The Hydraulic Interaction Between Passing Vessels, Called "Suction."
A Lecture Delivered Before the School of Marine Engineering, U. S. Naval Academy, Annapolis, Maryland, April 17, 1911
By Sidney A. Reeve, M.E., Consulting Engineer.
The subject of suction is regarded by most pilots with considerable fear. They all know, from hearsay, from practice, and from experience that suction exists, and that it is a force which, when uninterrupted, is usually unconquerable. Yet no one seems to be able to lay down the principles of its action for the guidance of pilots. Indeed, to most of them, it is an undefinable force. Out of nearly fifty pilots whom I have heard testify in these cases in court, the majority have never had any experience with a case of suction. Others have had merely one, usually the case in which they were called. So that the ordinary pilot or sea-faring man cannot, from his own experience and judgment, tell when suction is to be encountered and when it is not.
I shall show that the term "suction," which is a term uniformly used, is usually misleading. We speak of attraction, interaction, reaction and repulsion; and repulsion is about as common as is attraction. Repulsion, however, much less often leads to serious results, and so does not get into the records as frequently as attraction in the matter of collisions.
I will first run over briefly a few cases of suction collisions, in order to give you a general idea of how they happen, and then will take up the matter more in detail. The phenomenon has been known on the records from which I quote since 1869. But I should say at the start that what I have to say is not drawn from by any means an exhaustive investigation of the records. I have not had time to do that. The only records which are accessible to the layman for this purpose are the records of the courts, and frequently in the cases coming before the courts suction is not known as a contributing factor to the accident, and so no mention is made of it in the indexes. I have not had time to make an exhaustive search, but I have a sufficient number of instances to illustrate all which we need at the present time.
The phenomenon has increased in later years with the growth in tonnage and speed of vessels, and a corresponding increase in depth of channels. At the same time, the records go back a good many years. The earliest case I have is one which happened in 1869. In that year two Sound steamers, the Narragansett and the Providence left their piers in North River and proceeded around through Hell Gate, the Narragansett being the slower vessel and in the lead. She was overtaken by the Providence, the faster vessel, and then occurred a phenomenon which is a part of the suction principle, and that is that the slower vessel was towed by the faster vessel by hydraulic interaction all the way up through East River. She was enabled to maintain a speed equal to the faster vessel without any change in her steam or engines. When the vessels got into Hell Gate the equilibrium became too unstable and they collided. The slower vessel, as usual, plunged into the quarter of the faster vessel.
The next instance on the list happened in 1872, and was a collision between tugs, the McCandless and the Unit. This was in comparatively deep water. There are no other details given, except that the collision was of the most common type, that is, the slower vessel had been almost overtaken and passed and then plunged into the quarter of the overtaking vessel. I might explain, in going over these, that we can immediately classify them into two types of collision, by attraction and by repulsion. All these which I am running over now are in the attraction class.
The next case happened in 1878 or 1879. A fleet of excursion boats going to one of the International races off Scotland Lightship was led by the tug Hartt; this tug was overtaken by the excursion steamer City of Brockton. As the Brockton went by and had almost passed the tug, the bow of the Brockton being about one hundred feet ahead of the bow of the Hartt, on the starboard hand of the Hartt, the tug took a sudden sheer to starboard, toward the Brockton, then straightened up again. Then the commander put her under a starboard helm, and she took a very sudden plunge into the port quarter of the City of Brockton, running into the port paddle, and was almost overturned.
In 1880 there was a collision off Corlear's Hook, in the harbor, between the tug Imperial going up near the wharves and the ferryboat Garden City, following after. In that case the vessels were on converging courses.
In 1885 occurred the first case of two large trans-Atlantic steamers, the Aurania and the Republic, leaving New York Harbor, just about to enter the turn into Gedney Channel, near the Fairway Buoy, There again the overtaken vessel got under the quarter of the vessel leading. It is interesting to note that just fifteen years later there was a similar collision of two vessels of about the same tonnage, in almost exactly the same place, and under almost exactly similar circumstances, the case of the Martello and the Mesaba. I will take that up later, in detail.
In the '90s there was a collision between the steamer Owen and the smaller steamer Atlantis, in the Detroit River. In 1898 there was also a collision between two freighters in the Delaware River, the Aurole and the Willkommen.
The last one that has come to my attention was a collision between two liners in New York Lower Bay, the aftermost vessel again colliding with the one preceding. That case is more interesting because there was plenty of water, and the vessels were going at half speed, or at about three-quarters speed.
