Risk Due to Miscellaneous Hazards

Abstract

This chapter has the main focus on objects dropped from cranes, while the other hazards accidents during tow, man over board, structural failure and subsea gas release, are dealt with in less detail.

1 Crane Accidents

Accident and incident experience may be used in order to illustrate the risk picture. If we start with dropped objects , the main characteristics in the North Sea are as follows:

• Several fatalities have been caused when the entire crane has toppled overboard, but this was before 1990.

• Equipment damage has been caused by falling load impact on the deck.

• Subsea wellheads have been damaged especially as a result of BOPs falling during exploration drilling .

Fatal crane accidents were quite frequent in the period 1975–1985, with approximately one such accident per year in the North Sea. However, only one accident has occurred since 1985. This occurred in 1988, when the crane hook caught on a vessel due to heavy swells and the crane was dragged overboard. All of the fatal accidents have occurred in the British sector of the North Sea.

One somewhat special crane accident occurred on 3 December 1998 during installation of the production deck for the compliant tower platform on the Petronius field in the US Gulf of Mexico, in some 530 m water depth, where one of the two deck modules for Texaco’s Petronius project fell into the Gulf of Mexico during installation as it was being lifted into place by J. Ray McDermott’s DB50 barge. The north module was hoisted by the DB50 into place on the compliant tower structure earlier the same day, and was in the process of being secured. The north module weighed 3,876 tons and contained wellbay, power and compression equipment. The south module which fell into the sea weighed 3,605 tons. The module contained the production equipment, water flood facilities and crew quarters.

The south module was in the process of being lifted when it suddenly broke from its support. The module struck the transport barge as well as the DB50 before falling to the sea floor. Both barges sustained some damage .

It is claimed sometimes that less serious crane accidents are quite frequent, but it appears that systematic recording is not performed. A detailed study of causes of dropped load s has been performed for PSA , but is only available in Norwegian [1].

1.1 Modelling of Dropped Object Impact

The modelling of risk associated with dropped objects is often formulated as follows:

$$ P_{FDI} = \sum\limits_{I} {N_{i} P_{Di} \sum\limits_{J} {P_{Hij}\,P_{Fij} } } $$

(11.1)

where

P _FDI :

probability of equipment failure due to dropped object impact

N _i :

number of lifts per load category, i

P _Di :

probability of load dropped from crane for load category i

P _Hij :

probability of equipment j being bit by falling load in category i, given that the load is dropped

P _Fij :

probability of failure of equipment j given impact by load in category i.

The probability of hitting equipment in particular requires further modelling depending upon the type of equipment that could be hit, the process equipment on deck, the support structure above or below water, and/or the arrangement of subsea installations.

In addition to the probabilities of hitting equipment, the energy of the impacting load and the energy transfer both have to be established as the probabilities of failure due to impact are very dependent on energy levels. Each of these aspects is discussed separately below. The physical modelling of the fall is considered first.

1.1.1 Crane Load Distributions

Cranes are of vital importance to the operation of an offshore platform. During the drilling period in particular the cranes are operated almost continuously. Even during normal production operations the cranes are used regularly.

The loads handled with cranes differ in weight from light loads to multiple drill collars with weight to up to about 30 tons. In addition blowout preventers (BOPs), which can weigh up 150–220 tons, are handled with the derrick drawworks. The mass of an object and its velocity determine the energy it will gain through a free fall, and thus the damage it might cause. It is also likely that the probability of crane failures increase with increasing weight. Therefore, it is important to obtain statistics on the load distribution for crane activity. Table 11.1 presents two load distributions that are considered to be representative.

Table 11.1 Load distribution for different phases of production

This table shows a typical number of crane operations per crane during one year for a production installation, both for simultaneous drilling and production, and for normal production operations. Other surveys have given a range from 2,700–30,500 lifts per year, depending on the number of cranes (from one to four cranes).

BOPs are mainly moved during drilling or workover, and usually not by the crane. During the drilling of one well, the BOP may be moved 1–5 times, by special lifting/transporting equipment and derrick drawworks. There are several known instances where a BOP has been dropped during such movements. This may cause damage to the BOP itself, but also cases of damage to the subsea installations are known. A fatality has also occurred in one instance when a BOP was dropped.

