JP2002288386A - Risk evaluation method for ship disaster - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a risk evaluation method for ship disasters with high practicality by which the number of disaster development scenarios for performing the simulation of result analysis, is reduced, and yet the certainty of the estimation of life loss is enhanced. SOLUTION: In this risk evaluation method for ship disasters to be used for risk evaluation when evaluating the safety of a ship probabilistically, an event tree is prepared for elements having any influence on results in events which are likely to happen in disasters to be evaluated (S101), and the probability density function of a successful time for disaster suppressing countermeasures is prepared for each event sequence obtained from the event tree (S105), and the elements for classification by case leading to the complication of the event tree are taken out of the event tree (S106). Then, disaster development simulation and evacuation simulation is carried out based on the probability density function (S107). Moreover, the upper limit and lower limit of the risk of each event sequence is calculated based on each simulation result (S110), and the upper limit and lower limit of the whole disaster risks can be calculated from each of the limits of the risk (S111).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for assessing the risk of a ship disaster, in particular, a simulation for estimating the progress of a disaster such as a fire or inundation. The present invention relates to a risk assessment method for ship accidents that evaluates risk by reflecting simulation results.

[0002]

2. Description of the Related Art In the event of a fire occurring at sea, for example, there is no place to escape, and it is extremely dangerous. Therefore, it is necessary to take measures to prevent a fire or to minimize the occurrence of a fire. Therefore, international safety standards for ships have been established. However, unexpected safety accidents can occur even with safety standards established for all conditions. So, in order to make the safety standards more certain,
Safety standards have been revised in the wake of a major accident that has actually occurred. With each revision, ships have been equipped with many additional equipment and strengthened structures based on new safety standards. However, their effectiveness is hardly scientifically proven, and there is a possibility that excessive investments have been made for safety or ineffective equipment has been installed.
There is a danger among concerned parties that this may increase the cost of the ship.

On the other hand, a probabilistic safety evaluation method (PSA: Probabilis) which is regarded as a standard safety evaluation method in the field of nuclear power.
tic Safety Assessment or PRA: Probabilistic
Risk Assessment (also called Risk Assessment) has begun to be applied to other fields, such as chemical plant safety assessment, and is regarded as an excellent method for scientifically assessing safety.

In consideration of the above situation, the 1993 IMO
(International Maritime Organization) / MSC (Maritime Safety Committee) 62 uses the probabilistic safety evaluation method to express the safety of ships in numerical values of "risk" to ensure safety. The UK has expressed its intention to help assess the effectiveness of various standards, as well as assess the costs of setting them. The evaluation method for that is called FSA (Formal Safety Assessment). After that, in the IMO / MSC, a correspondence group and a working group for that purpose were formed, and the study was continued.
A. The provisional guidelines of A were approved, and based on that, each country started to apply FSA on a trial basis. In Japan, the Standards Research Group was established at the Japan Shipbuilding Research Association,
Development of a technique for realizing A is underway. The following (Table 1) shows the outline of the FSA.

[0005]

[Table 1]

[0006] The FSA performs step 1 shown in Table 1 above.
This is a method for deliberation of safety standards consisting of a five-step procedure from Step 5 to Step 5. FSA stands for RCO (Risk Control Option)
: Safety measures) are quantitatively expressed in terms of risk, and RCO
This refers to the procedure for selecting and standardizing the optimal RCO, taking into account the costs added to multiple parties required to install (or implement) the RCO.

[0007] In FSA, the difficult and most important step is the risk assessment of step 2 including the change in risk by the RCO of step 3. Risk is the concept of the product of the probability of an accident and the severity of the consequences.To find it, determine the probability of the accident leading from the danger that leads to the accident, and then the various It is estimated by estimating the progress of disasters, their probabilities, and the rescue process by evacuation and rescue, and evaluating the significance of the results.

That is, in step 2, a fault tree indicating a process leading to an accident, an event tree from the occurrence of the accident to a final result (loss of life, environmental pollution, loss of property, etc.), and a result indicated by each end of the event tree The risk is quantified by quantifying the risk. The current risk can be obtained from the analysis results of the accident statistics already available.

