Flooding Survivability Analysis - Early Implementation Results

IMO (SLF 47/48) Passenger Ship Safety

2.2 RBD Case Story: Large Passenger Vessel

2.2.2 Early Implementation Results

2.2.2.1 Flooding Survivability Analysis

Restaurant Prescriptive Design

"Evacuability"

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

450 470 490 510 530 550 570 590 610

Egress time (s)

Cumulative distribution

Effect of fire &

smoke scenario Fire Scenario

Fig. 2.57 Prescriptive design – evacuability – normal vs. fire scenario

Table 2.3 Alternative design (ALT 2) – consequences analyses

Spaces Egress time Time to reach UC No. of passengers affected

Upper deck 5–7 min 8–9 min 0

Lower deck 6–7 min 8–9 min 0

• Transient-, cross- and progressive-flooding assessment – Static vs. dynamic stability

– Time to flood

• Time to Capsize

– Probabilistic approach for selection of damage (collision and grounding) cases – Vulnerability approach for survivability assessment

• Systems availability for each flooding scenario

– Geometrical and topological evaluation of main ship systems

• Evacuability assessment

– Assembly and evacuation performance

– Assessment of time to capsize against total evacuation time

• Evaluation of casualty threshold/return to port capability

– Probabilistic approach; link to system availability post-casualty

Statutory Assessment

Acknowledging that emphasis on preventing a casualty from occurring in the first instance must take priority, focus on risk reduction by passive means (in-built safety) must come next and this must start at the beginning. To this end, the dilemma of prescriptive SOLAS-minded designers, illustrated in Fig. 2.58, in the simplest of levels, must be overcome. It is obvious that internal subdivision arrangement is a key issue affecting ship performance, functionality and safety, all of which have to date been catered for through the provision of rules and regulations that reflect, in essence, codification of best practice. Throwing this away and leaving on the table a blank sheet, makes ship subdivision a very difficult problem indeed. This was essentially the problem addressed in the EU project ROROPROB, (ROROPROB 1999–2002).

Building on the understanding of Index-A as outlined in Sect. 3.1, affords a straightforward way of determining the relative collision∩flooding risk profile of a vessel at an early design stage and hence devise an effective means of risk reduction by focusing primarily on the high risk scenarios.

The fully automated optimisation process typically produces several hundred design alternatives depending on the complexity of the ship’s layout and the number of variables. Typical variables of the optimisation problem include: type of subdivision, number, location and height of watertight bulkheads, deck heights, tank arrangement, casings, double hull, and position of staircases, lifts and escape routes.

Using the Attained Subdivision Index, payload capacity, steel weight and other regulatory requirements as typical objectives/constraints, the optimisation problem outcome typically includes: reduced number of bulkheads, reduced deck heights,

Machinery space bulkhead

Collision bulkhead

New requirements for double bottom Aft peak bulkhead

Minor damage concept (still deterministic) for passenger vessels, but no specific requirements on location of watertight subdivision. Required index to be met

A > R

Fig. 2.58 Largely “unguided” subdivision (probabilistic rules)

reduced void volume, reduced number of escape ways and required staircases, reduced steel weight, reduced complexity in tank arrangements, increased crew and service areas, improved functionality and, if required, improved Attained Subdi- vision Index. In order to make the process effective, participation by all decision- makers (designer, owner and yard) is essential to properly define the optimisation variables, objectives and constraints as early as possible in the design stage. Us- ing this approach, known as “platform optimisation”, high survivability internal ship layouts can be developed, without deviating much from the current SOLAS practice, this making it easier for ship designers to relate to the proposed proce- dure. The actual process for platform optimisation as it is currently being applied to

Design Variables -Height of Fb. deck -No., position and

height of bulkheads - Double hull Objectives/Constraints -Maximum payload -A R

NAPA parametric model

Optimisation Modelling Specification

Layout Parametric Model Input

Optimisation Problem Setup

Basis Platform Design Filtering Internal Layout

Optimisation

Outcome

Consultation

Genetic Algorithm Basis Ship

R=0.8

9 Acceptable Designs (A ≥ 0.8)

1,300 designs

Fig. 2.59 Platform optimisation process

0.91 0.915 0.92 0.925 0.93 0.935 0.94 0.945 0.95

0 600 800 1000 1200

0.955

200 400

Feasible Designs Pareto-optimal Designs

Fig. 2.60 Platform optimisation process – concept designs

newbuildings design is illustrated in Fig. 2.59. A sample of the optimisation problem outcome is presented in Fig. 2.60.