In addition to falling loads from a crane, there have been various cases of boom fall and crane fall in the North Sea. The first type of accident arises when the crane boom (typically 25 tons) falls from the crane, and the second when the entire crane structure (typically 60 tons) breaks loose from its base.

The main hazard associated with the fall of a crane structure or boom is that the crane driver may not be able to escape in time, and thus be dragged under the water with the crane.

1.2 Physical Aspects of Falling Loads

There are principally two cases that need to be considered separately in modelling the fall of a load from a crane. These are:

• Loads that are dropped onto equipment/structures on the deck or otherwise above the sea surface.

• Loads that are dropped over the sea with the possibility to hit structures in the water or on the sea bottom.

The first case has only one phase, whereas the second case has three phases, the fall through air, the impact with the sea surface, and the fall through the water. The following discussion is focused on these three phases which, implicitly address the first case as well.

1.2.1 Fall Through Air

Friction loss during the fall through air is negligible, due to the high specific weight of loads and thus a falling object will accelerate towards the sea surface in accordance with the force of gravity. The sideways movements will be determined by possible movements of the platform (applicable to floating units only) and the crane hook. Typically the dropped object will hit the sea at an angle within 3° of the angle it was positioned when on the crane hook.

1.2.2 Impact with Water

A falling object will hit the sea surface with the velocity v₁, and proceed through the water with the velocity v₂. These two velocities are given by the following equations:

$$ v_{1} = \sqrt {2gh} $$

(11.2)

$$ v_{2} = v_{1} - \int\limits_{0}^{t} {\frac{P(t)}{{m_{{f_{0} }} }}dt} $$

(11.3)

where

g :

gravity acceleration

h :

height from which the drop occurs

P _(t) :

impact force

m _fo :

object’s mass.

The integral represents the loss of momentum during the impact with the water surface. It is shown that this integral is a function of:

• the density of the object

• the impact angle with the water surface

• the mass of the falling object

• the density of water.

After the impact the object will accelerate towards its terminal velocity, v_t, given by:

$$ v_{t} = \sqrt {\frac{2(W - O)}{{C_{d} \cdot A \cdot \rho }}} $$

(11.4)

where

W :

gravity force (in air)

O :

buoyancy force

ρ :

density of water

A :

cross-section area

C _d :

shape coefficient of the object depending on the Reynolds number

It is also known that an object will tend to oscillate sideways during the fall through water. These oscillating movements are determined by the impact angle with the water surface and the external shape of the object. ‘Barlike’ objects and objects with large surface areas will oscillate more than massive and spherical objects. An oscillating object will have a lower terminal velocity than a non-oscillating object.

The path of the object through the water is also influenced by the currents that are present. After passing the sea surface, the object will move a distance s in a horizontal direction where s is given by the equation:

$$ s = \mathop \smallint \limits_{o}^{t} v_{0} \frac{Xt}{1 + Xt} dt $$

(11.5)

$$ X = \frac{{\rho \cdot CAv_{0} }}{{2m_{{f_{0} }} }} $$

(11.6)

where

v ₀ :

current velocity

C :

drag coefficient

The drift caused by the currents has to be taken into consideration when calculating the most probable landing point on the seabed of a falling object .

1.3 Probability of Dropped Loads

The probability of dropped load s during crane operations is considered to be dependent on the characteristics of the load and environmental conditions (when floating installations are involved). It is however, unusual to have sufficient data to discriminate between these differences. Typically, only one average frequency may be estimated, for instance, an average drop frequency per lift or per crane year.

WOAD ^® [2], probably the most commonly used data source for incidents involving dropped loads , falling crane boom or failure of the crane base itself. It is considered that events such as the failure of the crane boom or base, are unlikely to occur without being noticed and reported. When it comes to loads falling during handling, it is quite likely that these may not be reported if there is no subsequent damage . It is therefore quite possible that the frequency of falling loads based on WOAD^® reported events is an underestimate.

The typical frequency of dropped load s per crane is in the order 10⁻⁵–10⁻⁴ loads dropped per crane per year, it could even be up to one order of magnitude higher. For critical lifting operations, particular emphasis is sometimes placed on adhering to strict procedures and this is sometimes called a ‘procedure lift’. The frequency of dropped loads may under such conditions be typically 30–70% lower than the value for a ‘normal’ crane operation.

