V Chances of Loss of Merchant Ships .
By W. S. Abell , Chief Ship Surveyor , Lloyd 's Register of Shipping .
( Communicated by Prof. W. Burnside , F.R.S. Received August 14 , 1918 .
) The enormous losses of merchant ships arising from the operations of enemy submarines have caused considerable attention to be directed to the question of water-tight subdivision of such vessels .
One method of obtaining increased protection is to provide additional bulkheads , which would materially prevent loss of vessel if the intactness were likely to be preserved .
In view , however , of the considerable extent of torpedo damage in the longitudinal direction , it is obvious that there is an inferior limit to the spacing of the bulkheads below which the protection afforded is practically useless .
In addition to such objection , a close spacing of bulkheads materially increases the difficulties of loading and unloading , and consequently decreases the time-capacity of the vessel to transport goods .
The correct spacing to obtain the maximum efficiency of transportation must depend on a calculation of the probability of loss of one or more bulkheads , and of the consequent effect of the loss of these bulkheads on the ultimate chances of the sinkage of the vessel .
The first step to be taken is to obtain an approximate estimate of the chance of loss of bulkheads , and for that purpose it may be assumed that the intactness of water-tight subdivision will be destroyed within some radius , R , of the centre of explosion of a torpedo .
The great majority of cargo steamers can only be provided with vertical tranverse bulkheads , and consequently it is only necessary here to consider the aspects of the case where R is a horizontal distance from the centre of explosion .
Case I. Spacing \gt ; 2R = ( 2R 4- a ) .
If the explosion occurs at a distance , R , from the bulkhead , the intactness is destroyed , and consequently within a space ( 2R + \amp ; ) , bounded by a bulkhead at each end , the horizontal length wherein the vessel may be hit without damaging a bulkhead is a , and the length where the contrary effect is produced is 2R .
The " odds on " for loss of a bulkhead are therefore 2R/ a. The properties of the curve 2R/ \amp ; are simple , and when 2R-fa is great compared to 2R the odds on loss are small .
As the value of 2R-f-a decreases , the " odds on " rapidly increase until , where the spacing is within about 15 per cent , greater than 2R , the rate of increase of odds is extremely rapid .
When the spacing is very nearly 2R the " odds on " for loss of one vol. xcv.\#151 ; a. 2 i Mr. W. S. Abell .
bulkhead are , of course , very nearly infinite .
It is therefore important that the spacing of bulkheads shall be greater than 1*15 x 2R , in order that the probability of loss shall not be largely and unduly increased .
Case II .
Spacing \lt ; 2R == ( 2R \#151 ; a ) .
In this case one bulkhead will be damaged in every instance , and it is now necessary to calculate the chances that two bulkheads will be destroyed by one explosion .
Considering the space ( 2R \#151 ; a ) bounded by a bulkhead at each end , the length wherein two bulkheads will be damaged must be the space which is at , or less than , R feet from either bulkhead .
The length of this danger region is obviously a , and the remaining length is ( 2E \#151 ; 2a ) ; consequently , for the loss of two bulkheads the " odds on " are a/ 2R\#151 ; 2a .
The variation of this relation is generally similar to the previous one , except that when the spacing is = 2E , or a = 0 , the " odds on " loss for two bulkheads is zero .
On the other hand , when a approaches E the odds on loss approach infinity ; in fact , as in the previous case , it is desirable to avoid spacing of bulkheads lying between 1*1.5 E and E in order to confine the probability of loss to reasonable limits .
Case III .
Spacing \lt ; E = ( R \#151 ; a ) .
It is obvious , apart from the resistance offered by a bulkhead to distortion under the effects of an explosion , that whenever the spacing is less than E , two bulkheads are rendered ineffective .
Case III resolves itself into a calculation to determine the probability of damaging three bulkheads .
By similar reasoning to that previously adopted , it will be seen that within a region , a , on each side of a bulkhead , it is possible to destroy three bulkheads .
Consequently , since the remaining space is ( R \#151 ; 3a ) , the " odds on " for loss of three bulkheads are 2\amp ; /R \#151 ; 3a .
The form of this fraction follows the same general nature as the previous case , and is subject to the same approximate limitation that spacings between 1*15 ( 2E/ 3 ) and %R/ 3 should be avoided .
Application of Results to Actual Vessels .
As the conditions of flotability vary considerably with the type of vessel , which again is related to the service for which a vessel is designed , it is preferable to confine our attention to ordinary cargo steamers .
Such vessels are usually provided with a forecastle , bridge , and poop , and in the discussion which follows , it is assumed that the freeboard is the minimum permitted under the Load Line Acts .
The spacing of bulkheads , in an actual ship of given dimensions , draught , Chances of Loss of Merchant Ships .
and type , is dependent on the nature of cargo carried and on the permeability , i.e. , the amount of water which will fill the spaces unoccupied by cargo .
It is extremely probable that the conditions obtaining at the time of damage will differ radically from the assumptions made , and there will be in general an increased risk of loss arising from such difference .
The " floodable length , " or the length which may be flooded without loss of ship under assumed conditions of loading and of permeability , bears a fairly constant relation to length of vessel , since the type is fixed .
The floodable length necessarily varies throughout the length , being greater at the ends and amidships than it is in the forward and after holds ; the approximate minimum values of the floodable length have therefore been taken , since these values will give the maximum probability of loss . .
In vessels of the ordinary cargo type the permeability , coefficient of fineness of form , and ratio of sheer of the bulkhead deck to length are practically constant for varying length of ship .
The minimum freeboard permitted by the Load Line regulations varies so that the ratio of freeboard to draught actually increases with length of vessel .