[0009] The RCO changes the risk by affecting various parts of the process from the occurrence of a dangerous situation to the occurrence of an accident, and further from the occurrence of an accident to the final state. Its effect does not appear in accident statistics before implementation, and must be predicted in some other way. In the UK's trial application of FSA, expert judgment is used to estimate the effect of RCO, and the change in risk is determined from expert judgment (RID: Regulatory Impac)
t Diagram) has been created and implemented. This method is simple and applicable to a wide range of RCOs. However, IMO has pointed out that it is arbitrary and uncertain. Therefore, the present inventors have proposed the Transactions of the Shipbuilding Society of Japan, Vol. In 186 (6-3), "Probabilistic Safety Assessment Method for Ships", a risk assessment method with little arbitrariness, which does not depend only on human judgment, has already been announced.

[0010] In this paper, a risk assessment methodology developed based on an analysis of the process from the occurrence of an accident to its consequences for realizing FSA, and a risk assessment system developed based on this methodology are shown. As an example of risk evaluation, the evaluation of a sprinkler, which is a risk control means for preventing fire spread, performed by applying the system is shown. Here, as a method for estimating human life loss due to a disaster at the time of a ship accident, a simulation for estimating the progress of a disaster such as a fire or flooding and an evacuation simulation using the result are performed. This method has the advantage that the RCO is reflected in the simulation as a design change in a natural manner, and there is little arbitrariness.

[0011]

However, according to the conventional risk assessment method as described above, a huge disaster progress scenario (a time series of various state quantities of a ship as a result of performing a simulation in accordance with a certain event sequence). Say)
Need to be simulated over all of the above, which is not practical and requires further study.
In addition, in the case of ship accidents, the life and death of the crew and passengers of the ship is a major problem, and the fire progresses at almost the same speed as the evacuation speed. It is expected that. Therefore, the estimation of human life loss by simulation involves not only the probability of success / failure of fire extinguishing and fire prevention means, but also the time to success in the case of success. However, there is a limit in improving the accuracy of estimating human life loss.

An object of the present invention is to solve the conventional problems as described above, and to estimate the loss of human life while reducing the number of disaster progress scenarios for simulating the result analysis. An object of the present invention is to provide a method for assessing the risk of ship accidents that can improve the accuracy and is highly practical.

[0013]

Means for Solving the Problems To solve the above-mentioned problems, a method for assessing the risk of a ship accident according to the present invention uses a risk of a ship accident which is used for risk assessment when probabilistically assessing the safety of a ship. In the evaluation method, an event tree is created only for elements that directly affect the result among events recalled in the disaster to be evaluated and that mutually affect success and failure, and obtained from the event tree. In addition to creating a probability density function of the success time for disaster prevention measures for each event sequence selected, the same branching will be applied to all the obtained sequences, and the case elements that complicate the event tree will be separated from the event tree. And perform a disaster progress simulation and evacuation simulation based on the probability density function for each combination of those elements. The calculated upper and lower bounds of the risk of the event sequence based on the results of each simulation, in which so as to assess the risk by determining the upper and lower limits of the overall disaster risk from the value of each.

For the probability density function of the success time,
It can be obtained by generating a probability density function such as a normal distribution based on the known minimum success time and the latest success time.

The upper limit and the lower limit of the risk of each event sequence are composed of small sections obtained by dividing the success time in each event into a plurality based on the number of life losses obtained from the simulation for each event sequence. The occurrence probabilities of the upper limit set and the lower limit set of the scenario group formed based on one of the groups in the event are obtained as the scenario upper limit set occurrence probability and the scenario lower limit set occurrence probability, and the scenario upper limit set occurrence probability and the scenario lower limit set occurrence probability are obtained. It can be calculated based on the probability.

The scenario upper limit set occurrence probability (P
_j (U)) and the scenario lower limit set occurrence probability (P
_j (L)) can be obtained by the following equations.

[0017]

(Equation 9)

[0018]

(Equation 10)

The risk (P) is set for each combination (i1, i2,..., In _pl ) of elements belonging to a set (Epl) consisting of case-specific elements for the event sequence (les).
The upper limit (U) and the lower limit (L) of (LL) can be determined by the following equations.