1.4 Probability of Hitting Objects

It is useful to distinguish between different types of objects that may be hit, mainly on the basis of the potential worst case consequences of such occurrences:

• Topside equipment: May cause loss of integrity of hydrocarbon containing equipment possibly causing a process fire .

• Subsea installations: May cause loss of containment of production (HC containing) equipment, possibly causing a significant spill.

• Structural components: May cause structural failure or loss of stability or buoyancy.

1.4.1 Dropped Loads on Topside Equipment

The probability of hitting topside equipment is usually based on geometrical considerations reflecting the areas over which the lifting is performed.

Lifting over process area s is usually not permitted by operational procedures unless special restrictions are implemented. If a load is dropped under such conditions, it may be a critical event. The probability of being hit may be expressed as follows:

$$ \mathop P\nolimits_{Hij} = \frac{{A_{lij} }}{{A_{criti} }} \mathop f\nolimits_{crit} $$

(11.7)

where

A _lij :

area of equipment j over which loads in category i may occasionally be lifted

A _criti :

total area of hydrocarbon equipment over which load category i may be lifted

f _crit :

ratio of critical area to total area over which lifting is performed

1.4.2 Probability of Impact on Subsea Installations

The probability of hitting subsea installations is also usually based on geometrical considerations, which will then reflect the areas over which the lifting is performed.

When lifting over subsea installations lifting or lowering is frequently performed with a horizontal offset, in order to avoid damage to the subsea facilities if the load is dropped. The probability of the subsea equipment being hit may be expressed as follows:

$$ \mathop P\nolimits_{Hij} = \frac{{A_{lij} }}{{A_{subsi} }} \mathop f\nolimits_{subs} $$

(11.8)

where

A _lij :

area of equipment j over which loads in category i may occasionally be lifted

A _subsi :

total area of subsea equipment over which load category i may be lifted

f _subs :

ratio of area over subsea installations to total area over which lifting is performed

1.5 Consequences of Impact

1.5.1 Consequences for Topside Equipment

The principles are the same as for subsea equipment, outlined in the following section. It should be noted, however, that the probability of loss of containment and subsequent fire is often considered in a simplified manner.

1.5.2 Consequences for Subsea Equipment

The most important installations which may be subjected to falling object s, are underwater production system s (UPS) and pipelines.

Underwater production systems are mechanical equipment units, consisting of pipework, valves and controls, mounted on a frame or in an enclosure on the seabed. Their purpose is to connect the wells to the pipelines or risers which convey the well fluids to the process module. Typical UPS-modules are Xmas trees, made to control and shut down the wellstream, and control modules. UPS-modules normally have masses up to 30 tons. A Xmas tree has a height of about 4 m, and covers a horizontal area of about 8 m². Pipelines on the seabed for oil and gas, have diameters up to 40’ inner diameter.

The actuators on the Xmas trees are among the most vulnerable subsea components, and these are considered as an illustration of damage to subsea equipment .

The actuators are intended to operate the valves on the X-mas tree. A typical actuator consists of a stem which keeps the valve open. Hydraulic pressure is used to keep the valve open against a spring. Thus, if problems lead to a loss in the hydraulic pressure, the actuator will operate the valve, which will isolate the well. The actuators have a length of about 1 m, and they are often mounted in relatively unprotected positions on the Xmas tree.

The consequences of an impact are dependent on how a falling load actually hits subsea equipment i.e., the velocity of the falling load, where the subsea equipment is hit, the impact angle , the impact time, and the contact area. For a specific load, these values are difficult to estimate, and concentrates on the amount of energy which is transferred between the objects, and the deflection this energy causes on the equipment. For an actuator the deflection is:

$$ y = P \frac{{\mathop l\nolimits^{3} }}{3}E \cdot \mathop I\nolimits_{0} $$

(11.9)

$$ P = 2m_{{f_{0} }} \frac{v}{{\mathop t\nolimits_{d} }} $$

(11.10)

where

l :

length

E :

elastic tension module

I ₀ :

moment of inertia

m _fo :

mass of falling object

v :

velocity of impacting object (before contact)

t _d :

duration of energy transfer during impact.