Of the cases where the vessels were diverted, instead of being mutually attracted, as in the above cases, and no collision resulted, the records do not give us such information; because, in a majority of these cases, nothing happened after the diversion. If there happens to be sea-room, the diverted steamer recovers her course, with plenty of time to do so. Occasionally, there are obstacles in the way, and then happens an accident which brings the matter into the records.
The first one of this type occurred in the early '70s, in Hell Gate. The steamer Doris had passed around on the starboard of the tug Minnie, having two barges lashed to the beam. The tug was moving very slowly and the steamer attempted to pass. She had full speed on, driving the tug over on the rocks of the Sunken Meadows, so that the outermost barge was sunk.
In 1874 there also occurred a collision in New Haven Harbor between the steamer C. H. Northam and a tug hauling a string tow. The tug was hauling three barges abreast and following them two more barges abreast. In that case the suction of the steamer dropped the two farthermost barges astern, so that the tow lines were broken, and the stern-wash fetched them up so as to collide with the forward barges, one of which was sunk.
In 1890 occurred the triple accident between the Siberia, the Ohio and the Mather in the Detroit River. Two of the vessels were going in the same direction, at unequal speeds, the Siberia was passed by the Mather, and the Siberia was diverted from her course to such an extent that before she could recover she collided with the Ohio.
The last case on the list is a case so recent that I shall not mention the names of the steamers, the case being still before the courts; but it is most interesting because of the size of the steamers it concerned, and because of the completeness of the data which we have.
I have one instance here of a third class of suction collisions, and that is in which the suction deflects the vessels when they are passing in opposite directions. The courts have always held that when vessels pass in opposite directions, there is not time for the interaction between the hulls to take effect.
If we turn now to the question of the hydraulic theory, you will find that it is comparatively simple. The only hydraulic principle involved you are already familiar with, and that is the principle of the conservation of energy in a mass of water, whereby the sum of the kinetic head and the static head is a constant, although either is readily convertible, within certain limits into the other.
The hydraulic interaction, of course, is produced and governed by the water-flow from the bow of the stern of each vessel, through a temporarily restricted channel in which pressure variations are set up in order to produce motion. In order to bring that to your attention I have put on this diagram an outline of a venturi meter. The chief difficulty, I think, encountered in attempting to discuss these matters with laymen is the one which they have in relation to all hydraulic action except that of direct impulse in the direction of motion of the water. They are constantly looking for the swash of displacement waves (Fig. 2), or the current from a propeller or paddle-wheels and attempt to account for the action by the direct impulse of the current of water. So far as I have been able to understand the matter, that has seldom or never anything to do with these accidents. The only case in which that does play a part, so far as I am aware, is in a case where a large vessel leaves a slip—there have been several accidents of that sort—a slip in which barges are moored. In that case the action of the current from the propellers or the paddle-wheels in a restricted area empties the slip of water, and really brings a fair fraction of the thrust of the engines to bear upon any other vessels in the slip, drawing them away from their moorings and leading to later collisions. In that case, it is a very simple affair of direct impulse from the engines. The speed of the vessels is usually almost zero and has nothing to do with it. Those cases are of no guidance in the avoidance of collisions in navigation; they are to be avoided by the use of common sense.
What we have to deal with in all suction cases is to determine, not the forces alined with the direction of motion of the water, but the forces at right angles to the direction and motion of the ship and the water. That is the point which is hard for the layman to grasp, but you are all of you sufficiently familiar with hydraulics to understand that.
In order to bring out what I have in mind, I have drawn this diagram of the venturi meter (Fig. 3), with the head A, the throat B, and the outlet C. The velocity is comparatively low in entering, and comparatively rapid in the throat. It is then reduced to the original figure at the end. In these diagrams I have used red arrows to indicate velocities and black arrows to indicate forces. There is normally in that conduit, under all velocities, an unlike condition of pressure. As the velocity is increased by the reduction in area at the throat, that static pressure is more or less drawn upon to supply kinetic energy to the water. Ordinarily, it still represents a positive pressure, above atmospheric; but if the dimensions are chosen for the purpose, you can easily have a sub-atmospheric pressure in the throat, shown by the arrows on the outside of the conduit. That is commonly taken advantage of in hydraulic suction-apparatus. The dimensions are made such that there is a distinct deficit of pressure at that point, and then, if it is open here, D, there is a condition in which water will be drawn into the throat of the conduit, although the rest of the conduit may be under comparatively heavy pressure. As the water slows down again that kinetic energy again converts itself into static, and we have the regaining of the internal pressure.
