Woodyard D. (ed.) Pounders Marine diesel engines and Gas Turbines

Подождите немного. Документ загружается.

862 Gas Turbines

can have a signiﬁcant effect on average maintenance costs. Total cycle times

for very short applications are usually determined by the turnaround time at

each end of a route. The additional speed is less beneﬁcial for the short cycle

because a smaller percentage of time is spent in transit. Gas turbine propulsion

generally yields greater beneﬁts in longer cycle applications.

High reliability is a requirement imposed by military ship applications

because of the potentially critical nature of missions. In commercial operations,

however, time in service translates into revenue production; factors crucial to

the commercial success of the vessel design are therefore minimal downtime

and short maintenance intervals.

High availability results from two design considerations. First, the mainte-

nance requirements must be minimized through appropriate rating of the gas

turbine, and mission planning must allow for maintenance that minimizes dis-

ruption of the operating schedule. Second, appropriate installation design must

foster quick removal and replacement of modular components, minimizing

time out of service.

Commercial vessels are usually designed with a particular service proﬁle in

mind, but the design must be ﬂexible enough for diverse deployment if resale

to other operators is planned later. The propulsion plant must therefore allow

ﬂexibility in operation over a range of power levels while maintaining operat-

ing efﬁciency with respect to fuel and lube oil consumptions.

Gas turbine engine maintenance may be considered broadly in two groups:

routine tasks to be carried out onboard (Figure 31.25) and routines that call for

service engineer support or scheduled removal of the GTCU.

Routine onboard maintenance is limited, consisting mainly of daily visual

checks of the running engine. An internal water wash is executed after a ﬁxed

number of operating hours to remove any salt deposits in the compressor which

would otherwise eventually result in degraded performance.

The engine itself can be divided into two main elements: the GTCU and the

power turbine with ancillaries package. The GTCU can be readily removed from

the vessel, within 6–12 h, when dictated by condition monitoring or scheduled

removal. The power turbine and ancillaries are designed to remain for longer

periods, up to the life of the vessel.

In general, the operator will regard the repair of the GTCU as repair by

replacement; it will be removed and replaced, and the vessel returned to serv-

ice in hours rather than days. An owner will therefore normally choose to own

a calculated number of spare engines or major engine components (MACUs:

maintenance assembly change units) to support a particular operation with a

required availability.

A maintenance regime can be provided through a power-by-the-hour

arrangement directly with the gas turbine manufacturer. An indicative ﬁgure

for such an arrangement is often requested by operators, the ﬁgure typically

including scheduled removal and overhaul of the engine and its replacement

by a spare during the overhaul cycle. It is invariably misleading to quote such

a ﬁgure, which is not a feature purely of the engine but highly dependent on

the usage. The calculation will vary strongly with the following factors, notes

Rolls-Royce:

l The operating proﬁle of the vessel.

l The number of starts, stops and idling periods during the route.

l The ambient operating temperatures.

l The maximum and minimum powers required from the engine.

These factors are fundamental in deciding engine life and therefore the

maintenance costs. The resulting calculations are based on creep and cyclic

performance. With a knowledge of the operating proﬁle an assessment can be

made of the GTCU life at which the blading must be replaced. At one extreme,

on a longer ferry route with operation at high power and high ambient tem-

peratures, engine life may be creep limited.

The multiple compressor and turbine stages of the GTCU comprise indi-

vidual discs with blades attached. Any metal disc that is continuously and

repeatedly accelerated from rest to a high speed, then decelerated, will eventu-

ally crack. All the discs have a declared cyclic life, a full cycle being deﬁned

as from rest to full speed and back to rest. The declared life falls well within

that at which a disc might be expected to crack, but when this life is reached

the disc must be replaced.

Cyclic life may become a limiting factor on short ferry crossings, particu-

larly if this is carried out at high engine speeds; it may dictate whether the

Gas turbine maintenance 863

fiGure 31.25 routine maintenance of the rolls-royce MT30 gas turbine (shown

here on the testbed) is limited to checking uid levels and visual examinations

864 Gas Turbines

engine can be shut down or should be left running at idle in harbour, largely

for economic reasons based on engine liﬁng.

It is the combination of creep and cyclic life calculations in conjunction

with condition monitoring of the engine that decides the periodicity of the

required GTCU changes, and therefore impacts directly on any ﬁred-hour

maintenance support arrangement. By working with a prospective operator, the

engine manufacturer is then able to make an accurate assessment of the main-

tenance costs. The combination of engine characteristics and selected operation

determines these costs rather than the engine alone.

