*Dave McKeown is Manager, Corporate Projects for TWI, the research and technology organisation specialising in welding, joining and allied engineering.
This article first appeared in the September 2006 issue of Shipping World & Shipbuilder, which is published ten times a year by the Institute of Marine Engineering, Science and Technology (IMarEST).
During transportation by ship, LNG is maintained at atmospheric pressure at a constant temperature of around -163°C. The tanks in which it is contained are usually vented, and vapour (natural gas) boiling off is allowed to escape so as not to create pressure above the liquid. This boil-off gas can be used to power the ship's engines.
In order to contain a liquid at -163°C, attention must be given to the material of containment and to its insulation from the ship's hull. Even then, significant cooling of parts of the ship structure is likely, so use of Grade D and E steels with good low temperature toughness is mandatory. Such steels present no unique welding and fabricating problems over normal hull grades but careful attention to welding procedures and the quality of welding is required to ensure that welds display the same properties as the base plate.
The materials of the containment tanks themselves are, however, far removed from regular shipbuilding materials. None present insurmountable fabrication problems, provided the workers know the materials, their behaviour and the methods of joining them.
The first commercial LNG carriers had tanks that were self-supporting, prismatic types made from aluminium alloy. This design survives as the IHI SPB system (Ishikawajima-Harima Heavy Industries self-supporting prismatic shape,IMO Type-B). Alloy 5083, Al - 4.5% Mg, is used for the tanks as it has good strength and weldability. The prismatic shape fits within the ship's hull, resulting in a conventional flat deck and the manufacturer claims advantages of stability and easy maintenance. Only two ships (Polar Eagle and Arctic Sun) trade today, operated by Marathon between Alaska and Japan, and there are no new orders declared.
The main fabrication concerns are typical of welding any thick aluminium vessel. Care must be taken to keep the material clean and to scrape or brush the weld preparation just prior to welding. This is because aluminium forms a very adherent oxide that, if incorporated into the weld metal, can give rise to 'lack of fusion' defects and porosity formation. MIG welding is used, traditionally manually, although there is no intrinsic reason why mechanised welding could not be employed.
One advantage of the IHI design is that internal structure within the tanks cuts down the risk of damage through 'sloshing' (creation of a periodic wave in the liquid inside the tank due to the motion of the ship). This is at its worst with tanks filled between 10 and 80% and can create severe pressure on the walls of a tank. The IHI design has internal bulkheads that disrupt the wave formation.
Insulation is attached to outside of the cargo tanks and forms no part of the load-bearing structure. The IHI ships have a capacity of 85 000-90 000m 3 and designs exist for 130 000m 3 .
Moss Rosenberg (later Kvaerner Moss) pioneered a design of spherical tanks that sat half above the deck level, half within the hull. To support the tanks in this position a thick, forged ring runs round the equator of the sphere. This is shaped to be welded into the sphere and to have an outside rim from which a 'skirt' is attached ( Fig.1). The bottom of this skirt is welded into the ship's hull.
The initial ships were built with 9% nickel steel tanks, but in the 1990s aluminium 5083 became the material of choice.
9% nickel steel is especially designed to give excellent strength and toughness at very low temperatures. It is commonly used for onshore storage tanks for LNG.
The main difficulty of using 9% nickel steel is that, although it is readily weldable, welding consumables of similar composition to the base plate are not practically possible.
During plate manufacture special heat treatment is given, either quenching and tempering or double normalising and tempering. This cannot be reproduced in as-cast weld metal. The usual alternative consumable is a nickel alloy,which has excellent toughness and good strength - but not matching that of the base material.
TWI carried out much work on the toughness and performance of the intrinsically dissimilar joint thus produced. Good resistance to brittle failure was found and experiments also showed a good crack arrest when an artificial defect was induced to propagate.
TWI also demonstrated the limits of welding parameters before lack of fusion or hot cracking occurred, resulting in a clearly defined area of operation to give good quality welds. One feature of the dissimilar material junction between the high nickel alloy weld and the steel plate remains problematic - that of inspection.