Some calculations have been made for ideal situations. These indicate that a falling load with a mass of 2 tonnes could easily damage an actuator, and for heavier loads a blowout would be a most probable consequence. The same loads applied to a pipeline may cause damage and leakages.

1.5.3 Consequences to Structural Components

Loads due to falling object s and equipment are a result of the impact energy, direction and geometry of the contact area. Hence, it is natural to distinguish between loads due to long cylindrical objects (pipes) and loads from bulky objects because they have different drop rate, trajectory/velocity in water, and effect on the structure. The impact loads on the following structural components are of interest:

• topsides

• module support beams

• supporting structure

• buoyancy compartment s.

The elements supporting structure and the buoyancy compartment s will often need to be subdivided further, due to strength variations, different hit probability, etc. In principle, the probability of an impact by a falling load should be based on:

• Frequency of lift operations.

• Frequency of dropped loads as a function of lifting procedures, precautions, etc.

• Conditional probability of drop location and height.

• Conditional probability of a particular dropped object hitting a particular structural component, given the drop location and height. For underwater parts of the structure due account needs to be taken of the behaviour of the load in water (which depends upon the object’s angle with the water surface when hitting the water, its shape etc.).

• Conditional probability of impact geometry (e.g. the angle between the axis of a pipe and the impact surface), given a dropped object and a hit.

• Conditional probability of velocity, with a given dropped object and a hit.

Studies are often based on several simplifications, due to lack of data.

1.6 Impact Energy Distributions

The energy distributions may be calculated based on geometrical distributions, frequencies, and probabilities of failure and hit. Three examples are presented below, for columns of a floating production vessel , module supports beams, and topsides modules. Table 11.2 applies to impact on columns and the top of the column, while Table 11.3 applies to impacts on cantilevered structures and exposed beams.

Table 11.2 Cumulative hit frequencies for cylindrical objects and columns

Table 11.3 Cumulative hit frequencies and energies for dropped objects on module support beam

RNNP has collected an extensive amount of dropped object incidents on production installations.

Figure 11.1 presents the conditional exceedance probability distribution for impact energy for the dropped objects. The curve applies to all types of dropped objects, not only from cranes.

Fig. 11.1

Impact energy distribution for dropped objects on production installations, 2013–2017, NCS

1.6.1 Impact Energy Exceedance Curves

Figure 11.2 shows an example of an impact distribution of loads dropped from a crane. The annual impact frequencies shown relate to hits on a subsea installation, as a function of the impact energy.

Fig. 11.2

Impact distribution for dropped objects from crane

For design purposes, this would not be sufficient information, as the type of object, its velocity etc, would also be needed.

The next diagram, Fig. 11.3, presents separate impact energy distributions for three different objects, two ‘barlike’ objects (light and heavy) and medium ‘boxlike’ (edged) objects. The distributions correspond to the maximum energies shown in Table 11.4 below.

Fig. 11.3

Impact distributions for different objects

Table 11.4 Energies of falling object s at seabed level

2 Accidents During Tow

Accidents during tow are particularly relevant for jack-up platforms. An overview of accidents to jack-up platforms for the period 1980–1993 revealed that 11 accidents out of a total of 69 accidents were due to towing problems [2].

Serious accidents during towing of jack-up platforms have also occurred in the North Sea, in August 1990. The West Gamma jack-up was being towed between two locations in the North Sea, when it capsized during severe weather conditions. No fatalities occurred.

3 Man-Overboard Accidents

Man-overboard accidents may be considered a subcategory of occupational accident s, but are an important subcategory, particularly because a person falling overboard is in need of assistance from emergency response services, by means of a Fast Rescue Craft (FRC ), or Man-Overboard (MOB) boat. Lately, Daughter Crafts are also being used, i.e. a large MOB boat, with a small steering cabin that also serves as a shelter for crew and survivors, and dual propulsion systems.

The emergency resources may be installed on the installation itself, and/or on the standby vessel . There is a requirement in Norwegian legislation for production installations that two independent systems be installed. A daughter craft will satisfy this requirement due to its dual propulsion systems.