Gas turbine compressor washing is an important maintenance routine in

sustaining power output. Since the compressor consumes up to two-thirds of

the total engine power even slight blade fouling signiﬁcantly reduces the net

available output and hence the drive power available or increases fuel costs.

Off-line washing is an effective countermeasure in restoring performance and

overcoming the efﬁciency drop caused by compressor fouling. Online washing

is available when engines cannot easily be shut down for an off-line wash (as

with cruise ships during long voyage legs).

Acknowledgements to GE Marine Engines, Mitsubishi, Rolls-Royce,

Siemens, BP Marine and Altair Filters International.

865

C h a p t e r | t h i r t y t w o

Engineroom Safety

Matters

A ship owner operating a ﬂeet of 20 vessels can expect one major engine-

room ﬁre every 10 years, Det Norske Veritas (DNV) statistics reveal. Avoiding

ﬁres, or at least preventing engine-room incidents from developing into serious

emergencies, is therefore paramount in ship management. A research study by

the Oslo-based classiﬁcation society found that two-thirds of 165 ﬁres onboard

the DNV-classed ﬂeet started in the engineroom. Of these engineroom ﬁres,

some 56 per cent (excluding shipyard repairs) originated from oil leaking onto

a hot surface. The direct cost of repairing engine-room ﬁre damage is com-

pounded by the offhire penalty and loss of goodwill.

From 1 July 2003 all ships of 500 gt and above in international trading have

had to comply with IMO SOLAS regulations covering the protection of hot

surfaces in the enginerooms as well as fuel oil systems. The requirements were

made retroactive from that date for tonnage built before 1 July 1998. Water-based

ﬁreﬁghting systems have been required for all new cargo and passenger tonnage

from 1 July 2002, and for existing passenger ships from 1 October 2005.

Most fuel oils will ignite spontaneously when hitting surfaces with tem-

peratures above 250°C, DNV explains, and class rules therefore require that all

surfaces above 220°C should be shielded or insulated. However, its surveyors

note that such protection is often impaired in service.

A number of methods can be used to detect hot surfaces, DNV recom-

mending: surface/contact thermometers, laser-based infra-red heat tracers and

infra-red thermo-scanning video equipment.

Fuel leakage sources include damaged ﬂexible hoses, couplings, ﬁlters and

fractured pipes, calling for attention to the installation, location and condition

of all such components. Operators are urged to inspect engine-room fuel sys-

tems, which are also subject to class inspections.

Oil mist can be formed by minute leaks in oil lines which, under pressure,

produce a very ﬁne atomized spray; or it can be formed by oil boiling off in

contact with a hot surface and producing a ﬁne vapour. In the ﬁrst case, the

danger occurs when the particle size formed is between 3 and 10 microns and

allowed to form a mist in the atmosphere. When oil vapour reaches the range

866 Engineroom Safety Matters

of ﬂammability the condition can be classed as truly hazardous and, if no

action is taken, a ﬁre may result. The ignition temperature for this type of oil

mist can be extremely low, depending on the type of oil being atomized.

Oil mists generated by being boiled off can produce particles of between

3 and 10 microns. The mist is visible and known as ‘blue smoke’. The tem-

perature and area of surface contact affect the rate of oil mist generation; at

this stage, a temperature as low as 150°C could result in ignition. Among the

sources of oil mist are pump seals, leaking injectors, loose or incorrectly ﬁtted

pipe elements, weld fractures and poor machinery maintenance. Possible heat

sources causing ignition include heat exchangers, exhaust pipes, turbochargers,

electrical contacts, static electricity and faulty wiring.

Acknowledging that most engine-room ﬁres result from the formation of

oil mist, the IMO’s Maritime Safety Committee in 2003 approved a Code of

Practice for Atmospheric Oil Mist Detectors, giving advice on their location,

setting alarm levels, test procedures and maintenance.

Detection systems can be single sampling units or multiple sampling systems.

In each case, the number of detectors or sampling points depends on the size and

layout of the application. Sample lines in a multiple sampling system (ﬁtted in a

suitable location away from the application) are fed to a common manifold with

a control unit allowing alternative samples to be taken from continuously ﬂowing

sample streams. One unit can thus be used to monitor several points, the oil mist

drawn into the unit by a built-in fan or an independent blower.

To determine suitable positions for mounting detectors or for ﬁtting sam-

pling lines, a smoke test is required to verify air movements in relation to

application. In general, air will move towards ventilation extractors and tur-

bochargers, so any detector or sampling line should be positioned as closely

as possible to the machinery. Similarly, detectors or sampling lines should not

be sited next to ventilation blowers as these will prevent mist formation from

being drawn into the unit.