Using the hypothetical cruise vessel of Table 2.4 and Fig. 2.61 (Version 1) as a basis, Version 5 (Fig. 2.62) is produced using the process outlined above with A=0.92.

Taking additional measures from the available array of current best SOLAS practice, it was possible to further increase the attained A-Index to 0.985, without sacrificing any of the vessel’s functionality. Time domain simulations with PROTEUS3 have shown that such a vessel would survive all probable damages up to 3-compartment damage for all sea states up to 4 m Hs (the international voyages limit for Stockholm Agreement).

Flooding Vulnerability Assessment

Following from Sect. 3.1, the risk profile of Version 1 of the example cruise ship is illustrated in Fig. 2.63 for all the statistically possible damage scenarios deriving

Table 2.4 Principal particulars of example cruise vessel

Length 270 m

Breadth 35.5 m

Draught 8.3 m

Displacement 56,500 tonnes

Metacentric Height 2.35 m

Number of passengers 2,300

Attained Index of Subdivision, A 0.8 Required Index of Subdivision, R 0.8

–100102030405060708090100120110130140150160170180190200210220230240250270280290300310320330340350360380370260 –100102030405060708090100120110130140150160170180190200210220230240250270280290300310320330340350360380370260 –100102030405060708090100120110130140150160170180190200210220230240250270280290300310320330340350360380370260 –100102030405060708090100120110130140150160170180190200210220230240250270280290300310320330340350360380370260 Fig.2.61Examplecruisevesselsubdivision,ver1

DS /S2 – 4.1.0 DS /S6 – 8.2.0 DS /S8 – 10.0.0 DS /S10 – 12.1.0 – 1 DS /S14 – 16.2.0 0.002

0.004 0.006 0.008 0.01 0.012

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250

2 Comp 3 Comp 4 Comp 5 Comp

3 Comp sample p*(1-s)

–10 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350360 370380

Fig. 2.63 Distribution of relative contribution to risk per damage case, ver 1

from the probabilistic rules for damage stability (Hs, loading condition, collision and grounding – the latter in addition to the current set of scenarios, which relate only to collision damage statistics).

As indicated in Sect. 3.1, these scenarios could be supplemented by using relevant experiential knowledge judiciously and through HAZID/brainstorming ses- sions with designer/yard/owner participating, aiming to identify any design vulnerability. Numerical simulations can then be used in calm water and in waves (as required) to establish the exact flooding mechanism and identify cost-effective changes for the local watertight arrangement using, for example, the PROTEUS3 software suite. The results are analysed in terms of occurrence of potentially dan- gerous behaviour or attitudes by addressing the following three modes of flooding explicitly, on a case by case basis and using a much more complex (in terms of

Proteus 3.1

Fig. 2.64 Typical model used for flooding survivability analysis

number of compartments and number of openings) and more complete model (up to 5 decks are being modelled – see Fig. 2.64):

(a) Initial (transient) Flooding (b) Cross-Flooding

Focus during intermediate stages of flooding targets risk associated with the following hazards:

Transient and Intermediate Flooding. Having to deal with such a complex ge- ometry, explicit dynamic flooding simulation of a damaged ship in waves is a must.

Static analysis simply will not do. Moreover, in some cases where cross-flooding through intricate connection arrangements becomes a problem in terms of long cross-flooding times, results from simplified time-domain simulation codes need to be supported using CFD as the only viable option for a proper treatment of such a problem. The fact that industry appears to be pre-ordained to use static analysis when addressing damage survivability could at best affect adversely the design process and at worst severely undermine safety. Figures 2.65 and 2.66 demonstrate two such cases. In Fig. 2.65, the s-factor results in zero, because the angle of inclination exceeds the statutory range, which does not reflect what actually happens.

Conversely, Fig. 2.66 shows a damage case where the s-factor results in zero based on the SOLAS 2009 formulation whilst numerical simulation results indicate progressive flooding, likely to result in capsize/sinking.

