The ever-present hazard of storage-battery explosions in submarines has never seemed satisfactorily explained. My interest in this subject was increased during a postgraduate course in Diesel engines and storage batteries. In connection with this course an inspection of the Gould Storage Battery Company was made. At that plant, tests were being made on submarine storage-battery cells. A temporary and open ventilation duct had been installed. Sparking occurred within this ventilation pipe. From this observation, I drew the conclusion that the same sparking probably occurred in the ventilation ducts of a submarine. It seemed possible to devise some method of measuring the voltage that caused this sparking. An opportunity presented itself to conduct a research when I was ordered as division engineer of Submarine Division Eleven. Using this sparking as a hypothesis, I endeavored to prove its existence, magnitude, and result.
When a lead-acid battery is charged, about 80 per cent of the current goes into useful work. The specific gravity of the electrolyte rises as the sulphate is forced from the plates. However, a certain portion of the current breaks down the water of dilution in the electrolyte into its component part, H2 and O. As the state of the charge progresses, less current becomes useful energy and more current is available for the breaking down of water electrolytically. Thus it may be seen that hydrogen is present in appreciable quantity. If air contains about a 4 per cent mixture of hydrogen, an explosive mixture is formed. This mixture will not explode unless the temperature is raised by heat as from the presence of a spark or fire.
Records show that there have been twenty-two battery explosions in our submarines. Each explosion has been followed by a board of investigation or a court of inquiry. The findings of these various boards invariably have been: “Explosion occurred due to an explosive mixture of hydrogen and air present in the system becoming ignited and exploding.” The second finding is in almost every case: “The spark necessary to ignite the explosive mixture is always present in a submarine battery.” No board has ever fixed the source of this spark. It has been universally accepted that the spark could come from any number of sources. It is the purpose of this discussion to show that the spark can come from only one definite source.
Because the source of the spark was not fixed, all preventive measures have been along two lines:
- Installation of a blower system of such capacity as to prevent the accumulation of hydrogen in explosive percentage. This system derived its power from a separate circuit to increase reliability.
- Explicit instructions issued by the Bureau of Engineering for charging batteries. These instructions were based on the data obtained from past explosions and test charges.
Despite these precautions, the spark still remains. Should a blower stop at any time, due to failure in the electric circuit, hydrogen accumulates rapidly enough to form an explosive mixture. Every condition necessary for an explosion is present whenever a blower is run too slowly or stops while the batteries are nearing the completion of a charge.
A submarine battery consists of one hundred and twenty cells divided into two groups of sixty cells each. Each group is arranged in four rows of fifteen cells. The ventilation of each group is identical. The cells of a group are placed in a lead-lined tank covered by floor boards. To prevent water from entering the tank a rubber and canvas covering is placed over the floor boards. Brass strips secure this rubber deck cover to the side edges of the lead-lined tank. Air enters the space between the floor boards and the cell tops by means of two air inlets at one end of the tank. The blowers are located at the opposite end of the tank. Each blower is enclosed in a lead-lined pipe that enters a lead-lined box. Two hard-rubber ducts connect to this box. Each duct runs over the tops of the cells between an outboard row and the next inboard row to the air inlet end of the tank. The ducts are closed at this end. Individual cells are connected to these ducts by rubber nipples. When the blowers are running they place a suction on these two ducts. Air enters the air inlets and passes over the cell tops. By means of four breather pipes on each cell the air enters the cell top. It passes over the electrolyte, cooling it, and is drawn out through the rubber nipple into the hard- rubber duct. The air passes from the ducts into the lead-lined box and thence to the blowers. The blowers discharge to a common pipe which exhausts to either the compartment or overboard.
Battery exhaust gases are usually termed evaporation. The product of evaporation is a gaseous matter consisting of water vapor and hydrogen gas holding in suspension minute particles of electrolyte. This mixture is carried away by the ventilation system. The electrolyte usually deposits in the ducts and ventilation system prior to discharge, forming an acid film which is a conductor of electric current.
The external surfaces of a cell cannot be kept free of acid. Due to seepage, spray, watering, and sweating, a moisture path is formed along these external surfaces to the hull of the ship. Hence a moisture ground exists at all times to the hull. As the ventilation discharge overboard is metal and connected to the hull, it is obvious that the ground extends up to this point. If electrolyte is deposited in the vent ducts in sufficient quantity to form a current path, there is no reason why this ground cannot back down this new path along the vent duct surfaces.