On installation, the IMO Code of Practice advises, a smoke test should

be carried out with all engines, ventilation and machinery fully operational

to ensure that detectors/sampling lines are correctly positioned. If the detec-

tor units are to be located close to the source of application, care is necessary

to avoid locations where: vibration is excessive; extremes of temperature may

be experienced; access for maintenance personnel is difﬁcult; high levels of

humidity and water may arise; and there is a risk of electromagnetic interfer-

ence. Locating any detector in an explosive atmosphere should not be under-

taken unless the unit is certiﬁed as intrinsically safe for the hazardous area.

PrEvEnting CrankCaSE ExPloSionS

Explosions caused by overheating of surfaces within the engine crankcase or

by leakage of hot gases past the piston have always posed a danger. Among the

components identiﬁed as the cause of crankcase explosions have been over-

heated main bearings, gudgeon pin bearings, chain wheel bearings, cylinder

gland rings seizing on the piston skirt, and seized pistons and piston rods. An

overheated component can convert the normal crankcase atmosphere of air and

oil droplets into an explosive oil mist, the same hot spot then acting as a source

of ignition of the mist. The primary explosion, often of limited severity, may

be followed by a more severe secondary explosion.

Serious crankcase explosions on the British passenger ship Reina del

Paciﬁco in 1947—killing 28 personnel and seriously injuring another 23—

highlighted the dangers and stimulated regulatory action and technical devel-

opments to combat the problem. The four main engines were 12-cylinder

four-stroke trunk piston designs with a combined output of 11 800 kW. The

subsequent court of inquiry held that the primary cause was the piston of No. 2

port outer engine overheating and igniting a ﬂammable mixture present in the

crankcase of that engine.

Prior to the explosions, the engines had been running at full power for

some six-and-a-half hours when it was reported that No. 2 cylinder liner was

hot. The control lever on the engine was brought to the stop position and, after

a few minutes, was moved to the starting position. Shortly afterwards, the ﬁrst

explosion occurred, spreading throughout the crankcase, blowing off the crank-

case doors and igniting the whole of the lower engine-room atmosphere.

It was not clear how the ﬂame from the port outer engine reached the

crankcases of the other three engines. The consensus was that the pressure pro-

duced in the engine-room atmosphere forced the unsupported lower ends of

certain crankcase doors of each engine inwards; and then, when the ﬂammable

vapour inside the crankcase was ignited, the pressure blew the doors outwards.

SOLAS regulations introduced in the 1960s aimed to improve the safety

of marine engines through monitoring by oil mist detection (OMD). Until the

mid-1990s, the adopted standards appeared to be effective as there were no fur-

ther incidents of crankcase explosions reported with severe loss of life.

Ongoing research intensiﬁed during the 1990s, however, when several

major ship owners and enginebuilders began to express concerns that the

number and scale of explosions appeared to be increasing. Estimates suggested

that a minimum of six crankcase explosions were occurring every month

worldwide. Between 1990 and 2001 Lloyd’s Register recorded 143 crank-

case explosions in its classed ﬂeet that caused damage in one form or another.

Of these incidents, 21 occurred in two-stroke engines and 122 in four-stroke

engines. The causes in two-stroke engines were identiﬁed as due to bearings

(43 per cent), pistons (38 per cent) and others (19 per cent); the causes in four-

stroke engines were due to bearings (39 per cent), pistons (47 per cent) and

others (14 per cent). The analysis showed that both engine types are affected in

almost the same way. Higher speciﬁc power ratings and inferior fuel qualities

were cited among the contributory factors.

The International Association of Classiﬁcation Societies’ (IACS)

Machinery Panel was tasked with reviewing the applicability and suitabil-

ity of the two Uniﬁed Requirements—UR M10 and M9—covering crankcase

explosions.

Preventing crankcase explosions 867

868 Engineroom Safety Matters

In January 2005 IACS agreed major revisions to both URs (M10:

Protection of internal combustion engines against crankcase explosions; and

M9: Crankcase explosion-relief valves for crankcases of internal combustion

engines). These URs set uniform standards of installation and functionality for

the equipment—OMD and alarm equipment (M10) and relief valves (M9)—to

be applied by all IACS members.