Installations in North-west European waters have a large freeboard, due to high wave heights in extreme conditions. This implies that a person falling over board may fall up to 30–40 m before hitting the sea. Persons falling over board may be injured due to hitting structural elements in the fall, or when hitting the water. They may also suffer from hypothermia if left in the water for a long period without protective clothing, and may also drown. It is therefore important to rescue such persons within a short period.

The North Sea practice is that a person should be rescued out of the water within 8 min from the alarm is sounded. This implies that MOB boat crews must be available rapidly when performing activities that may lead to man-overboard accidents, such as when erecting scaffolding over the side of the installation.

3.1 Frequency of MOB Accidents

The Risk Level project [3] is the source of overview of occurrences of MOB accidents in the Norwegian sector. In the UK, HSE has published overview of accidents and incidents on production and mobile installations. Since recent UK data have not been available, only data until 2003 are included for UK.

Figures 11.4 and 11.5 presents the available statistics for the Norwegian and UK sectors. It should be noted that the majority of the MOB incidents in Norwegian waters have occurred from attendant vessel s. These are not covered in the UK statistics. The frequencies are therefore not directly comparable.

Fig. 11.4

Occurrence of MOB accidents in Norwegian sector, 1990–2017

Fig. 11.5

Occurrence of MOB accidents in the UK sector, 1980–2003

There has been one fatality due to MOB incidents in Norwegian waters in the period 1990–2017. This occurred at Maersk Interceptor in 2017 where a person fell to the sea as a consequence of a dropped load when working over sea. If we restrict consideration of the UK sector to 1990–2005, there have been two fatalities, in 1990 and 1996, the former from a mobile installation, the latter from a production installation.

In order to compare the two sectors, we restrict the consideration to production and mobile installations in the period 1990–2003:

• UK: 10 incidents

• Norway: 5 incidents.

It should, on the other hand, be observed that more incidents have occurred on attendant vessels in Norwegian waters, compared to production and mobile installations. When it is considered that there are more UK installations than Norwegian, and that the average annual number of manhours in the UK industry is about 50% higher than in Norway, the ratio between the number of cases may not be very different, when normalised according to the manhours. If we calculate the incident frequency, the values are:

• UK: 1.5 incidents per 100 million manhours

• Norway: 1.3 incidents per 100 million manhours .

In the UK sector we may calculate a FAR value, based on the two fatalities that have occurred. The value is based on the period 1990–2003, as an average for production and mobile installations:

• UK: FAR  = 0.30.

3.2 Scenarios Involving MOB Accidents

The Norwegian data is, as noted above, the most detailed, which gives the best opportunity to consider scenarios where man-overboard incidents have occurred. Figure 11.6 shows an overview of the scenarios where such incidents have occurred in the Norwegian sector, whereas Fig. 11.7 shows the same for the UK sector.

Fig. 11.6

Scenarios for MOB accidents in Norwegian sector, 1990–2011

Fig. 11.7

Scenarios for MOB accidents in UK sector, 1990–2003

It is shown that work over [open] sea is completely dominating (six of nine) for the man-overboard occurrences in the UK sector, whereas the opposite is the case for the Norwegian sector (two of 23). If vessels are excluded, the Norwegian ratio becomes two of six occurrences, which is still considerably less than in the UK.

4 Structural Failure

Structural failure has been one of the difficult aspects when it comes to risk quantification. It has generally been omitted in QRA studies, but this is unfortunate with respect to presenting a complete risk picture. It should therefore be the aim to include risk due to structural failure in the risk results. It must at the same time be acknowledged that risk analysis is probably not the applicable source in order to identify risk reduction measures and their effects. With respect to the main sources of probability of structural failure it is practical to group the sources as follows:

• Probability of failure of the structure resulting from statistical variations in loads and structural loadbearing capacities

• Probability of failure due to accidents

• Probability of failure due to a gross error during design, fabrication, installation and operation of the structure.

The first element is identical with the usual scope of structural reliability studies. Risk in this respect should be controlled by appropriate design standards with specified load and resistance coefficients. A low probability of failure (i.e. <10⁻⁴ per year) is aimed for when design standards are developed. Thus, this probability of failure usually is small compared with the other risks for structural failure .

Probability of failure due to accidents reflects systems failures, such as ballast system failure, anchor system failure, collision impact, falling object s, etc. These mechanisms are addressed in other sections of this book.

