivity of long-term extreme calculations was

noted when comparing extreme values based on outputfrom different ship motion programs. The DLAprocedure provided significantly different long-termextreme results even when the basic motioncharacteristics of the input matched reasonably well withthe input of the other ship motion programs. The extremeroll angle calculated by SMP, based on model test-deriveddamping coefficients, gave an unrealistically high long-term extreme roll angle. The SCORES II program, usinga standard damping factor, estimated a maximum roll

20

angle consistent with what has typically been used onDLA analysis of similar ships.

Understand Design Impact on Engineering andProduction

One particularly interesting aspect of the MillenniumClass tanker design was the impact on engineeringsupport services and Production. As a result of the effortto achieve continuity of structure the overall ratio of partsderived from plate versus parts derived from shapes wassignificantly different. That is, the exclusive use ofbacking brackets and web frame stiffeners cut from flatplate was not typical of other vessels. The MillenniumClass tankers have a ratio of plate-originated parts toprofile-originated parts of approximately 70:30.According to the Litton-Avondale Lofting Departmentthis varies significantly from other vessel designs wherethe typical distribution is relatively standard at 50:50.This variation resulted in a redistribution of man-hoursspent in the lofting effort as well as in the production

loading of product lanes. In short, it was necessary to cutmore parts cut on plate burning machines.

Take Advantage of a Team ApproachThe team approach of addressing the structural

design of the Millennium class tanker proved to be anexcellent situation. Ideas were exchanged openly, fitnessof the design was supported by extensive experience andthe overall quality of the final product was enhanced.This carries not only to the ship itself but also to alldesign information that will be maintained throughout thelife of the vessel.

Constant communication was necessary to make thisapproach work. In the end, the Owner obtainedsignificant input on the design and the designrequirements were met. The Shipyard benefited from theOwner’s past experience and obtained information onnew design techniques from the contractors. Theconsultants derived benefits from the Owner’s extensiveexploration of details and new design requirements.

Fig. 18 Diagram Showing Selected Benefits of Team Approach to Millennium Class Tanker Project

21

CONCLUSION

The effort to achieve a high structural standard forthe Millenium Class tankers clearly departs fromtraditional thought processes. Design has been extendedbeyond Class Society requirements to incorporate lessonslearned as successful marine operators. Throughoutdesign the goal was to improve the safety and structuralreliability of the vessels and address the needs of futureoperators of the ships. While the physical result of theseeffort are visible now, in the end, time will be the ultimatejudge of success.

ACKNOWLEDGEMENTS

The authors would like to thank a number ofindividuals that made this paper possible. John Sullivanand Robert Levine from ARCO Marine, Inc. haveprovided visionary guidance and unending support fromthe start of the project. William Young and RobertKitney, the AMI steel inspectors, must be thanked fortheir tremendous efforts to ensure that the extensivedesign analysis was ultimately supported by goodworkmanship. George Knight, Al Bozzuffi and IvanZgajlic at JJMA have continuously shared new ideas andoffered recommendations on incorporating innovativedesign features. Dr. Maxwell Cheung at MCA Engineerscontributed his expertise in addressing complex structuralissues and challenging the traditional methods ofstructural design. Others that contributed significantlythroughout the project include Tom Longo at Litton-Avondale Shipyard, Keith Michel at Herbert Engineering,Bruce Hutchison at The Glosten Associates, and LarsGustafsson and Peter Tragardh at the SSPA model testfacility.

REFERENCES

1. Cheung, M. C., “Cost Effective Analysis for TankerStructural Repairs”, NRC Symposium andWorkshop on the Prevention of Fracture in ShipStructure, 1995

2. Sucharski, D. and Cheung, M. C., “Application ofSpectral Fatigue Analysis to the ARCO 190 MdwtTankers”, Ship Structures Symposium, SNAME andShip Structure Committee, 1993.

3. Sucharski, D., “Crude Oil Tanker Hull StructureFracturing An Operator’s Perspective”, NRCSymposium and Workshop on the Prevention ofFracture in Ship Structure, 1995.

4. Witmer, D. J. and Cheung, M. C., “The Use of FiniteElement Modeling Techniques in the Investigation ofStructural Failures in Ships”, SNAME West CoastTanker Operations Symposium, Joint CaliforniaSection Meeting, 1991.