The two materials have very different sound velocities so ultrasound, commonly used to inspect joint lines for lack of fusion, reflects from solid boundaries as well as defective ones. There is difference in the signal, but operators must be carefully trained to distinguish between them.
The use of radiography is no less confusing. The opacity of nickel is considerably higher than that of steel, leading to radiographs of dark plate with very light welds (viewed as a negative). Exposure must be regulated for the defect being sought. Checking for weld defects such as centreline cracking requires exposures that render the steel too dark on the image to discern any indications. Checking the fusion line for planar defects is also fraught as the strong transition from dark to light can mask defects.
Whether from these fabrication practicalities or economics, the material choice for Moss tanks eventually settled on aluminium alloy 5083. This requires very considerably thicker structures than similar capacity tanks in steel(30 to 150mm), which brings its own fabrication peculiarities. As well as the necessity for cleaning, aluminium is prone to distortion during welding so techniques need to be derived to balance the welding to avoid mismatch. The production of the petal shapes that form the panels of the sphere is also a costly exercise.
The Moss design requires relatively simple insulation. It is not structural and is usually created from polyurethane foam. Sloshing is not a problem in these spherical tanks. Firstly, the shape suppresses some of the tendency to create large waves, and secondly, the thick structural material of the tanks is better able to accommodate the loading.
Moss tanks have been used on ships up to 149 000m3 capacity. Mitsui OSK has two Moss ships currently under construction in Kawasaki of 153 000m 3 capacity.
Gaz Transport and Technigaz (later merged as GTT) developed two designs for thin walled tanks that relied on the ship structure for rigidity and stability. The designs have developed and a new combination, taking features of both, the CS1 system, has recently been introduced. They share a common principle of using thin metal (0.7 to 1.5mm thick) welded liners (or membranes), insulation panels and a secondary barrier layer, all firmly fixed to each other and to the ship's hull. The materials and methods of welding and joining vary between the three.
Gaz Transport No 96
Both membranes in this system are Invar, an iron/36% nickel alloy that has an extremely low thermal expansion coefficient. It shrinks very little on being taken to operating temperature, creating negligible stress. This material is expensive but the weight advantage, maybe 400t compared with a Moss design requiring 4000t of tank material for the same capacity, plays in its favour.
Insulation materials are abundant and cheap. Boxes (200 x 1000 x 1200mm) are constructed in plywood and filled with expanded perlite. The outer insulation layer is bonded structurally to the ship's inner hull. This is achieved with adhesive resin ropes. The operation is critical to the structure as load is transmitted through the joint.
Ships' hulls are notoriously uneven and a high integrity bond must be made over the whole surface - bottom and sides - to effect adequate performance. This is achieved by use of differing thickness of resin rope, then squeezing each box into place against the hull.
A layer of Invar is laid over the insulation. It is available in 0.7mm thick x 500mm wide coils. These are unwound to lay continuous strakes across the wall (or roof or floor). The edges of these strakes are bent up to form a protrusion into the tank. These are initially spot welded together, then a mechanised seam welder traverses the tank creating a leak-tight joint in this upstand.
At the tank corners, Invar boxes termed 'tubes', are welded into place, creating the build platform for the next wall and forming an integral part of the design to accommodate load.
Onto this outer membrane, a further layer of plywood and perlite insulation is laid. Bolts are welded to the Invar so that the insulation boxes can be anchored firmly. Then an inner membrane of Invar completes the tank ( Fig.2). This inner layer has additional webs, keyed into the plywood and welded between the strake edges leaving a 'tongue' protruding into the tank. As well as holding the lining onto the plywood insulation, these tongues add rigidity to the structure.
In principle, tanks built in this way can be very large but restrictions are applied, principally because sloshing can be a very significant problem inducing fatigue and excessive loading on the composite tank structure. GT designs have been used for vessels with capacity up to 146 000m 3 . Under construction are ships with capacities of 263 000m 3 .
Technigaz Mk III
In the early 1970s, Technigaz pioneered the use of the cheaper and more available 304L stainless steel as a membrane.