In the passage of a vessel through the water, we have, of course, the displacement of water from bow to stern, in order to permit the passage of the vessel. I have here a chart (Fig. i) in which I will call attention to the displacement of the water, the arrows showing the general direction of the flow. I call attention to the fact that the velocities must be least around the bow of the ship and greater amidships and towards the stern. That is the reason for the practice, in the old sailing ships, of building bluff-bowed vessels. You are well aware that the shape of the stern has a good deal more to do with the speed of the vessel than the sharpness of the bow.
It is those velocities that fall parallel with the axis of the vessel that develop suction between the two ships. If the vessel is in open water, its velocities extend in all directions from the vessel, and of course, theoretically and mathematically, they are inversely as the distance, and so would never become zero. Actually, they are limited to a sensible degree in all vessels. These velocities show the passage of water taking the shortest cut to its destination, except as it develops sufficient resistance.
When two vessels are proceeding in the same direction, the flow on the outer side of either vessel is, of course, normal. But between the two vessels we have action quite similar to the ordinary venturi meter. Here is water between the two stems of the vessd (Fig. 3), and that water has to find its way through this restricted channel between the vessels. Technically, it must have an increased velocity, a positive velocity, which can be determined from the nature and depth of the water, and the distance of the stems apart, and the speed of the vessel. That determines the deficit of pressure amidships. The two hulls of the vessel must develop this deficit of pressure, exactly as do the sides of the conduit of the venturi. That deficit is proportional to the square of the speed 01 the flow through, and that again depends upon the restriction of the channel.
The flow of water, however, cannot be accomplished without a gathering of initial static head to promote it. That is what develops the ordinary body-wave of vessels as they move through the water.
In order to avoid confusion, I wish to state clearly that I am not now speaking of the ordinary bow-waves that go off from either bow of a vessel, the waves of natural equilibrium in water as shown in Fig. 2. The bow-wave is dependent upon the speed of the ship only, and, unless the water is very shallow, it is an equilibrium wave. Nor am I referring to the waves that follow after the vessel, in two curved lines (same Fig.), the "echelon" waves. This wave I now refer to is not a natural wave of equilibrium, but (Fig. 8) is a constrained wave; it is not a true wave at all. It is a conformation of the surface of the water, enforced upon the water by the constraint of the ship, the bottom and the pressure effect. The theory of waves may be said not to apply to this phenomenon at all. This constrained wave, as I have called it, consists in a surplus of pressure above average mean sea-level at this point A, (Fig. 8 and 8a), which must be accomplished in order to set the water into motion. So here at A, we have water pushed ahead of the vessel, and rising as it is pushed—and that represents, by the way, the cross-section of a wave which extends in a fairly straight line, laterally. On river steamers you get the best chance to observe that. I have noticed it in the upper Hudson, where the reaches are fairly restricted. Standing in the bow, and looking ahead along the shore, you can see the surplus or excess following along either shore, away ahead of the vessel. The water is pushed ahead for quite a distance. The height is very little, but the tonnage of water moved is very great.
The principle of ordinary static equilibrium insures the fact that at the point of maximum velocity the level must be lower than mean sea-level. Therefore at that point (B, Fig. 8) we have water moving at minimum velocity. Here (C, Fig. 8) we have water moving at maximum velocity. Here (D, Fig. 8) the water is comparatively at rest; but because the equilibrium has once been disturbed, it does not come back to its original position. Then are developed the natural waves which follow after the vessel, the "echelon" waves, the length of which depends upon their speed. This diagram (Fig. 3) represents one of the positions of unstable equilibrium, the suction-position when two vessels, going in the same direction, have come to this point. The forces then are nearly equal to the surplus pressure outside either ship, and are drawing the ships bodily together. Since ships do not move easily sideways through the water, that particular phase of the situation very seldom has any real effect. Ships sailing side by side are not often brought bodily together, parallel with each other. The only interest in looking at this diagram is that it shows an intermediate phase between the two dangerous phases. That is the case where vessels are of equal length, where their bow-waves are in phase with each other.
The trouble is passing vessels comes, however, when the waves are out of phase—when the stern-waves are out of phase—with crest to trough, and trough to crest.