Multi-free Surface Effect. This mechanism of capsize is relevant to ships with complex watertight subdivision such as cruise ships. As the hull is breached, water rushes through various compartments at different levels (Fig. 2.67), substantially reducing stability even when the floodwater amount is relatively small. As a result the ship can heel to large angles, even for small damage openings, letting water into

–18 –16 –14 –12 –10 –8 –6 –4 –2 0 2

0 100 200 300 400 500 600 700

Time [sec]

Roll angle [deg]

STD DOORS CLOSD!

STD DOORS CLOSD!

Fig. 2.65 Numerical simulation of transient flooding behaviour (calculated s=0)

0 –1 –2 –3 –4 –5 –6 –7 –8

0 200 400 600 800 1000 1200 1400 1600 1800 2000

Time [sec]

Roll [deg]

Fig. 2.66 Numerical simulation of transient flooding behaviour (calculated s=1)

Fig. 2.67 Multi-free surface effect during intermediate stages of flooding

the upper decks that spreads rapidly through these spaces and may lead to rapid capsize at any stage of the flooding.

Bulkhead Deck Submergence and Progressive Flooding (Ducting, Piping, Doors, Windows, Shafts, etc). Scenarios of this nature demonstrate the need for explicit knowledge on how the flooding process evolves, as in many cases it proves to be rather straightforward to impede the evolution of flooding with easy and very cost- effective measures. Figures 2.68 and 2.69 show the post-processing that modern tools afford in this quest.

Time to Capsize

The results of the foregoing investigation is analysed in terms of the distribution for the time it takes the vessel to capsize/sink, one of the key parameters in flooding risk estimation. As outlined in the RBD Overview, accounting only for the damage

windows under water windows under water

SWT doors D2 holding SWT doors D2 holding

Fig. 2.68 Time-domain simulation of the flooding process (windows and SWT doors)

D1 D2

Fig. 2.69 Time-domain simulation of the flooding process (various openings)

case scenarios implicit in the new harmonised rules for damage stability (normally over 1,000) and considering the 3 loading conditions, also implicit in the rules, and some 10 sea states per damage case, one would have to deal with tens of thousands of scenarios. Therefore, as indicated earlier, two lines of action are being followed:

the first entails automation of the process using Monte Carlo simulation; the second

Fig. 2.70 Cumulative probability distribution for time to capsize within a given time for two ship layouts shown in Figs. 2.22 and 2.23

relates to the development of a simpler (inference) model for estimating the time to capsize for any given collision damage scenario.

For the example cruise vessel, results using the simpler model are displayed in Fig. 2.70, next.

A close examination of Fig. 2.70 reveals that a 15% increase in Index-A from Version 1 to 5 of the example cruise ship, results in a 60% reduction in the probability to capsize within 3 h. Knowledge of the probability of survival beyond [3]

hours in all relevant flooding scenarios would provide the basis for ascertaining safe return to port capability as shall be explained next.

Casualty Threshold

Putting forward an argument for having a comprehensive risk model and framework to evaluate the total risk of a ship, it is important to appreciate that the aggregate number deriving from such a process, which will represent the safety measure of a ship (safety level), could not be used as the sole criterion for design or approval purposes. Risk components contributing to the total risk will still have to satisfy hazard-specific risk acceptance criteria to avoid undermining safety through what is known as “compen- sation effect” (e.g. having some of the high risk flooding scenarios, say one compartment damage, being compensated by the contribution to aggregate safety level of all the other damage cases, which might satisfy the rules but would still leave the vessel vulnerable and this in itself would not be acceptable). In such cases, it needs to be appreciated that severity and frequency are normally not interchangeable in the high

stakes regime, meaning that individuals and society would (given an explicit choice) not act in accordance with normative decision theory.

In this respect, safeguards ought to be put in place for all component parts contributing to the calculation of this single number, right down to loss scenario level i.e., achieving a high Index of Subdivision by focusing on average safety standards must not be allowed in the knowledge that highly probable damage scenarios are non-survivable.

Therefore, focusing on the casualty threshold with respect to collision damage (damage extent) two complementary lines of action emerge. The first is to min- imise total risk as an aggregate statistic (here equivalent to maximising Index-A) in the knowledge that all components contributing to risk are likely to be reduced.