[0020]

[Equation 11]

[0021]

(Equation 12)

The upper limit (PLL _Ies (U)) and lower limit (PLL _Ies (L)) of the risk of each event sequence can be obtained by the following equations.

[0023]

(Equation 13)

[0024]

[Equation 14]

The upper limit of the overall disaster risk (PLL _ALL
For (U)) and lower (PLL _ALL (L)) can be obtained by the following equation.

[0026]

(Equation 15)

( _Where PLL _Ies (U) represents the upper limit of the risk that the event sequence is Ies).

[0028]

(Equation 16)

( _Where PLL _Ies (L) represents the lower limit of the risk that the event sequence is Ies).

As the simulation of disaster progress and the evacuation simulation, for each element, the interval between the shortest success time and the latest success time is divided into M equal parts, and the time points are ordered from the shortest success time. This can be achieved by performing the operations at corresponding time points (hereinafter, these time points are called boundary times).

According to the present invention as described above, an event tree is created for each element that influences the result, and the probability density function of the success time for disaster prevention measures for each event sequence obtained from the event tree is created. Create The probability density function of the success time means that, in addition to success / failure, the time to success in the case of success is added to the estimation of life loss. By adopting the probability density function of the success time, an event tree is created only for the elements that directly affect the result, and the effect is indirectly affected, in order to prevent the number of event sequences to be simulated from increasing. As for the factors that affect the above, the influence is reflected in the form of the success time probability density function. In addition, an element having only “case classification” is taken out of the event tree. Then, a simulation of disaster progress and evacuation simulation are executed based on the probability density function for each combination of those elements, and the upper and lower limits of the risk of each event sequence are obtained based on the results of each simulation, and the respective values are calculated. Assess the risk by determining the upper and lower limits of the overall disaster risk from Therefore, it is possible to improve the estimation accuracy of the life loss while reducing the number of disaster progress scenarios for performing the simulation of the result analysis, and it is possible to obtain a highly practical ship disaster risk evaluation method.

[0032]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to the drawings, taking a fire as an example of a disaster. To explain the outline of the method of creating a fire disaster progress scenario, the risk assessment usually involves creating an ET (Event Tree) after the occurrence of the accident, estimating the branch probability, and obtaining the resulting event sequence (from the initial event). Estimate the damage (estimate the number of lives lost) for each of the branches leading to the final state, which are the same as the number of the final branches of the event tree, and estimate the probability of occurrence of each event sequence and the degree of damage. , FN (Frequency-Number of fatality)
This is done by creating a curve. Estimation of damage in each event sequence, for example, when radioactive leakage from the reactor is estimated, estimation of the amount of leakage, diffusion of radioactivity from the leak, and the effect on human life due to it, by simulation Done.

On the other hand, in the case of a ship accident, the life and death of the crew and passengers of the ship are mainly a problem. Since fires evolve at a speed almost equal to the evacuation speed, it is expected that the number of deaths will change drastically with a slight difference in timing. Therefore, not only the probability of success / failure of fire extinguishing and fire prevention measures,
Time to success matters. In normal event tree analysis, the probability of a branch portion and the distribution function of the probability are estimated, but the time until the event succeeds is not estimated. However, estimation of life loss by simulation includes not only the success / failure of fire extinguishing and fire prevention, but also
If successful, the time to success matters.

Therefore, in the present invention, in order to reduce the number of event sequences to be simulated as much as possible and to make it possible to simulate smoke flow, the event tree is used to manage the main events of fire disaster management.
In other words, it consists of events consisting of fire detection, initial extinction, sprinkler extinction, fire door closure, and hydrant extinction, and the effects of other minor events are reflected in the form of a probability density function of the success time of those major events. By doing so, an event tree was developed. As a result, the overall event tree (GET: Generic Event Tre
e) is configured.

Since the smoke exhaust fan and the like operate independently of a fire disaster event, their effects are modeled and incorporated in a fire simulation in each event sequence. Further, in the related art, one of the factors for increasing the size of the event tree is to put “division” (fire place, initial positions of passengers / occupants) in the tree. Therefore, in the present invention, in order to make the size of the tree appropriate, “case division” which similarly affects any event sequence is taken out of the tree and considered in the result analysis. did. As a result, the number of disaster progress scenarios can be reduced.