I have here a diagram, drawn to scale, of one of the liners in this most recent collision (Fig. 8a), on the same scale as this diagram (Fig. 8). This is the slower vessel being overtaken (Fig. 8), and this the larger vessel (Fig. 8a). If I place these diagrams one over the other in this way (Figs. 8 and 8a), the water-lines coincide. That shows the first phase of a larger vessel, the faster vessel, overtaking a smaller vessel. The two crests coincide in this case, consequently the crests are intensified. The two vessels would be repelled one from the other by the accumulation of surplus pressure in between. That phase very seldom produces any dangerous effect.
If we slide the diagram 8a along until we come to this position (D, of Fig. 8, opposite C, of Fig. 8a), we get the first dangerous phase. We now have the crest of the stern-wave of the overtaken vessel abreast the trough of the wave of the overtaking vessel. The result is a dropping away of the pressure on the port bow of the faster vessel, and a surplus pressure on the starboard beam of the slower vessel. These two forces very seldom have any effect. The overtaking vessel is usually the larger, and also the longer vessel. Her tonnage and her moment of gyration about the horizontal axis are much greater than that of the smaller vessel. So the larger vessel is very seldom affected.
The smaller vessel, on the other hand, is affected amidships; the force acting on her has a tendency to draw her bodily sideways through the water to the larger vessel. That, of course, cannot have any appreciable effect. At the stern, however, we find that the stern-surplus of the smaller vessel has offset the natural deficit in lateral force amidships of the overtaking vessel. That cannot affect the overtaking vessel, because it is a force amidships. But on the smaller vessel it makes trouble. This force pulls the water away from the starboard quarter of the overtaken vessel, and sheers her stern in toward the overtaken vessel, and that sends the bow out correspondingly.
That leads to the first class of suction-action, the repulsion of the overtaken vessel. If the vessels continued in that position, and were near enough together, this vessel (Fig. 8) would be driven off to port, and would not ordinarily recover.
In fact, I might explain that the only state of stable hydraulic equilibrium for two vessels passing in this way is when they are at right angles to each other. In other words, this smaller vessel is not in stable equilibrium until it finds itself in the trough of these stern waves, athwart its original course. I heard one pilot testify to being in charge of a steamer going down New York Bay, when he was passed by a larger liner; and he was diverted in that way. Fortunately, he had plenty of sea-room, and it was only necessary to reverse his engines; but by the time the ship was stopped she was out eight points from her original course, right across the channel, and the other vessel had gone on. There was nothing for him to do but work his ship around again. That is, to say, equilibrium is obtained, and the suction-forces cease entirely only when the ships are at right angles.
If we advance these overlapping vessels to the next position (Figs. 8a and 8), we find here the counterpart of this diagram (Fig. 3). This position develops the most serious condition of all; that is, when the point B of the overtaken vessel comes opposite this point C of the overtaking vessel. That, of course, draws water on the port beam of the larger vessel. But this force can have no effect because it acts amidships, and because the vessel is so large; but it does draw the water away from the starboard bow of the overtaken vessel, and gives her a sheer toward the larger vessel. That gives rise to the worst possible state of affairs: because, if, in Fig. 3, instead of drawing the ships in parallel position, we should give either or both of them a slight sheer toward each other, you can see that we would very much restrict the area of cross section in proportion to the volume of water passed through. The most restricted area would not be amidship, but near the bow, and, being near the bow, is would create a deficit 0: pressure between the two vessels, and sheer them still further together.
We would therefore again have unstable equilibrium, the instability of which is increasing very rapidly (instead of decreasing as in the case of the diverted vessel), and which must inevitably lead to collision. The movement of the two vessels toward each other multiplies the force, and the force multiplies the motion. Matters are accelerated in geometric progression. There is not time to do anything. The force very quickly becomes uncontrollable.
Such an action as that— if we suppose this to be the overtaking vessel (Fig. 8a)—would occur when the vessel of Fig. 8a had it beam about opposite the bow of that of Fig. 8; and the instant the vessel of Fig. 8 begins to sheer, she throws off her own water from her bow, and sheers more and more, and usually gives 3 plunge into the other vessel, striking abaft the beam. To make sure that I make myself plain, I will imagine the vessel of Fig. & to have reached the position with C, of Fig. 8, opposite B, of Fig. 8a, and by the time she has got there the vessel of Fig. 8 has sheered over and has struck at D, Fig. 8a. That is the commonest instance of suction-collision.
Here I have drawn a diagram of a vessel being diverted (Fig.7). You can follow the venturi action, the water between these two stems developing an increased force on the beam of the vessel A, and on the quarter of B. The forces set up there—I have drawn the vessel with the helm hard a-port—are such as could not be overcome by any reverse helm action whatever, particularly as the reverse action is attempting to deflect water which has already been deflected in that direction.
