Article - Issue 44, September 2010

Shipping Liquefied Gas Safely

Stéphane Maillard and Nigel White

Download the article (536 KB)

An example of a Q-Flex vessel specifically designed as a membrane type liquefied natural gas carrier

An example of a Q-Flex vessel specifically designed as a membrane type liquefied natural gas carrier

Shipping high volumes of liquefied natural gas poses challenges for transporters including that of ‘sloshing’. With new large offshore terminals being built, companies are trying to get to grips with the sloshing phenomenon as Stéphane Maillard of GTT (Gaztransport & Technigaz) and Nigel White of Lloyd’s Register explain.

The maritime trade of liquefied natural gas (LNG) has become a big business since its inception 40 years ago. Today’s seas are busy with ships transporting LNG long distances from large natural gas exporters such as Qatar, Australia, Algeria and South-East Asia to importers like Japan and Europe. With nearly 400 million m3of LNG shipped each year, the number of LNG ships has risen from just over 100 to nearly 350 since 2000. At the same time, the cargo capacity of the largest of these ships has increased from around 140,000 m3to 266,000 m3.

In its liquid form, natural gas – predominantly methane (CH4) – takes up only 1/600th of its gaseous volume, which makes shipping the gas as a liquid commercially attractive. However, shipping a cargo in liquid form at cryogenic conditions brings many challenges for designers.

LNG boils at around -162 °C at atmospheric pressure, which means that natural gas needs to be liquefied by either cooling or pressurising or a combination of the two. The only technique used so far is to refrigerate the natural gas before loading with the ship maintaining the liquid state by the use of insulation. Refrigeration of the cargo aboard is complex and adds additional expense, so good insulation is important to limit the loss of natural gas through boil-off – although most LNG ships use some boil-off as fuel for propulsion and power production. The insulation also has to ensure that the surrounding ship’s inner hull does not get too cold as this can affect the integrity of the steel.

Containment designs

Two main containment approaches are currently used on LNG ships. One alternative is to have independent tanks with external insulation. The principal system using this technique is a spherical tank system, a MOSS system, and this is used for one third of the current LNG fleet, the largest ship built with this system was in 2008 and has a cargo capacity of 156,000 m3. Another system is a prismatic shaped tank (SPB system) of which only two vessels have been built, these were in 1993 with a cargo capacity of 90,000m3. The main drawbacks of these tank systems are that they are not space-efficient, requiring complex hull/tank interactions and are more costly to construct.

The majority of LNG ship owners have selected the second alternative; a membrane containment system. These ships are denoted as membrane tank LNG ships. In these ships, the tank is an integral part of the ship and the insulation is internal to the tank (see Membrane containment systems for LNG). Membrane containment systems have enabled ships with a capacity of 266,000 m3to be built.

The internal insulation system of membrane tanks allows normal shipbuilding steels to be used. The insulation system is fitted directly to the ship’s hull, so no additional steel structure is required. This allows a more lightweight ship to be constructed for the same cargo capacity and lower fees for passing through waterways that charge ships for access based on their size. In addition, there is less windage (the part of the ship above water) subject to wind loading, leaving it less vulnerable to wind effects than the spherical tank designs.

Liquid motion

In spite of these advantages, there are still challenges with the membrane design. One of the issues with transporting liquids is ‘sloshing’. This is where the motion of the liquid in the tank can induce severe impact pressures on the containment system.

Fortunately, the LNG shipping industry has not experienced any major incidents so far due to sloshing. However, there have been some minor incidents where sloshing pressures have caused deformation of the tank walls without compromising the safety of the ship.

All types of tanks, whatever their shape or size, experience sloshing when partially filled. However, the shape of the cargo tank and any internal subdivisions significantly influence the sloshing motion of the LNG. For spherical tank LNG ships, the thickness of the tank boundaries protect it from sloshing damage (in these ships there are no internal subdivisions or supporting structure). The independent SPB tank LNG ships have many internal stiffening members and divisions that break up the liquid flow in the tank so sloshing is not such an issue in these ships: the disadvantage is that these stiffening members and divisions introduce a number of potentially fatigue sensitive areas.