Statistics from incidents reported to DNV show that main bearing and jour-

nal problems are the most frequent cause of damage in medium-speed main

engines and high-speed engines, with crank bearings and pins the second major

contributor. Damages caused by bearing failure in low-speed engines are con-

siderably lower in percentage terms. An investigation of the causes of bearing

damage identiﬁed the following:

l Lubrication oil supply failure or contamination (from particles or water).

l Connecting rod bolt shear (material fatigue, over-stressing).

l Abnormal wear (ovality, pounding, corrosion).

l Overloading (revving, piston seizure, abnormal vibration).

l Faulty bearing material.

l Incorrect bearing assembly.

l Non-optimum engine construction.

Three different basic approaches to detecting and preventing bearing dam-

age are available: OMD; oil temperature measurement; and measurement of

metal temperatures.

OMD is the most well-known and applied system. Overheating of a sliding

surface under friction can be recognized by detecting the developing oil mist.

When the dry friction continues and the component reaches ignition tempera-

ture the oil mist concentration can rapidly increase to the level of creating a

powerful explosion if not detected early.

Oil mist (sometimes erroneously termed ‘oil smoke’) consists of very ﬁne

oil droplets—colloidal oil—that are suspended in the air, their size in the range

of approximately 1–10 microns. Oil mist that has arisen through mechanical

atomization has particles lying more in the upper size range; oil mist aris-

ing from oil vapour that has condensed again has particle sizes more in the

lower size range. The atmosphere of the crankcase compartment contains

both oil mist from re-condensed oil vapour and from mechanical atomiza-

tion, though the latter can be ignored because of the far greater amount of mist

that is generated even at the start of a damage event through vaporization and

re-condensation.

An important feature of OMD systems is that they cover all heated sur-

faces inside the engine. It is questionable, however, if such systems—being an

indirect bearing temperature measurement method—have the capability to give

the fast response necessary in the event of evolving bearing damage, suggests

Norwegian specialist Kongsberg Maritime.

The fact that engines develop oil mist during normal operation (conden-

sation from evaporating cooling oil returning to the crankcase from pistons,

bearings—including those of the turbocharger—and the blowby gases)

presents a difﬁculty in identifying the start of friction damage. If the setting

of the oil mist detector is too sensitive it can lead to false alarms. The other

extreme—when the detector is set to react at a higher mist concentration—is

also problematic: engine components may reach an advanced stage of damage

and become costly to repair.

An alternative approach is to measure bearing temperatures by measuring

the temperature of the associated lubricating oil using probes (also known as the

‘splash oil method’) or by direct measurement of the bearing metal temperatures.

The compromise here, according to German OMD pioneer Schaller

Automation, is that the temperature probe (such as a PT100) must be very close

to the heat source. In the case of crankshaft bearings in large engines, this is dif-

ﬁcult due to the fact that the bearing area is wide; and in the event of friction the

lubricating oil ﬁrst cools and the heat sink effect of the voluminous metal parts

(bearing cap or shaft) around the probe reduces the efﬁciency of this measure-

ment method. Ultimately, says Schaller, there is an economic restriction on the

number of temperature probes that can be installed in an engine, and not every

sliding surface can be provided with a temperature sensor.

In the splash oil method, the temperature of the lube oil splashing out of

each bearing is measured. Critics argue that a measurement point located at

some distance from the engine wall will not deliver any useful values for the

timely detection of bearing damage. Furthermore, in high- and medium-speed

engines piston cooling oil can be mixed with the splash oil from the crank

bearing, making the temperature measurements less reliable.

By comparing the results, Kongsberg Maritime concludes that a system

based on directly measuring the bearing metal temperatures has a better per-

formance both in terms of response time and measurement accuracy. It there-

fore pursued a system designed to be permanently installed inside an engine to

perform real-time monitoring of bearing metal temperatures (Figure 32.1).

An innovative wireless telemetry technology was exploited to create the

Sentry GB-100 sensor-based system capable of real-time temperature moni-

toring of crosshead and crankpin bearings. A combination of specialized radar

technology and acoustic wave technology facilitates the use of highly reliable

surface acoustic wave (SAW) sensors without wiring or internal power sources.

The passive GB-100 sensors are reportedly capable of working at a range of

1–50 mm between sensor and antenna as well as in harsh environments.

Sentry embraces a signal processing unit (SPU); a wireless temperature

sensor; and a stationary antenna. The SPU generates a low-energy, high-

frequency radar pulse which is transmitted to the wireless sensor via the sta-

tionary antenna. When the wireless sensor passes the stationary antenna the

radar pulse is transferred. The sensor will then immediately reﬂect a pulse sig-

nal, which is received by the SPU via the stationary antenna. The shape and

characteristics of the received signal are then used to determine the temper-

ature of the sensor, this information then being communicated to the engine

control and monitoring system.