That condition of unstable equilibrium is apparently a very complex one. It not only has to do with the constrained waves, but is frequently complicated by the natural waves of the vessel, so that the vessels would pass through two or more phases of unstable equilibrium. That is particularly true when one vessel is considerably longer than the other, so that the body waves of one are longer than those of the other.
This diagram is also drawn to scale (Fig. 6). That is to show, in the first place, the very scant water in which large liners find themselves in the channels of our harbors. There is almost no passageway for water beneath the vessels. There is some space there, but it is of very little use for transmission of water, and practically all the water has to find its way around one side or the other.
These figures show that sixty feet was the natural length of the wave for a boat of this type (Fig. 5). The vessel of Fig. 6, however, has a natural wave length of 130 feet, so that the echelon waves which combine are about two to one. In the large vessels shown that did not have any particular effect. But, in the case of a small vessel like a tug, it would. For instance, the tug Hartt was of about the right length to reach from one crest to the next crest of the natural wave of the City of Brockton, so that the tug was affected not only by the constrained wave, but also at the point where the crest of the natural wave came along. That is the natural explanation why the tug took a sheer toward the Brockton and then straightened up, and then, as the second crest came, took her second plunge and ran into the port paddle-box.
That primary and secondary sheer occurs in a number of these cases, and that is the natural explanation. We might possibly go into it mathematically, but we should not get a much clearer idea; because the one thing we cannot get in suction-cases is quantitative results, or exact data. In the first place, the position of vessels can never be established correctly. Whenever a collision does occur it means that the positions change so very rapidly that the exact point is almost impossible of discovery, so that we get no exact data whatever.
It has, however, brought out one interesting thing, which is that the ordinary seafaring man, even of the best class, is quite incapable of estimating distances over water in yards or feet. He handles his vessel by eye and judgment, in reference to direction and distance, from a light or buoy. But when it comes to translating that into feet, it is pretty hard for him to go into court and tell the exact position of the vessel.
In a case shown on these diagrams, one of the vessels was one of the most prominent liners leaving New York. Her commander came on the stand and testified about her course down New York Bay, and as he did so I charted his testimony. But if his ship had been where he said it was, she would have been dragging mud all the way down the bay. And that is not anything unnatural. It is usually safe, and is the common practice of the court to take a point half way between the testimony of the two sides as nearest the probable truth; but in no case is the data exact.
The only exact investigation work to determine the effects of suction-action has been done by Naval Constructor Taylor, and 1 have had some of his charts reproduced here to show how hi; experimental work verifies the theory of the matter.
Mr. Taylor made his experiments in a tank with two models, each about twenty feet long, and each having the same displacement of 3000 pounds and, with different lines for the midship across section in special cases. These diagrams (Plates 2, 3, 4, 5) show the results he got.
The little arrows show the deflecting forces which he measured In these experiments the vessels were fixed in their courses, and held in position, and were moved in the direction parallel to each other. The deflecting forces could not set up any sheer, of course They were measured in the initial stage of suction only.
These diagrams show the distance between the two vessels. Here the two vessels are far apart, and the forces are .7 at the stern (lower R. H. diagram, PI. 2), and .3 and .4 at the bow. The fractions are the deflecting forces, expressed as ratios in terms of the resistance.
Here (middle R. H. diagram, PI. 2) the models were .24 of their length apart, and the stern forces have been increased to unity. Those of .3 and .4 increased to .6 and .7. Here (upper R. H. diagram PI. .2) they are brought within .19 of their length and the deflecting forces have risen to 1.3 and 1.1 at the bow and 2.1 and 2.2 at the stern.
Mr. Taylor placed the hulls so close together that the water in the tank was quite deep, comparatively, in relation to the size of the models. Here are his experiments with another pair of models having fuller lines, showing the same thing, PI. 3. At seven-eighths of their length apart the forces are very slight, .002 at the stern and .03 at the bow (lower R. H. diagram, PI. 3). Those forces are so slight that they probably do not mean very much. In the upper R. H. diagram, PI. 3, the distance apart is one-half of the length, and the forces have risen to about .05 and .12. In second set from right, PI. 3, are the full line models at distances apart which correspond to those of the fine line models of PI. 2, with corresponding figures.