The ground-recording instruments are so connected that they read the ground from the external surfaces of the cell to the hull. No device exists to measure the voltage through this new path. In order to determine the possible existence and magnitude of such a voltage ground, a portable voltmeter was used. One lead was placed inside the cell where the rubber nipple connects to the vent duct. The other lead was placed on the metal at the discharge overboard. The existence of a current path was definitely proved as a reading on the voltmeter was obtained on the first trial.
The first reading obtained was 1.4 volts which was prior to starting a charge. After a charge had been started this voltage rose to 65 volts before the end of the first hour. Before the completion of the charge the voltage had increased to 85 volts. Subsequent trials showed the initial voltage to vary from the original 1.4 volts to as high as 40 volts. Final voltages at the completion of charges were obtained as high as 96 volts. On every charge this voltage ground increased as the charge progressed.
Much data was accumulated on tests but it only proved the existence and magnitude of the voltage ground. It was necessary to prove that the path was broken periodically to cause the 85 or more volts to arc with sufficient heat to ignite the explosive mixture. The first theory evolved was expressed as follows:
When the battery blowers stop on a charge, hydrogen accumulates at such a rapid rate that an explosive mixture soon forms. If a blower should be started with such a mixture present, the acid film would suffer the same effect as the surface of a lake when a strong wind blows over it. There would be ripples that would make and break the normal ground path and cause an arcing to take place at the break. As a result of the above arcing an explosion would follow.
This theory became known as the “Ripple Theory." It was sent to the Bureau of Engineering in the above form. It was requested that this theory be tested at the Naval Research Laboratory, Bellevue, Anacostia, D.C. This same letter suggested a device for testing the theory. In the tests at Bellevue a device almost identical to the one suggested was used. Soon after this theory was expounded, Dr. E. G. Lunn of the Division of Physical Chemistry, Bellevue, was on board the U.S.S. Holland. The experiments were later conducted by Dr. Lunn at Bellevue. The theory was discussed with him while he was on the Holland. The additional possibility that certain sections of the duct obtained a thinner coating of electrolyte than other sections gave rise to the fact that the path varied in resistance throughout its length. Thus the PR heat loss at a thin section was probably great enough to evaporate a section and cause a break with the resulting voltage arc.
At a later date correspondence was received from Bellevue requesting detailed information. Test data and additional assumptions were furnished. Emphasis was placed on the fact that, regardless of the rippling effect of the fan, the important feature was the acid film acting as a ground conductor and, hence, due to possible interruption of any kind, furnished a fruitful field for sparking.
To test out these theories Bellevue Laboratory performed the experiments in the order enumerated:
Experiment No. One.—A piece of hard-rubber pipe, similar to the type used aboard submarines, was fitted with internal electrodes, and the electrodes connected to a 110 volt D.C. line. The inside surface of the pipe was then liberally coated with electrolyte and the current turned on. After a few minutes distinct arcing was visible at several places on the inner surface of the pipe. This arcing heated the rubber surface sufficiently to cause it to ignite and char.
From this it was observed that acid film is a conductor of electric current and that unless the coating of electrolyte is being continuously renewed, the film will break causing arcing and evolve considerable heat.
Experiment No. Two.—The above piece of hard-rubber pipe was placed in a bake-lite tube with a fan attached through which it drew air. A lead was constructed so that an explosive mixture of hydrogen could be introduced into the tube. With the fan running, the current of 110 volts on, and the rubber wet with electrolyte, hydrogen was admitted. An explosion immediately occurred. This was repeated several times with identical results. The fan was then stopped and the experiment repeated without the fan running. Explosions were again obtained without the fan running. From the above it was deduced that the action of the air current induced by the fan had no controlling effect on the explosions obtained.