FIG. 1 shows the overall event tree (GE
It is explanatory drawing which shows the content of T). Figure 1 is roughly divided
It consists of "case classification", "tree part", and "success time probability density function". The tree part branches according to success or failure in each process (event) in a series of flows that occur at the time of a fire, that is, a flow of fire detection, smoke detection, initial digestion, sprinkler operation, fire door closure, and fire hydrant digestion. The steps are shown in a tree structure. Seven sequences of sequences 0 to 6 are set for “success” and “failure” of the last event and “success” until the middle event. Then, a success time probability density function (p0 to p4 in the figure) is obtained for each sequence, and the time until the event succeeds is estimated based on this. Hereinafter, a description will be given of a smoke flow simulation to be performed for each event sequence of the event tree created as described above, and a method of estimating the probability of occurrence of a fire disaster event sequence corresponding to the simulation results.

First, a method for utilizing the order relation of the criticality of the scenario will be described. FIG. 2 is a probability density function of a normal distribution from the shortest success time (TiE) to the latest success time (TiL) in the intra-event group of each event. It is assumed here that the longer the success of any event, the greater the damage. A small section obtained by dividing the success time in each event into several is defined as an intra-event group, and one intra-event group is extracted from each event, and a combination (= scenario group) composed of these is extracted. When it is made, the order of the degree of success between the scenario groups, that is, the order of the upper limit and the lower limit of the number of life loss is obtained.

FIG. 3 is an explanatory diagram showing the relationship between a group within an event and a successful event. When all the successful events are divided into M event groups, the upper limit of the scenario group where all the event groups are j = 5 in each event (portion surrounded by a thick black frame)
Are larger than the upper limit of the scenario group in which some are “5” and others are less. Therefore, the upper limit of the scenario group in which all the intra-event groups are n can be set as the upper limit of the scenario group in which some are n and others are less than n. Similarly, the lower limit of the scenario group in which all the intra-event groups are n can be the lower limit of the scenario group in which some are n and others are larger than n. Scenario groups that are all n are referred to as a representative scenario group, a set of scenario groups in which the representative scenario group is the upper limit is referred to as a scenario group upper limit set,
The set of scenario groups that is the lower limit is called the scenario group lower limit set. The number of representative scenario groups is M, that is, the number of event divisions. From the above, the risk of the event sequence is obtained by multiplying the number of life loss obtained by simulation at the earliest time and the latest time of the representative scenario group by the probability of occurrence of the upper limit set and the lower limit set obtained by a method described later. be able to.

Next, a case where the result is influenced by before and after the success time of the event will be described. When passengers in a fire compartment that is closed after the fire door is closed start to be blamed, when those evacuees move from each living room to the aisle, they are wrapped in smoke and cannot move. As described above, when the result greatly differs between before and after the procedure, it is not possible to assume that the later the success time of the event is, the greater the number of human lives lost is. In such a case, a combination of events in which such a case occurs is taken up, and the risk is obtained by dividing the case into two depending on before and after the success time. In order to simplify the explanation, it is assumed that there are only two events having different numbers of human lives before and after the procedure (evacuation and fire door closing), and a method of deriving a risk in this case will be described.

The overall fire event tree (GET)
However, such a problem does not exist because the order of events such as fire extinguishing and fire door closing is considered in advance. These problems exist between events that can be considered to occur independently. In practice, it is naturally assumed that the fire door is closed after confirming that all the passengers and occupants in the fire compartment have left the compartment. However, if the fire compartment is filled with smoke and the passengers in the living room are sleeping deeply or are already immobile due to the smoke, spending time on checking will hinder the spread of fire. Therefore, it is necessary to consider that a situation may occur in which the fire door is closed without waiting for confirmation of the departure of the passengers, etc., and it is necessary to consider measures to prevent such a situation.

The probability density function of the evacuation start time is f (x),
Let g (y) be the probability density function of fire door closing success, and x = y
Assuming that a probability density function of t satisfying + t (t> 0) is h (t), h (t) is obtained as in the following equation (1).