Preventing crankcase explosions 869

870 Engineroom Safety Matters

A key merit cited by Kongsberg Maritime is the system’s high ﬂexibility

in arranging the sensor and antenna with respect to the gap, angle and lateral

position between these units. In addition, there is no need to calibrate the tem-

perature sensors and there is no degradation of sensor accuracy over time.

Final development of the Sentry system was carried out in close co-operation

with Wärtsilä, which performed a series of tests including a crank bearing seizure

on a W4L20 medium-speed engine. Seizure was affected by cutting the supply

of lube oil to one of the cylinders while the engine was running at approximately

1050 rev/min. The main conclusion from the test was that the response time of

the system was fast enough to save the crankpin from being damaged.

Several long-term performance and reliability tests were also conducted,

and the ﬁrst installation made on a large two-stroke engine to measure cross-

head-bearing temperatures. Wärtsilä adopted the Sentry system as a standard

ﬁt for temperature monitoring the big end bearings of its new 46F medium-

speed engine. The SPU can be linked to the group’s condition-based mainte-

nance system or any alarm and monitoring system.

Approved by classiﬁcation societies as an alternative to traditional oil

mist detectors, the Sentry system can be installed to monitor the big end and

crosshead bearings of two-stroke engines, such as MAN Diesel and Wärtsilä

designs, at newbuild stage or by retroﬁt.

In the early 1980s, Schaller Automation found that interacting machine

parts in a state of friction and made of different metals (e.g., a steel shaft

running in a bronze bearing) produce a thermo-electric effect which can be

FigurE 32.1 Continuous big end bearing temperature monitoring is performed by

kongsberg Maritime’s Sentry system on Wärtsilä’s 46F medium-speed engine

exploited to produce instant measurable signals. With an appropriate method,

this effect can be measured to provide a diagram showing whether there is a

normal operational condition or serious damage is evolving.

Schaller’s Bearing Overheating Monitoring System (BEAROMOS) was

developed primarily for protecting sliding bearings and surfaces in combus-

tion engines. A sensor is incorporated to obtain information on signals that

become noticeable when the rotating crankshaft is touching the metallic sur-

face of the main bearings. Applied to engines, the basic principle of the system

is to measure the thermo-electric effect produced by a thermocouple formed

by the crankshaft (steel) when sliding on the soft metal alloy of bearing shells.

This physical effect takes place for a few seconds at each start and stop of the

engine when the crankshaft is still making contact with the bearing shell at the

instant that the oil ﬁlm is partly interrupted.

As soon as the oil ﬁlm is complete, the thermo-electric current is inter-

rupted and BEAROMOS registers normal operation. Should the ﬁlm be inter-

rupted again at any moment (e.g., due to dirt or metal particles within the

bearing clearance, or lack of pressure or overheated oil) the sensor will at an

early stage show signals which can be used for taking operational precautions:

a pre-alarm or shutdown procedure. An important aspect of such a measure-

ment, says Schaller, is that when integrated in a system it is free from false

alarms. This is due to the fact that the signal is only generated when the oil

ﬁlm is interrupted and frictional heat begins to develop in a particular bearing.

The XTS-W bearing wear sensor system from UK-based Amot addresses

two-stroke engine applications, promising the capability to detect the onset of

plain bearing problems before expensive secondary damage can occur. The

simple-to-install proximity-based system is claimed to indicate problems well

in advance of temperature sensors or OMD systems.

Wear is detected by the system in the main, crankpin and crosshead bearings

by accurately measuring the vertical position of the guide-shoe at bottom dead

centre: wear occurring in any of these bearings causes the guide-shoe position to

drop. The relevant data is processed by the system, which provides the operator

with a graphic display of current status and trends. A bar chart display shows the

current status: a small green bar is normal; a larger amber bar is a warning that

calls for further monitoring; and a red alarm bar is the ﬁnal stage.

The alarm levels are user-conﬁgurable but the factory-set limits of 0.3 mm

warning and 0.5 mm alarm avoid the potential for nuisance alarms while retain-

ing sufﬁcient headroom on a typical 1 mm-thick bearing shell to allow time

for corrective action. Trend monitoring provides the rate of degradation data

that allows the operator to determine whether immediate action is required or

whether the issue can be addressed at the next port of call.

Delivering this level of functionality makes it necessary to compensate for

normal engine conditions that can affect measurement accuracy, such as engine

load and temperature. The algorithms in the Amot system were reﬁned during

extensive engine testbed and sea trials to provide a repeatability of, typically,

50 microns. A typical XTS-W system may comprise: two 4–20 mA output

Preventing crankcase explosions 871