In addition, high-risk damage scenarios, normally associated with one- and two- component damage ought to be catered for. In fact, the argument could be taken one step further by stating that for design purposes a (large) passenger ship ought to survive up to two-compartment damage in all loading conditions and sea states.

In fact, this was the intention in designing passenger ships under the deterministic SOLAS regime (e.g. SOLAS 1990), despite the fact that in reality this intention is not completely fulfilled. Such a safety-related performance criterion will provide a clear-cut, unambiguous threshold that could serve to reduce uncertainty drastically at a time when a decision to abandon ship in a flooding-related casualty is needed.

Additional information available when the said casualty occurs will help to further aid decision making.

Indeed, an introspective look into the results of the example cruise ship, shown below in Fig. 2.71, reveals that this requirement is not difficult to achieve. Even

Fig. 2.71 Defining a damage threshold

with Index-A of 0.8 the risk contribution of 2-compartment damages is just over 2%, reducing to zero for Version 5(A=0.92). In the latter, even for 3-compartment damages the risk contribution falls below 2%.

At this stage of development, fire-related casualty threshold is much more clear cut, considering that the effect of fire is normally space specific and does not influ- ence the whole ship in the way a flooding scenario might do.

Safe Return to Port

Having explored all issues pertinent to safe return to port, an attempt will be made here to synthesise these in accordance with the IMO framework, Fig. 2.15. Deriving from this, it will be helpful to classify flooding-related casualties as proposed in Table 2.5 below (covering the whole risk space of interest):

The functional requirements corresponding to the same categories can also be classified as shown in Table 2.6.

As a next step, quantifiable performance criteria would need to be developed that reflect the specific functional requirements but from the point of view of decision making in emergency situations, following a casualty, relevant criteria could readily be adopted as shown in Table 2.72.

Adopting the framework presented in Table 2.72 as a basis would reduce a complex and serious problem to manageable proportions, and this could be supplemented by additional information, specific to the casualty in question (actual extent of damage, sea state and so on). The proposed categorisation could be further fine-tuned, particularly concerning the more extensive damages (Categories III-4

Table 2.5 Flooding-related casualty classification Casualty severity Safety objectives Category I

(1-compartment damage)

Vessel remains upright and afloat and is able to return to port under own power (RTP)

Category II

(2-compartment damage)

Vessel remains upright and afloat, but unable to return to port under own power/wait for assistance (WFA)

Category III-3

(3 or more-compartment damage)

Vessel likely to capsize/sink. Abandonment of the ship may be necessary (AS)

Table 2.6 Flooding-related functional requirements classification Systems/functions availability Functional requirements

Category I (Full) All necessary habitability and transport functions available for (5) days

Category II (Partial) All necessary habitability functions available for (5) days Category III-3 (Basic) Basic habitability functions and emergency systems availability

for (3) hours.

Fig. 2.72 Decision making in flooding-related casualties Casualty Severity Design Criteria (Safety- and

Outcome Category I

(1-compartment damage)

P[tc 3 hours] = 100%

Heel, trim limits as per SOLAS Systems Availability

Full

RTP

Category II

(2-compartment damage)

P[tc 3 hours] = 100%

Heel, trim limits as per SOLAS Systems Availability

Full

RTP

Systems Availability

Partial WFA

Category III-3

(≥3-compartment damage)

P[tc 3 hours] = 100%

Heel, trim limits as per SOLAS Systems Availability

Full

RTP

Systems Availability

Partial WFA

P[tc 3 hours] < 100%

Heel, trim limits as per SOLAS

Systems Availability Basic

AS Functionality-related)

and III-5), based on results of purposely undertaken studies and on open discussion and debate at IMO.

Finally, as indicated in the foregoing, functions/systems availability at the “Ba- sic” level is a requisite for all scenarios but what constitutes “Basic” as well as what is considered necessary at “Partial” or “Full” levels is primarily a decision of economics in need of full consideration in the design stage.

Similar developments with regard to fire-related casualties are already embedded in SOLAS.

Dalam dokumen Risk-Based Ship Design (Halaman 81-94)