[0042]

[Equation 17]

In the region where t ≧ 0 in h (t), the boundary times ts, tsm, tm are set so that the occurrence probabilities are almost equal.
l and tl are determined (see FIG. 4), a probability density function of y is determined so that t becomes those values, and their maximum values are determined as ys,
ysm, yml, and yl. Xs = ys + t
s, xsm = ysm + tsm, xml = yml + tm
Assuming that l and xl = yl + tl, a simulation is performed when the fire door closing success time is earlier than the evacuation start time at those y and x values (see FIGS. 5A to 5D). In addition, in case evacuation start time is earlier than the fire door closed success time, it is assumed that the assumption of the late as slow as the loss of life the number of increases is _{true, x E, x EM, x} M L, x
_L and y _E , y _EM , y _ML , y _L are determined (FIG. 6).
(See (a) and (b)). Incidentally, the sum of the occurrence probabilities of the two events groups in this case is the probability of the case of x <y, the ratio x _E, y _E etc. E at the time of seeking, M, the same as the probability of L .

For the sake of simplicity in calculating the probability of occurrence, for two event groups whose results are expected to differ depending on the context, the combined probability density of the two event groups with a positive or negative difference in their success times The function is divided into two parts, and the probabilities of those parts are defined as p1 and p2 (p1 + p2
= 1), and the division of each part is performed at the same ratio as the division of the other parts.

Next, the achievement procedure in this embodiment will be described. Here, the calculation of the risk for a fire in a living room of a ship is taken as an example. (1) The earliest success time, the latest success time of fire extinguishing and fire prevention events in each event sequence from the time of successful smoke detection,
Listen to success probability experts and create their success time probability density functions. (2) Create a probability density function of the smoke detection success time. (3) Boundary time (TiE (shortest success time), TiEM (medium short success time), TiML (medium delay success time), TiL of each group within the event of each event obtained from (1) and (2) above (Latest success time)) and occurrence probability (Pi
E, PiM, PiL). (4) For each classification of evacuees, the probability density function of the evacuation start time, the occurrence probability of the group within the event, and the boundary time are obtained. (5) If there is a possibility that the evacuation start time is later than the fire door closing success time, for these two events, (i) the difference between the success times of the two events when the evacuation start time is later than the fire door success time And (ii) the probability density function of the fire door closing success time when the fire door closing success time does not become slower, and determine the time point of TiE or the like to be used in the simulation in those two events. . (6) Obtain n initial distributions of evacuees and obtain their occurrence probabilities. (7) For each initial distribution of evacuees, the number of deaths and the scenario group probability are combined, and an evacuation simulation and a smoke flow simulation are performed. It will be determined for each event sequence.) (8) From the results obtained in the simulation of (7), the upper and lower limits of the risk of each event sequence are obtained, and the upper and lower limits of the overall fire risk are obtained from the respective values. With these upper and lower limits, the risk can be evaluated with a certain range, and the overall processing amount can be reduced.

FIG. 7 is a flowchart showing a disaster risk evaluation method according to the present invention. Here, (1)
The flowchart created based on the procedure of (8) is shown. In the figure, S means a step.

First, in the disaster to be considered, elements whose success / failure affects the success / failure of other elements in the disaster to be considered are selected, and a set composed of those elements is described as Edp (S101). As described above, the elements belonging to this Edp include smoke detection, initial fire extinguishing (extinguishing by a fire extinguisher), fire extinguishing by operating a sprinkler, fire door closing, and fire hydrant extinguishing, as described above. Next, an event tree is created from elements belonging to Edp. At this time, the elements belonging to Epl are taken out of the tree to form one table as shown in FIG. 1 (S102). In addition, the FT (Fault Tree)
Define and quantify it, seek expert input,
By such a method, the branch probability of the event tree obtained in S102 is estimated (S103).

Next, for each event sequence obtained from the event tree obtained in S102, the shortest success time (TiE) and the latest success time (TiL) are obtained in order to obtain the probability density function of the success time of the success element. Is heard from an expert (S104). Then, a probability density function of a normal distribution in which TiE is −3σ and TiL is + 3σ is created, and this is set as the success time probability density function of the element (i) (S10).
5).