These diagrams (PI. 4 and PI. 5) show the effect of these forces in the form of curves. Mr. Taylor has taken all of his results in terms of the overtaking vessels. I think that is misleading, because it is only in unusual cases that the overtaking vessel is affected; it is nearly always with the overtaken vessel. He explains, as to this, that the names might be interchanged, and it would simply mean a rearrangement of the diagrams. Not until the position is reached shown in upper .6, .4 and .2L diagram, PI. 3, do we find the repulsion at work on the stern, and attraction on the bow. Out here (upper middle diagram, PI. 3) the bow is under repulsion. An interesting thing appears with the stern under strong repulsion, when the vessels are practically clear of each other (lower .6L diagram PI. 3).
In the positions shown in lower .2, .4 and .6L diagram, PI. 3, the overtaking vessel has passed and is in the lead, and the forces which then act on the two vessels are clearly indicated. None of these diagrams show how rapidly the forces must increase as the vessels sheer toward one another.
The diagram of PI. 4 suggests the same result in another way. The figures on the base-line show the relative position of the two vessels, the two vessels being initially just beginning to overlap. The ordinates are repulsion above and attraction below. As the vessels come abreast there is a turning-moment or tendency- to sheer the vessel from its original course, and that tendency is proportional to the distance between the two curves. When we get to the point shown by ordinate .1 ahead, we get a maximum moment of sheer. The result of that moment is the deflection of the vessel, but if there is plenty of sea-room, there can be no harm.
The diagram of PI. 5 is the corresponding one for two other models. They show the same thing, except in the diminution of the forces, and give corresponding results. You will notice that there is considerable difference between this curve and the one we just examined, although the models are being towed at the same speed, in the same water and at the same distance apart, showing that the lines of the vessel have a good deal to do with it. It is some such effect as that which makes it hard to predict whether a suction-collision is going to take place or not. We have not yet reduced this to numerical form. At the points on these curves shown at .4 astern the moment tending to bring them into collision is much greater than elsewhere, and would make the danger of collision much greater, although an almost equal deflective effect is shown at .4 ahead.
I shall not attempt to bring Mr. Taylor's findings into any concise mathematical laws, but I think it is important to point out that the experimental data, so far as they go, corroborate the simple theory of the matter. So I have taken the figures of Mr. Taylor's experiments and tried to draw a few simple deductions from them.
The thing of primary interest is the distance apart of the two vessels. That, of course, is what every navigator wants to know, how closely he may approach with safety. There again, we cannot give any exact statement to cover all cases; all we can say is that whenever you get in what is known as a suction-position your vessel will be in unstable equilibrium, and you must look out.
However, in following this theory (Fig. 3 of lecture) the deflecting forces ought to be proportional to the ratio between the water trapped between the stems and the area of the passage-way in the center. Let B represent the distance from stem to stem, D the distance between the rails amidships, 6" the deflecting force at the stern in terms of the resistance of the vessel and A a coefficient; then the deflecting force would be proportional to the square root of B over D minus one.
S = deflecting force = A√(B/D) - 1
If I attempt to get a coefficient for A of that expression it will give me some sort of mathematical relationship for the thing. I find from Mr. Taylor's experiments given in PI. 2 that the values for A lie between 0.787, 0.952 and 0.783 ; and from the first three experiments of the second set (PI. 3) I find the figures 0.662, 0.613 and 0.625. That means that these values fall fairly near together. They are in rough agreement, and I draw the inference from that that the value of A is from 0.6 to 0.8 for those models, under varying distances apart. Whether that would have a corresponding value for different models of larger vessels, we have no data at all. It is worth while to say that the matter is consistent and harmonious.
If we abandon that theory and go upon the assumption that you merely find the numerical relationship between results, we find that we can state roughly that when the distance apart between rails is stated as a fraction D of the length, and the total attraction at bow and stern together is stated as a fraction of the resistance we have
A = C/(100D3)
wherein the constant C has a value between 1 and 2. That is purely an arbitrary relationship, but that will give some sort of an idea. A applies, of course, only to models 20 feet long, abreast, moving at 2 to 3 knots, the equivalent of ships 400 feet long, abreast, moving at 9 to 14 knots, in deep water.
Now, again, if we compare the results between these two sets of models, one of full lines and the other of finer lines, we should see from diagram such as this that the effect must be proportional to the water displaced. If we take the ratio between the aggregate midship cross-section for the two models of PI. 2, and that of the models of PI. 3, towed abreast, we find that ratio to be 1.27. If we take a similar ratio between the maximum suction-moments of the respective cases, we find that to be 1.20. That also confirms us in the idea that the explanation which is drawn from the theoretical basis is about correct, and we find that the experimental results, so far, are in consonance with that.