The probability of failure due to a gross error is the most difficult to handle, partly because it is outside the scope of structural reliability studies, and partly because such gross errors are impossible to analyse with a normal risk analysis approach. Normally, a detailed plan for verification and quality assurance of important items in the design, fabrication and installation process is required in order to keep the probability of gross errors in a project at a low level. Gross errors are understood to be [4]:

• Lack of human understanding of the methodology used for design,

• Negligence of information,

• Mistakes such as calculation errors (this can be input errors to the analysis programs used and also errors in computer software that are used for design),

• Lack of self-check and verification,

• Lack of follow-up of material data testing, welding procedures, inspection during fabrication, etc.,

• Mistakes resulting from lack of communication or misunderstanding in communication,

• Lack of training of personnel onboard the installation that may lead to maloperation of ballasting systems,

• Errors in systems used for operation of the installation.

Thus, gross errors are understood to be human errors. The nature of the failures as listed above is such that all these scenarios should be possible to detect and rectify in time. Gross errors have been a significant contributor to the failure of structures, and a focus on these issues is considered to be important in order to ensure project success. Two examples of gross errors are the sinking of the Sleipner GBS structure during construction [5], and the overpressure of the cargo system on an FPSO [6].

The approach used by Lotsberg et al. [3], as outlined in Fig. 10.1 may be useful in order to indicate the risk contribution from gross errors.

5 Subsea Gas Release

The possible sources of subsea gas leak s are subsea gas wells, as well as subsea leaks from risers and pipelines. Subsea oil leaks are not considered in this context. The special aspect associated with a gas plume in the water is that a flammable gas cloud may be formed above the sea surface. The possible consequences for buoyancy of floating objects are discussed in Sect. 10.5.

A subsea gas leak may be observed on the surface, if the leak rate is significant, such as in the Snorre Alpha subsea gas blowout (see Sect. 4.9). Other leaks may be difficult to observe, and thus hard to detect, except with ROV. Visually the following parameters may be observed when relevant:

• The diameter of the plume on the surface

• The swell of the water within the plume

• The horizontal water speed.

There may also be a gas cloud formed above the gas plume in the sea. This may have concentrations above LEL, depending on the size of the gas leak and other parameters. An idealised representation of the gas plume and associated gas cloud is shown in Fig. 11.8.

Fig. 11.8

Subsea gas leak which may be observed on the surface

For one 1000 kg/s leak at a depth of 100 m, the void fraction may be around 40%, the water rise may be up to 4 m, and the diameter may be around 250 m. The following factors will, in general, affect the behaviour and shape of the plume:

• Release rate (kg/s)

• Gas density (kg/Sm³)

• Depth of release

• Diameter of release opening

• Release direction

• Currents

• Vertical sea temperature and salt variation.

A gas cloud from a subsea gas leak may cause ignition, if an ignition source is present within the zone with concentration above LEL. The most obvious source of ignition could be a vessel entering the cloud. A passing vessel could enter the zone without knowledge of its presence, or a vessel engaged in emergency response actions if proper safety zones have not been established. See also Sect. 4.9.1, pipeline rupture at the Jotun field, in which case emergency actions were delayed due to the need to establish the extent of the danger zone. In this event, a plume with approximate diameter 100 m was observed by the standby vessel. The water depth at the location of the leak was 126 m. The gas flowrate was calculated to be initially 25–30 kg/s, later dropping to 3–5 kg/s.

In the case of the Jotun leak, the incident was detected due to the effect on the export line pressure on the Jotun FPSO, and the leak location was detected visually by the standby vessel, when following the line in order to search for leaks.

An important source for modelling of the gas plume in the water is Fanneløp [7]. When the behaviour of the gas plume is established, CFD simulation may be used in order to model the behaviour of the gas cloud above the sea surface, and thus the dimensions of the danger zone.

The extension of the danger zone is strongly dependent on the wind speed. For a 1000 kg/s gas leak, the danger zone has a typical downwind extension of around 200–300 m at sea level, with a low wind speed. With moderate wind speed (8–10 m/s) the extension will exceed 1000 m. The zone will reach 30–40 m above sea level.

PSA organised a joint modelling effort in the period 2006–2008, from which two reports have been published [8, 9].