Next, for each element, the interval between TiE and TiL is divided into M equal parts, and their time points are ordered from TiE.
1, Ti2,... Ti (M + 1). If there is an element that is related before and after the procedure, those success time probability density functions are synthesized, and divided into cases before and after the procedure and cases where it is not, and Ti1, Ti2,... Ti (M +
1) is determined (S106). Next, a disaster progress and evacuation simulation is performed at the time of the corresponding order of the elements (S107). As a result, when the number of event sequences is Nes (total number of event sequences composed of elements belonging to Edp), the total number of simulations is Nall (the number of disaster progress and evacuation simulations, and the following equation (2)) ).

[0050]

(Equation 18)

(In the equation (2), M is the number of divisions of the success time of the success element, and m (j) is the number of possible elements (j) belonging to the element Epl.)

Next, the probability of occurrence of a set of combinations of small sections divided into M equal parts (scenario group upper limit set, scenario group lower limit set) with the number of life loss obtained from the simulation for each event sequence as the upper and lower limits, respectively ( The scenario upper limit set occurrence probability P _j (U) and the scenario lower limit set occurrence probability P _j (L) are calculated by the following equations (3) and (4).
(S108). The simulation may be performed only at the boundary time, that is, only at the upper limit value (latest time) of the M scenario group upper limit set and at the lower limit value (earliest time) of the scenario group lower limit set.

[0053]

[Equation 19]

[0054]

(Equation 20)

Next, the event sequence is Ies (identification number of the event sequence), and the elements belonging to Epl for each event sequence are _ij (j = 1, n
The upper limit and the lower limit of the risk (PLL) in the case of _P1 ) are obtained by the following equations (5) and (6) (S109).

[0056]

(Equation 21)

[0057]

(Equation 22)

Next, the upper limit PLL _Ies (U) and the lower limit PLL _Ies of the risk (PLL) of the event sequence (Ies)
(L) is obtained by the following equations (7) and (8) (S1
10).

[0059]

(Equation 23)

[0060]

(Equation 24)

Next, the upper limit PL of the overall risk (PLL)
L _ALL (U) and the lower limit PLL _ALL (L) are obtained by the following equations (9) and (10) (S110).

[0062]

(Equation 25)

(In the equation (9), PLL _Ies (U)
Represents the upper limit of the risk of the event sequence Ies. )

[0064]

(Equation 26)

(In the equation (10), the PLL
_Ies (L) represents the lower limit of the risk of the event sequence being Ies. )

[0066]

Embodiments of the present invention will be described below. As shown in FIG. 8, it is assumed that a ship is provided with each deck of A, B, C, D, E, and F, and with the conditions shown in the following (Table 2). First, an event tree as shown in FIG. 9 and a probability density function of the success time at each branch (the earliest success time and the latest success time at each event; Assuming a value of ± 3σ).

[0067]

[Table 2]

Next, the relationship between the distance from the center of the smoke detector to the fire source in the horizontal direction and the operation time was modeled by a quadratic equation as shown in the following equation (11), and the center of the smoke chamber was smoked. Assuming that the detector is installed, the probability density function of the time when the smoke detector in the firebox operates is shown in FIG.

T = 5.72r ² + 31.2 + ε μ (ε) = 0.0 σ (ε) = 24.7 (11)

In the above equation (11), t is the time from the start of the fire to the operation of the smoke detector, r is the horizontal distance from the fire source to the smoke detector, and ε is the deviation (assuming a normal distribution).

Next, the boundary time and occurrence probability of each event in each event sequence were estimated. Here, since the success time of each event in each event sequence is the time (Tse) from the operation of the smoke detector, the time (Tig) from the time of the fire to the success of each event in each event sequence is calculated from the time of the fire to the time of the smoke. The time to sensor activation (Tigs) must be added. Therefore, the following equation (12) is obtained.