I think it will be interesting to go over some of these actual collision cases and see how they follow out our ideas. The first one on the list is the collision of two Sound steamers at Hell Gate, but we have no data in regard to them which we may discuss. The interesting thing to note is how a slower vessel hung on to a faster vessel and was carried up East River by this hydraulic interaction. That position is one of unstable equilibrium. It is exactly the same thing as a sailing vessel running before a gale, on a high sea. If the commander has that in mind it can be avoided, as a rule; it is a dangerous position, and the slightest lapse of care will lead to serious results. She is not in equilibrium until she is fairly in the trough of the sea; after she has come to a stop she can be gotten on her course again.
The next one was of chief interest because it was the first instance of a primary and secondary sheer and collision between vessels of different dimensions. The next one, the Imperial and the Garden City collision, occurred some thirty years ago near the western shore of the East River, quite close to pier 54 (see chart, New York Harbor), when the ebb tide was coming out. The tug Imperial was running up quite close to the wharves and the ferry-boat Garden City overtook her. The vessels were close together, in deep water. That case was confused by two contributing causes. The courses were slightly converging, and there was a tidal eddy, which would give the tug a sheer toward the Garden City. For that reason the court did not discuss any question of suction-sheer as a primary cause. But to my mind, because the courses were converging, and the distances were very restricted (I think there was only some 50 feet between two rails), the probabilities of suction were very great.
The next case is a larger and more important case, that of the Aurania and Republic, two large liners leaving New York. Your familiarity with the charts will enable you to follow the argument. (See chart, New York Lower Bay, lower main ship channel, lower Swash Channel, out past the Hook and past South Channel, and turning into Gedney Channel). Large vessels coming that was have to make a turn of about four points to get in through the channel, and the width of the channel is restricted for large vessels There have been two collisions at the entrance to Gednev Channel, under identical conditions. The Aurania, the faster vessel, passed the Republic well up in the Upper Bay and came down in the Lower Bay ahead. She had to take the Horseshoe Channel, whereas the Republic got through the Swash Channel and came out ahead. The Aurania overtook the Republic after they had begun to make the turn in the Gedney Channel. We will find that the mean depth to the westward of the buoy runs about 40 feet; just at the turn that shoals quite suddenly to 30 feet. A restriction of depth of water of 10 or 11 feet, when the ground was only 15 feet below the bottom of the ship anyway, is equivalent to a very sudden diminution of throat-area in the venturi meter. That is always a dangerous case, for this reason: The natural speed is less in shallow water than in deep water, and when the ship runs from deep into shallow water the kinetic energy stored in the entire hull converts itself into suction-energy. She runs into a condition which will be very quickly seen to be dangerous.
In cases like this the causes producing the suction-conditions which I have described could not be very well anticipated. The energy which was contributed to deflecting the vessels would be the momentum-energy of the two hulls. The speed would be changed suddenly from the natural speed in deep water to the natural speed in shallow water. The kinetic energy between these two speeds would be converted suddenly into a constrained wave outside the ship. That gives rise to unstable equilibrium of the worst sort.
I may say here that the chance incident which drew me into the whole line of this discussion was that I happened to see something in the daily press about one of these collisions that occurred in the Gedney Channel, and I thought we might get some interesting data if we could place the vessels accurately at that point.
One other collision quite similar to this was that between the Mesaba and the Martello, which occurred in 1900, in the same place, and almost in the same way.
The collision, or rather the encounter, between the two large vessels represented to scale in the diagrams I have shown, the most recent one that occurred, was just at the mouth of the Swash Channel, going down the Lower Bay. The vessel overtaken was on the easterly side of the channel. Just before she reached the mouth of the Swash Channel she was overtaken by the larger vessel. The larger vessel caused the smaller vessel to sheer to the eastward quite a good deal. When the captain of the larger vessel reached the other side at the end of his trip he was very much surprised to find that his own vessel had been libeled for having forced the other vessel ashore.
The general depth in mid-channel in this case is fairly even: but there was a shoal spot at just about the place where the collision occurred, and I should say, in passing, that we might locate the collision at about that spot in the shoal. We were not able to get sufficiently exact data to do this with certainty, however.
In contradistinction to that idea there were two cases in New York Upper Bay, where the water is from 70 to 90 feet deep, and where the vessels were not going at full speed. That shows that suction-action may occur where the water is of considerable depth.