A membrane LNG tank has no internal sub-divisions and is prismatic in shape. This means that sloshing liquid motions can occur in the tank. This has the potential to create high impact pressures on the insulation system. The introduction of internal sub-divisions into a membrane tank is not structurally practical, so the prediction of sloshing loads is a key design issue.

Empty or full

Studies have shown that sloshing pressures in membrane LNG tanks are greatest for medium fill-heights. For this reason, it is recommended that tanks are not filled in an intermediate filling range (known as the barred fill range – see Figure 3), except for some specific cases.

This requirement does not inconvenience the ship-owner as most ships operate with tanks that are usually near full on laden voyages and nearly empty on the return journey – ships usually retain a small amount of LNG on their ballast journeys to keep the tanks cool ready for the next load of LNG. Currently, the usual allowed filling heights at sea are above 70% and below 10% of the cargo tank height. Membrane LNG ships have four or five cargo holds so the crew can distribute almost any amount of cargo between the holds while still keeping clear of the barred fill range in each tank.

However, the barred fill range will present more of a challenge when membrane LNG containment systems begin to be used for moored floating offshore loading or discharging terminals (known generically as floating LNG vessels). These are being planned for certain parts of the world as an alternative to land-based terminals. The tanks in these floating terminals will need to have full fill height flexibility – that is to say, no barred-fill range – in most or all sea states to operate efficiently.

To answer this particular need, GTT has introduced the two-row arrangement: instead of one row of tanks down the length of the ship, the vessel is equipped with two parallel rows of tanks. These tanks are separated by a longitudinal cofferdam on the ship’s centreline (see Figure 4). The two-row arrangement results in reduced sloshing within the tank at all filling heights. This, combined with a strengthened version of the containment system, will allow membrane floating LNG systems to be operated at any location worldwide.

Predicting sloshing effects

Such developments are requiring continuous advances in the methods used to predict sloshing pressures and in evaluating the strength of the containment system under this form of loading.

One of the key problems with sloshing is that it is not easy for the ship’s crew to see or hear it happening in the way that they can with other kinds of ship motion. In a very few cases, sloshing impacts have been heard but this does not give any information on their magnitude. This means that containment designers and ship operators must use other ways to observe or model what is going on in the tanks.

Computer models, specifically computational fluid dynamics simulations, can give a good idea of how liquid moves in the tank as a whole. This gives a picture of the sloshing activity through the observation of tank geometry, wave period, wave height, relative ship/wave heading, ship speed, the ship’s sea-keeping characteristics and the liquid height in the tank.

Depending on the filling height, researching engineers can distinguish two types of wave within a tank for the same excitation. For low liquid heights, the liquid motions are mainly progressive waves, possibly leading to breaking waves travelling from one side of the tank to the other. As the filling height increases, the liquid motions become of a standing wave type. This forms a ‘jet’ of liquid at its leading edge and travels parallel to the tank sides, impacting only at the tank corners (see Figure 5).

Key developments from recent research work have revealed that the roughest seas are often not the most significant factor for producing high sloshing loads. Higher impact loads can be expected in much more commonly-occurring lower sea states. This is illustrated in Figure 6 which shows a simple CFD assessment of the liquid motion in a tank for different sea states, the wave heights have been increased but the wave period has been kept the same. The figure shows the motion of the liquid in the tank a few milliseconds before impact for four wave heights. The figures show that beyond a wave height of 6m, the bore wave formation becomes less pronounced as the motion of the liquid becomes more chaotic, as seen by the greater diffusion of the colour; hence the energy in the bore wave is reduced, resulting in lower sloshing impact loads.

Model behaviour

Several phenomena affect an impact such as the escape of the gas in front of the liquid just before impact, the compression of gas which may have become trapped by the sloshing wave, possible liquid-gas phase changes, and fluid/structure interaction such as hydroelasticity. The effects linked to these various phenomena are the main cause of the variability of the pressure loads, known as stochastic behaviour.