Tig = Tigs + Tse (12)

As described above, the T of each event_iE, T
_iEM, T_iML, T_iLCan be requested. Ma
T, P_iE, P_iM, P_iLIs determined appropriately. here
Then, P_iE= 0.3, P_iM= 0.4, P_iL= 0.
3, T_iE, T_iLIs 0.5% of time on each side
Then, T of each event for each event sequence_iE, T
_iEM, T_iML, T_iLIn the following (Table 3) to (Table 7)
It began to show.

[0074]

[Table 3]

[0075]

[Table 4]

[0076]

[Table 5]

[0077]

[Table 6]

[0078]

[Table 7]

Next, in order to consider the case where the probability density function of the evacuation start and the order of closing the fire door are reversed, the probability density function of each event is used, The evacuation and fire door closing time to be simulated were obtained.

Next, the initial distribution of evacuees was determined (two types), and their occurrence probabilities were determined. Here, the initial distribution of the occupants and the passengers is divided into two types: a meal when the dining room of the deck C on the trial calculation target vessel is full (3 hours in total) and a case where most of the rooms are in the living room (21 hours remaining). Was.

Next, for each initial distribution of evacuees, EE, E
Evacuation and smoke flow simulations were performed using a combination of M-EM, ML-ML, and LL (see FIG. 9). From the obtained results, the upper and lower limits of the risk of each event sequence were determined, and then the upper and lower limits of the overall fire risk were determined. The results are shown in the following (Table 8). However, in some smoke flow simulations, the blame simulation was carried out on the assumption that the smoke had diverged after a certain time, and in that case, the distribution of the smoke concentration immediately before the divergence continued thereafter.

Thereafter, the above-mentioned (Table 3) to (Table 7) and the Transactions of the Shipbuilding Society of Japan, Vol. 186 (6-3) Obtain the lower limit and upper limit of the risk of the trial calculation ship by obtaining the adjustment values for calculating the fire accident probability and the room fire occurrence probability described in “Probabilistic Safety Evaluation Method for Ships”. Was completed.
The following (Table 8) shows them, including the risk calculation process.

Here, P _ACD is the probability of occurrence of an accident (ACD), pf-I _es (j) belongs to the Edp whose identification number is Ies, and the failure probability p of the j-th element among them.
f-Ies (j) = 1-ps-Ies (j), Nf-Ie
s is the total number of failed elements whose identification number belongs to the Eds of Ies, and ps-Ies (k) is the probability of success of the k-th element among those belonging to the Edp of the identification number Ies, Ns -Ies is the total number of successful elements whose identification number belongs to the Eds of Ies, and Nes = Ns-Ie
s + Nf-Ies, and pIes is the occurrence probability of the event sequence Ies represented by the following equation.

[0084]

[Equation 27]

[0085]

[Table 8]

As a method of estimating human life loss due to a ship accident, when performing a simulation for estimating the progress of a disaster such as a fire or flooding and performing an evacuation simulation using the result, conventionally, a huge amount of It is necessary to carry out simulations for all of a large number of disaster progress scenarios, but according to the present invention,
By eliminating simulations for elements that do not affect the results, and by setting upper and lower limits on the risk PLL for evaluation, the number of disaster progress scenarios can be reduced, and the number of simulations executed accordingly Can be reduced, and it is possible to easily perform the disaster progress simulation and the evacuation simulation.

[0087]

As described above, according to the ship accident risk assessment method of the present invention, there is provided a ship accident risk assessment method used for probabilistic risk assessment of ship safety. Then, an event tree is created for each element that influences the result, and a probability density function of the success time for disaster prevention measures is created for each event sequence obtained from this event tree. The event tree is only used for elements that directly affect the results, so that time is reflected in the loss of life estimate and the probability density function of this success time prevents the number of event sequences from increasing. And, for indirectly affecting elements, reflect the effect in the form of the success time probability function, and
With respect to only those elements, the simulation of the disaster progress and the evacuation simulation are executed based on the probability density function for each combination of those elements, and each event sequence is performed based on the result of each simulation. The risk was evaluated by determining the upper and lower limits of the risk of the disaster and setting the upper and lower limits of the overall disaster risk from the respective values.Therefore, the number of disaster progress scenarios for simulating the result analysis was determined. The accuracy of estimation of human life loss can be improved while reducing the risk, and a highly practical risk assessment method for ship accidents can be obtained.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing the contents of an overall event tree (GET) used in a ship disaster risk evaluation method according to an embodiment of the present invention.