5. Henn, A.E. (RADM), Commandant (G-MVI-1),United States Coast Guard, U.S. Department ofTransportation, Navigation and Vessel InspectionCircular No. 15-91, “Critical Area Inspection Plans(CAIP’s)”, October 1991

6. Tanker Structure Co-operative Forum, “GuidanceManual for Tanker Structures”, Witherby & Co. Ltd.,London, England, 1997

7. Tanker Structure Co-operative Forum, “Guidelinesfor the Inspection and Maintenance of Double HullTanker Structures”, Witherby & Co. Ltd., London,England, 1995

8. International Chamber of Shipping Oil CompaniesInternational Marine Forum, “Guidance Manual forthe Inspection and Condition Assessment of TankerStructures”, Witherby & Co. Ltd., London, England,1986

9. Lloyds Register, ShipRight, “Fatigue DesignAssessment Procedure, Structural Detail DesignGuide”, January 1996

10. American Bureau of Shipping, “Rules For Buildingand Classing Steel Vessels”, New York, 1997

11. American Bureau of Shipping, “SafeHull System forTankers, Guide on Improvement for StructuralConnections and Sample Structural Details – ServiceExperience and Modifications – For Tankers”, NewYork, July 1995

12. Sullivan, Levine, “The Millennium Class TankerDesign”, Presented at the 1999 Motor ShipConference, Athens, Greece

13. United States Code of Federal Regulations,“Navigation and Navigable Waters”, 33 CFR165.1303(b)), Office of the Federal Register, 1997

14. International Maritime Organization, IMO MaritimeSafety Committee Circular.686, “Guidelines on theMeans of Access to Structures for Inspection andMaintenance of Oil Tankers and Bulk Carriers”,London, England, adopted June 1995

15. International Maritime Organization, SOLASConsolidated Edition, London, 1997

22

16. ASTM Designation: F1166-95a, “Standard Practicefor Human Engineering Design for Marine Systems,Equipment and Facilities”, American Society forTesting and Materials, April 1996

17. International Maritime Organization, MARPOL73/78 Consolidated Edition, London, 1997

18. SPLASH, A Free-Surface Panel Code Flow Solver bySouth Bay Simulations, Inc.

19. SCORES II, Strip Theory Ship Motion Program Usedby ABS

20. Lundgren, J., Cheung, M. C. and Hutchison, “Wave-induced Motions and Loads for a Tanker.Calculations and Model Tests”, PRADS 98Symposium on Practical Design of Ships and MobileUnits.

21. Stenseng, A., “Cracks and Structural Redundancy”,Marine Technology, October 1996

22. Gordon, R. and Kirk, M., “Fracture Mechanics andFitness-for-Service”, Fracture Mechanics andFitness-for-Service Short Course at AMI, By EdisonWelding Institute, 1994.

23. Stambaugh, K., A., Lawrence, F. and Dimitriakis, S.,“Improved Ship Hull Structural Details Relative toFatigue”, Ship Structure Committee #379, 1994.

24. SMP91, A strip theory shipmotion program, NavalSea Systems Command, Code: 05H3, Washington,DC

25. UK Department of Energy, “Offshore Installations:Guidance on design, construction and certification”,Fourth Edition, 1990

26. Edison Welding Institute, Proprietary Report ofARCO Marine, Inc., “Specification for HammerPeening of Fillet Welds”, February 1998

23

APPENDIX

Fig. A1 View Looking Aft Along Transverse WebFrames in Cargo Tank

Fig. A2 Detail of Transverse Web Frame ToeAttachment to Tank Top

Fig. A3 View Looking Aft Along Hopper To TransverseBulkhead

Fig. A4 Typical “Mushroom” Cutout at Erection Joint inBallast Tanks

24

Fig. A5 View of Hopper After Erection, Looking Aft

Fig. A6 Detailed View of Hopper Connection, Looking from Below with Forward Unit Removed

25

Fig. A7 Photo from Bow Looking Aft

Fig. A8 Photo from Stern Looking Forward (Note Twin Skeg Arrangement Aft)