Of the deflections, except for this one I mentioned, I will describe only one. This is a collision on the Delaware River, just below Newcastle, (Southwest Range, see chart, Delaware River, with a shoal, pretty close to the range). A vessel coming down was passed by a vessel, overtaken itself by the Willkommen, which was overtaken by the Aureole. After the Aureole had almost passed, the Willkommen was drawn in, and there was a collision upon the port quarter. The Willkonimen's engines were first slowed and then stopped. Whether that helped or harmed, we cannot say. It is the most natural impulse to shut off power. But the point is. that whereas the depth along the channel is four fathoms, the depth on the outer sides is three and three quarters, and three and a half: and the Willkommen naturally went toward the west to give the Aureole a chance to get through easily without running into shoal water. These boats were both drawing about twenty-four feet, and there must have been five or six feet of tide above the plane of reference, so that there was inshore about 27 feet, and further out about 30 feet.
The case of the Ohio (Fig. 9) was in the Detroit River. There was plenty of water sideways, but not so very much underneath. In that case the Mather overtook the Siberia, going in the same direction; the Siberia sheered to port, under influence of suction, ran off and collided with the Ohio, which was coming down in the opposite direction. That is one of the combinations that must be looked out for. There is a good rule in the west that if you meet two men coming down the pike, never let them pass on both sides of you at once, and it might be the same way in connection with vessels passing.
The only other case I shall speak of at all is that of two vessels passing in opposite directions (Fig. 10). This was the case of the steamer Devereaux, coming down Lake George, St. Mary's River. There was a wide dredged channel which was restricted at a certain point by a natural restricted channel, and then a dredged channel on beyond. The Devereaux coming down passed the tug Folsom towing the schooner Mitchell. When the Devereaux got opposite the Folsom she took a sheer and came into collision with the Mitchell, breaking the tow line and driving her ashore. I have sketched out the dimensions to scale with the cross sections drawn to scale also. The Devereaux is drawn in the position where the court assigned her, but that does not seem quite correct. At any rate she was over near the bank. A vessel near a bank like that is spoken of by the pilots as " smelling the bank." She is deflected from her course, and it is altogether likely that that was the beginning of the phenomena there. The vessels were so close together that I feel quite sure, particularly in view of the position of the Devercanx when she sheered, that suction there was a contributing cause of the collision.
The conclusion which I have tried to draw from this is that we ought to have a better recognition of the subject. We ought to have more data to work with in this matter of collision between vessels. What I have said is all very plausible, but how far it would be justified by experience extended to a large number of cases of different dimensions is not certain. From these I cannot give you any quantitative rules as to how near vessels may approach and at what speed.
We ought to have a definite rule to cover such cases, particularly as steamers are getting larger and the channels are not changing: the dredged channels are multiplying; they are no smaller, but vessels are getting larger and the channels are not, so that suction cases appear to be on the increase. It would not necessarily be an international rule, because these forces occur only in restricted waters, and each country could enforce its own rule, though the rule ought to be international and uniform. I am not a pilot at all, and I would not suggest what that rule should be, except that when an overtaking vessel signals for permission to pass a slower vessel it should be within the power of the overtaken vessel to signal, "I am slowing my engines; I fear danger," and that should enforce upon the larger and faster vessel the slowing of her engines also. I think, in all these cases that I have considered, if both vessels had been at half speed the collisions could have been avoided. It does not do any good for the overtaken vessel to slow her engines if the overtaking vessel passes at full speed.
Extracts from a Letter of Mr. Reeve, Dated November 11, 1911
So soon as I get into the work on the problems connected with suction I feel the need for more data. At the same time I am impressed with the realization of what a splendid machine for the collection of data the U. S. Navy might be, without expense, if directed somewhat to that end.
For instance, two items about which we need data is the thickness of the layer of eddies along the ship's skin, and the outline of the water-level along her side, both observed when the ship is under way at different speeds—or even at her standard speed only.
The midship cross-section of the hull effective for fore and aft displacement of water, as you well know, is greater than that of the metal itself, by a layer of eddies which are dragged along by the ship; but I have not yet run across any data as to the thickness of this layer, amounting to a foot or more, I should judge, on fast vessels, although it is easily observable over the side.
The distance at which suction-effect could be felt, transversely to the ship's course, could be known, if we had data as to the height of displacement-wave, at bow, mid-length and at stern, usual in vessels of her size and speed. But I have seen no collection of such data.
Indeed, the Department may already possess such data about its ships, of which I am not informed. I must go to Washington when I get this work a little further along, to look up a number of such things. If you have any suggestions as to where data may be found I should be grateful for references. Yours very sincerely,
Sidney A. Reeve.