FIG. 2 is a graph showing the relationship between the minimum success time (TiE) and the latest success time (T
It is a characteristic view showing the probability density function of the normal distribution leading to iL).

FIG. 3 is an explanatory diagram showing a relationship between a group within an event and a successful event, used in the risk evaluation method.

FIG. 4 is an explanatory diagram showing a boundary time and a probability density function used in the same risk evaluation method, in which the occurrence probabilities of events having different results before and after a procedure are substantially equal.

5 (a) to 5 (d) are explanatory diagrams showing simulation results when the fire door closing success time is earlier than the evacuation start time based on the probability density function obtained based on FIG. 4 and used in the risk evaluation method. It is.

FIGS. 6A and 6B are explanatory diagrams showing simulation results when the evacuation start time is earlier than the fire door closing success time, which is used in the risk evaluation method.

FIG. 7 is a flowchart showing the risk evaluation method.

FIG. 8 is an explanatory diagram showing a ship assumed in an example of the risk evaluation method.

FIG. 9 is an explanatory diagram showing an event tree used in the embodiment and a probability density function of a success time in each branch.

FIG. 10 is an explanatory diagram showing a probability density function at an operation time of the smoke detector used in the embodiment.

Claims (7)

[Claims] Claims: 1. A method for assessing the risk of a ship disaster that is used for stochastically assessing the safety of a ship, in an event recalled by the disaster to be evaluated, the element directly affecting the result. And create an event tree for only those factors that affect each other's success and failure,
In addition to creating a probability density function of the success time for the disaster prevention measures for each event sequence obtained from the event tree, and a similar division in all the obtained sequences to complicate the event tree, Out of the event tree, for each combination of these elements, perform simulation of disaster progress and evacuation simulation based on the probability density function, and based on the result of each simulation, the upper limit of the risk of each event sequence and A risk assessment method for ship disasters, wherein a lower limit is obtained, and an upper and lower limit of the entire disaster risk is obtained from each value to evaluate the risk. 2. The method according to claim 1, wherein the probability density function of the success time is obtained by generating a probability density function such as a normal distribution based on the known minimum success time and the latest success time. Risk assessment method for ship accidents. 3. An upper limit and a lower limit of the risk of each event sequence are calculated from a small section obtained by dividing the success time in each event into a plurality based on the number of lives lost obtained from the simulation for each event sequence. The probability of occurrence of the upper limit set and the lower limit set of the scenario group formed based on one of the groups within the event is determined as the scenario upper limit set occurrence probability and the scenario lower limit set occurrence probability, and the scenario upper limit set occurrence probability and the scenario lower limit set 3. The method according to claim 1, wherein the risk is calculated based on the probability of occurrence. 4. The scenario upper limit set occurrence probability (P
_j (U)) and the scenario lower limit set occurrence probability (P
_{4. The} method according to claim 3, wherein _j (L)) is obtained by the following equations. (Equation 1) (Equation 2) 5. An upper limit and a lower limit of a risk (PLL) for each combination (i1, i2,..., In _pl ) of elements belonging to a set (Epl) of elements classified into cases for each of the event sequences (les). 5. The method according to claim 1, wherein (L) is calculated by the following equations. (Equation 3) (Equation 4) 6. An upper limit (PLL) of a risk of each event sequence from an upper limit and a lower limit of a risk for each case.
_{6. The method according} to claim 1, wherein _Ies (U)) and the lower limit (PLL _Ies (L)) are obtained by the following equations. (Equation 5) (Equation 6) 7. The overall disaster risk upper limit (PLL)
The risk assessment method for ship accidents according to any one of claims 1 to 6, wherein _ALL (U)) and a lower limit (PLL _ALL (L)) are obtained by the following equations. (Equation 7) ( _Where , PLL _Ies (U) indicates that the event sequence is I
Indicates the upper limit of the risk of es. ) (Equation 8) ( _Where PLL _Ies (L) is the event sequence I
Indicates the lower limit of es. )