GTT has developed a scale model test facility to study sloshing in its membrane tank designs. The Hexapod model test rig (shown in figure 7) is able to move the tank in all six degrees of freedom just as it would be moved as if it was in a ship in a seaway. The ship motions are predicted using state-of-the-art prediction software and scaled to the model tank scale.

Pressures within the tank are measured with up to 300 sensors that simultaneously record pressure signals at a 20 kHz sampling rate. Typically the large impact pressures last less than a few milliseconds. This large number of sensors enables the formulation of a detailed description of impact pressures within the tank in both time and spatial domains. Each model test runs for around five hours full-scale duration or 45 minutes for model scale.

It is necessary to investigate the sensitivity of the sloshing pressures to different wave heights and wave periods. As well as sea states, the pressure predictions are affected by different ship loading conditions, cargo fill heights, ship speeds and wave headings. As a consequence a very large number of model test runs are needed, typically more than a thousand for one tank. The data is then processed statistically to produce a long-term distribution of sloshing pressures. A scaling factor is also derived to determine what will happen on full-size ships.

Full-scale studies

Lloyd’s Register and GTT have been involved in several joint industry projects to investigate LNG sloshing in membrane tanks. One such project is SlosHel. The aim of this is to create and measure wave impacts at full scale on a wall fitted with the NO96 containment system in order to study hydroelastic effects, scaling laws and the physics of wave impacts. Another example of joint cooperation is a Full Scale Measurement project, which has developed instrumentation to measure actual sloshing impacts onboard an LNG carrier. Since 2008, this system has been measuring in real time the complete ship motions and recording the pressures due to sloshing impacts in two corners of a tank. The insulation boxes and the supporting steel structure in the tank corners are equipped with fibre-optic strain gauges. The aim is to build a database linking ship motions and impact pressures and to investigate scaling laws between model and full scale. GTT’s facilities will be used to reproduce at model scale the sloshing sequences measured during the ship’s navigation.

To answer the growing need for operational advice, GTT is also currently working on the development of a real time Sloshing Prediction and Monitoring System (SPMS) aimed at assisting the crew to mitigate sloshing risks. This system includes sea-state estimation based on ship motion measurements, in situ sloshing detection, and sloshing prediction tools, which will inform the crew on the effect of change of speed or heading on sloshing activity.

Lloyd’s Register made in 2008 the following recommendations to Masters of membrane tank LNG ships to minimise sloshing activity. When a tank is filled in the upper range, the advice is to: “Avoid regular rolling in large angles, especially when combined with regular pitching. If this occurs, then the Master should consider changes of course and/or speed to minimise these motions.” When a tank is filled in the lower range then: “The most risk is from rolling motion, particularly regular large angle rolling in beam seas. If these conditions occur, changes of course and/or speed should be taken by the Master to reduce the large roll angles and the regularity of the roll motion.”

Other applications

Figure 7: Hexapod test rig © GTT

Figure 7: Hexapod test rig © GTT

The extensive efforts to understand the way LNG moves in ship’s tanks may also bear fruit in designing ships to transport other liquids. For example, GTT systems can be used for carrying liquid petroleum gas (LPG). LPG is denser than LNG and so will cause greater sloshing pressures. LPG’s boiling point is also higher than LNG (between -50 °C and around 0 °C depending on the composition), meaning that the containment system has to be designed accordingly. However, the computational and observational approaches are the same.

Having a good understanding of sloshing pressures from simulations and models as well as the feed-back from in-service experience gained with the existing fleet of membrane tank LNG ships equipped with its membrane systems, enables GTT to design containment systems for ships that can withstand sloshing whatever the sea is like. This in-service feedback is also used by Lloyd’s Register to improve the classification rules for LNG ships as well as providing advice to bodies such as IMO (International Maritime Organization) and SIGTTO (Society of International Gas Tanker & Terminal Operators Ltd) with regard to the regulatory framework for the operation of LNG ships. This leaves ship operators free to worry about the commercial and logistical aspects of shipping LNG rather than being concerned about sloshing issues.

[Top of the page]