The minimum floodable lengths are obtained from the Board of Trade standard curves , which give the percentage of length which can be flooded in terms of coefficient of fineness of form , ratio of freeboard to draught , and ratio of sheer of bulkhead deck to draught .
These minimum floodable lengths when plotted on a base of length of ship lie approximately on a straight line .
When the bulkheads throughout a ship are so spaced that any one compartment can be flooded and the ship remain afloat , the flotability is defined as a " one-compartment " standard .
When the arrangements are such that any two adjacent compartments can be flooded without loss of vessel , the standard is designated " two-compartment , " and a similar notation is given to higher degrees of subdivision .
Such definitions necessarily imply that the bulkheads bounding the flooded space remain intact .
Since the damage inflicted by a torpedo explosion is limited in extent , it is evident that the larger the vessel the greater the floodable length and the greater the chance of remaining afloat after a hit .
On the other hand , starting from a " one-compartment " standard , where the spacing of bulkheads is the floodable length , and where the loss of one bulkhead means the sinkage of the vessel , it is obvious that the chances of damage of a bulkhead , i.e. , the chances of loss of vessel , continuously increase as the bulkhead spacing diminishes .
The risk of loss from torpedo explosion therefore increases very rapidly until a " two-compartment " standard is obtained .
With the type and freeboard assumed the " two-compartment " standard can only be obtained in two ways:\#151 ; 2 i 2 Mr. W. S. Abell .
Chances of Loss of Merchant Ships .
387 1 .
By building ships of such size that the iloodable length is somewhat in excess of twice the diameter of the explosion circle ; or 2 .
By increasing the freeboard , and thereby deliberately sacrificing the carrying power of the vessel , in order to provide increased protection .
The further discussion of the subject will probably be more intelligible if figures be taken for the various quantities involved .
Numerical Results .
The value of R , the radius of damage from a torpedo explosion , has to be assumed , and for the purposes of the argument it is desirable to keep the estimate on the large side .
Experience indicates that R = 20 feet would be a good approximation .
Fig. 1 illustrates the " odds on " for loss of one , two and three bulkheads , expressed as ordinates on a base of bulkhead spacing .
This diagram shows the desirability of avoiding spacings which lie between 40 and 47 feet , 20 and 23i feet , and 13\#163 ; and 15 feet for the one , two , and three bulkhead losses respectively .
Fig. 2 is a scale diagram giving the approximate relation between Fig-are 2 .
Diagram giving approximate relation between " Floodable Length'\amp ; ^Length of VesseiI* for a Three Island type of Cargo Steamer , with normal sheer \amp ; .
minimum freeboard permitted by the Load Line regulations .
LENGTH or SNIP IN feet Mr. W. S. Abell .
" floodable length 99 and " length of vessel " for a cargo steamer , with forecastle , bridge , and poop of normal extent , with the minimum freeboard permitted by the Load Line regulations .
Fig. 3 is obtained by combining the results of figs. 1 and 2 , and gives as ordinates the " odds on " for loss of vessel , i.e. , the " odds on " for loss of one bulkhead for a " one-compartment " ship , two bulkheads for a " two-compartment " standard , and three bulkheads for a " three-compartment " vessel , Fig* ur 3 .
Curves showing " Odds on " Loss of Vessel .
LENGTH OF SH/ P / /V FEET The following deductions may be drawn :\#151 ; 1 .
Decrease of size of a large vessel only slightly increases chances of loss .
In a " one-compartment " ship , where the bulkhead spacing is equal to the floodable length , the " odds on " for loss of vessel are 1 to 1 for a vessel 460 feet in length , and about 2 to 1 for a length of about 385 feet .
With a " two-compartment " standard , the " odds on " for loss are practically " nil " at about 460 feet , although owing to the limitation of spacing already mentioned , the length of ship should preferably be nearer 500 feet .
For vessels of length of about 340 feet and below , the risk of loss of ship is relatively high , and with very little diminution in length the " odds on " for loss increase very rapidly .
From the diagram , ignoring the resistance Chances of Loss of Merchant Ships .
389 afforded by a bulkhead to distortion under torpedo explosion , it is extremely doubtful whether any subdivision by transverse water-tight bulkheads , however great , is likely to be effective when the length of the vessel is about 320 feet or less than that amount .
3 .
As illustrating the increase of safety arising with the larger vessel it may be observed that:\#151 ; ( a ) A vessel of 460 feet length with the " one-compartment " standard is as saf$ as one of 360 feet length subdivided on the " two-compartment " system .
( b ) A vessel of 500 feet length with bulkhead spacing equal to floodable length is as safe as one of 345 feet length with bulkheads arranged for a " three-compartment " standard .
4 .
The effect of the adoption of intermediate bulkheads in a one-compartment ship ( in order to comply with , say , the two-compartment standard ) , varies considerably with the length of vessel .
Whereas , in the 500-feet vessel the two-compartment standard is practically immune as compared with the one-compartment , yet at 340-feet length the three-compartment vessel is only twice as safe as the two-compartment system , and the latter is only times as safe as with the one-compartment standard .
In addition , spacing of intermediate bulkheads should avoid more particularly the dangerous limits such as 20-23^ feet already mentioned .
General Remarks .
The discussion of this particular phase of the problem suggests that , with reasonable assumptions as to the value of R , it should be possible to arrive at some conclusion respecting the amount of subdivision to be provided to meet conditions attached to ordinary marine risks .
Further , that if any means could be devised to limit the longitudinal extent of damage arising from torpedo explosion , a great increase in safety would result .
For example , if the value of R could be confined to , say 16 feet , as compared with 20 feet , the " odds on " for loss of one-compartment vessels would be reduced by 33 per cent , for a ship of 460 feet length , by 40 per cent , for 400 feet length , and by 60 per cent , when the length is 340 feet .