Experiment No. Three.—An experimental apparatus was next constructed simulating on a small scale the ventilating system of a submarine. A full-sized standard submarine duct was not used, because none was available, and because it was felt that the explosions expected would break the apparatus, leaving it unavailable for further experiments. A diagram of the apparatus used is shown in Fig. 1. The apparatus consisted of alternate sections of lead and hard-rubber pipe with fan and flapper valve at opposite ends. Suitable connections for connecting the ground current (110 volts D.C.) along the length of the composite duct, as well as a lead for introducing the explosive mixture, were attached. Moisture was simulated by spraying electrolyte from an ordinary medical atomizer into the duct with the fan running. This was continued until it was observed that the inner surfaces were well coated. Tests were then undertaken as follows:
- Duct dry, ground electrodes disconnected, flapper valve open, fan running. Explosive mixture introduced—no explosion resulted.
- Duct dry, ground connected to 110 volts D.C., flapper valve open, fan running. Explosive mixture introduced—no explosion resulted.
- Duct wet with electrolyte, flapper closed, ground connected, fan stopped. Explosive mixture introduced—fan was then quickly started, at the same instant
opening the flapper—explosions resulted in almost every trial.
Duct wet with electrolyte, flapper closed, fan stopped. Explosive mixture introduced—explosions obtained in the same manner as experienced in (c). In this test a small ammeter was placed in the ground circuit to measure the current flowing through the film. Its reading with the duct wet varied from one-half to two amperes. Just prior to each explosion this ammeter was observed to fluctuate violently, showing that the current path was being interrupted, and arcing presumably going on inside.
Experiment No. Four.—It was next desired to measure the difference of potential at different points along the duct and between the different sections of lead and hard-rubber piping. This was done by boring several small holes in the top of the hard-rubber pipe, to permit the introduction of a wire probe of a voltmeter. The other probe of the voltmeter was connected to a section made of lead pipe. A sketch of this arrangement is shown in Fig. 2. The following readings were obtained:
During the period over which the readings were collected the following conditions existed: duct wet, flapper open, fan running, ground current on, the probe being introduced successively into the holes as above. The violent fluctuation of the voltmeter indicated that the current path was being broken. The voltage dropping to zero at No. 1 while remaining near 110 at points No. 2 and No. 3, indicated the break as being between holes No. 1 and No. 2.
| Voltage | Explosions Obtained |
110 |
| Consistent |
90 |
| Intermittent |
70 |
| Intermittent |
60 |
| Occasional |
50 and | below | None |
The above experiment was repeated with the duct dry but no voltmeter readings were obtained, showing conclusively that the acid film was the current carrier in the previous test.
Experiment No. Five.—In an effort to determine a ground voltage below which no explosions were obtainable, experiments were conducted progressively reducing the voltage across the electrodes shown in Fig. 1. It was appreciated that it was the current through the acid film which directly caused the arcing; however, it was assumed that the potential across the electrodes could be used as an index of this current. Keeping this in mind, experiments were proceeded with, and the following results obtained:
This would then indicate that between fifty and sixty volts is the control voltage desired. In considering these results, however, due regard should be paid to the fact that, with the experimental apparatus, it could not always be assured that an explosive mixture of hydrogen was present in the tube. This was due to wind conditions, and also to the fact that after several explosions leaks developed in the apparatus. However the results obtained form a basis for the hypothesis that a considerable reduction in the battery ground voltage aboard submarines would have a direct effect on the explosion hazard caused by arcing in the ducts.
Summary.—From the experiments undertaken it is felt that the following facts are established:
- That the acid film on the interior of battery ventilation ducts is a conductor of electric current.
- That arcing takes place in the duct shortly after the supply of moisture is stopped.
- That this arcing will ignite an explosive mixture if it is present.
- That the starting or stopping of a blower does not appear to exercise any control over the explosion except that on board a submarine the stoppage of a fan would stop the supply of moisture, thus giving the acid film an opportunity to evaporate, leading up to the condition under (b).
The foregoing experiments conducted at Bellevue seem to be a proof of the assumption that there is only one source of the spark that causes the explosion. The “Ripple Theory” as originally advanced should now be reworded. A careful investigation of the resistance of electrolyte shows that at normal battery specific gravity the resistance is low. Hence, a high I2R loss will result in a section having a thin film. The heat at such a section would be sufficient to drive off vapor and sulphurous fumes so that this section would tend to approach the physical properties of an insulator. Thus a gap would occur at this point in the film path with its resulting spark.
Since the design of a lead-acid battery does not permit the possibility of eliminating gassing, it seems to follow that the logical step in the prevention of battery explosions should be toward the elimination of the film path with the result that the spark would then be eliminated.