Stainless steel has a high coefficient of thermal expansion, so will shrink markedly on cooling to -163°C. To accommodate this, corrugations are formed into the sheet, creating an orthogonal pattern to take care of shrinkage in both planes. The stainless sheets are 1.2 x 3000 x 1000mm and are welded together by automatic TIG welding.
At corners and complex joints, manual welding was used but much development of mechanised machines for specific tasks has been carried out, especially in Japan, and welding is almost entirely mechanised on new builds.
The outer barrier (or secondary membrane) in this design is a composite material, Triplex®, comprising of aluminium foil sandwiched between glass cloth and resin. Two layers of insulation are used - between the hull and the outer barrier and between the two membranes, as in the GT design. In the current system, this insulation is polyurethane foam panels, faced with plywood.
Initially, bolts are welded to the ship's hull to locate and hold 300 x 3000 x 1000mm polyurethane panels. The panels are coated with adhesive resin rope and pressed into place so that they are bonded to the hull.
The bolt-holes in the panels are plugged with polyurethane and the gaps between the panels are filled with fibre insulation, leaving a planar surface for the mounting of the Triplex membrane. This is bonded to the insulation. First, the gaps between the panels are covered with strips and then sheets of Triplex are adhesively bonded, overlapping these, to create a leak-tight barrier.
Adhesive bonding of materials such as Triplex and plywood is a very different skill from traditional shipbuilding and, as it is mostly manual, requires dedicated training. TWI has been involved in ensuring that those involved display the necessary competence. As they are required to lay 90km of Triplex strip and bond it without leaks to the thousands of sheets for each ship, their skill is unquestionable.
A further layer of plywood-faced polyurethane insulation is then bonded to the Triplex layer and stainless steel anchor strips are fixed over the joins. The membrane panels are continuously lap welded to these strips, creating a leak-tight container.
Technigaz tank ships have been built with similar capacity to those of the GT design (147 000m 3 ), with ships under construction up to 270 000m 3 capacity.
The Combined System 1 has a primary barrier is of Invar and the insulation and secondary barrier are the polyurethane and Triplex of the TZ system. No ships have yet been completed with this combination but three are under construction in France.
The track record for LNG ships is exemplary, with no accident or collision causing product loss. If leakage should occur, the environmental damage would be limited. LNG is less dense than water and is totally immiscible, so floats on the surface where it boils to natural gas. This is non-toxic and lighter than air so easily dissipates.
Natural gas is, of course, flammable, and in certain mixtures with air, can be explosive so a severe incident is not without possibility. The key is to make sure that the tanks are sound and remain so throughout their life.
LNG is not corrosive, nor does it deposit residues. This means that tanks have long life expectancy. Long life, however, means that fatigue must be considered as a possible means of deteriorating the structure.
The concern with thick steel or aluminium is that any developing crack might trigger catastrophic failure and, as there is no secondary containment, the product would be lost in abundance. A huge spillage of LNG over the ship structure could take sections of the hull well below expected service level temperature with the attendant possibility of embrittlement and risk of brittle fracture.
Mechanical testing has proven the acceptability of both 9%Ni steel and aluminium 5083 with respect to both crack initiation and crack arrest. Monitoring for any signs of fatigue damage remains a concern as the fleet grows older.
There are several proposals for Floating Production, Storage and Offloading (FPSO) vessels. These are very likely to be created from ships that are nearing the end of their service life. FPSOs are likely to spend much time with partially filled tanks, and if these are already 30 years old, detailed work on life extension will be necessary.
Several organisations are studying the phenomenon of sloshing of liquid in partially filled tanks in order to gain accurate data for assessing the loading and its likely effects on the fatigue performance and general structural stability of the tanks.
This is of particular concern for the large box-shaped membrane tanks, especially as the current building programme has ships with around twice the capacity of the average ship in service today. Here again, if membrane tanks are to be used as FPSOs, there is need for much more data collection.
The construction of LNG ships uses materials and techniques not seen in other cargo vessels. The forty-year successful track record is a testament to the quality of design and fabrication but, as is pointed out in this article, expansion of size, life extension and change of use require further extensive testing to be carried out.