Charles Prelini

CHAPTER XVIII. SUBMARINE TUNNELING (Continued); THE COMPRESSED AIR METHOD.--THE MILWAUKEE WATER-WORKS TUNNEL.

Tunnels excavated at shallow depth from the bed of the river are liable to cave in under the great weight of the water and material above the roof. Besides, the progress of the work will be greatly interfered with by the water which may reach the tunnel passing through the loose soil in large quantities. To contend with these two sources of trouble, different methods of constructing subaqueous tunnels have been devised; they are: by compressed air, by shield, and finally by a combination of these two methods, viz., by shield and compressed air.

The compressed air method was suggested by Mr. Haskin, the promoter and the first builder of the Hudson River tunnel. In 1874, when he began to sink the shaft for the construction of his tunnel, several subaqueous tunnels had already been driven by means of shields. Mr. Haskin had ideas of his own, and thought he could dispense with the shield and could trust to compressed air, since he was firmly convinced that compressed air alone could expel the water and temporarily support the roof of the excavation prior to the building of the lining masonry. In other words, he expected to substitute a core of compressed air for the core of earth removed. In the patent granted him for this method of tunneling, he expresses himself as follows: “The distinguishing feature of my system is that, instead of using temporary facings of timber or other rigid material, I rely upon the air pressure to resist the caving in of the wall and infiltration of water until the masonry wall is completed. The pressure is, of course, to be regulated by the exigencies of the occasion. The effect of such a pressure has been found to drive water in from the surface of the excavation, so that the sand becomes dry.”

The compressed air method was soon found to be inefficient, even in the construction of the Hudson tunnel where the roof of the excavation was supported by timbering in the manner indicated in the pilot system. Thus large subaqueous railway tunnels cannot be driven exclusively by the compressed air method; still it has been successfully employed in the construction of small tunnels driven for aqueduct purposes. But the use of compressed air marked a great progress in the art of submarine tunneling.

THE MILWAUKEE WATER-WORKS TUNNEL.

The following description of the Milwaukee Water-Works Tunnel is an example of subaqueous tunnels driven through good soil in the usual manner employed in land tunnels; but afterward when treacherous material was encountered, the work was continued by means of compressed air.

The new water supply intake tunnel for the city of Milwaukee, Wis., is one of the most difficult examples of tunnel construction which American engineering practice has afforded. The difficulties were in a large measure unexpected when the work was decided upon and put under way. The tunnel began and ended in a hard, impervious clay, practically a rock, and all the preliminary investigations led to the conclusion that the same favorable material would be encountered for its entire length. With such material a brick-lined tunnel 7¹/₂ ft. in diameter presented no unusual problems; but after about 1640 ft. had been excavated from the shore end the tunnel ran out of the hard clay, and for the next 600 ft. or more a variety of water-bearing material was encountered, which tried the skill and patience of the engineer to their utmost. Other difficulties were indeed met with, but these were of minor importance in comparison with that of safely and successfully penetrating the water-bearing drift.

The work of sinking the shore shafts and excavating the first 1600 ft. of tunnel did not prove especially difficult. A hard, compact, and rock-like clay, bearing very little moisture, was encountered all along, and was blasted and removed in the ordinary manner. The only mishap which occurred with this portion of the work was the destruction of the contractor’s boiler plant by fire on Jan. 12, 1891, which allowed the tunnel to fill with water, and delayed work about a month. By Oct. 21, 1891, 1640 ft. had been driven, averaging about 6²/₃ ft. per day, all in the hard clay. No timbering had been necessary, and except for the first 100 ft. of the tunnel there was very little seepage. On the afternoon of Oct. 21 water was observed coming out from one of the drill holes in the heading, but no attention was paid to it. Shortly after a blast was fired, and was immediately followed by a rush of water from the heading. An unsuccessful attempt was made to check the flow, and the pumps were started; but they were unable to keep the water down, and after seven hours’ hard work the tunnel was abandoned. By the next morning the tunnel and shaft were full of water.

Several attempts were made to empty the tunnel; but the limited pumping capacity was not equal to the task, and it was finally decided to install larger pumps. The pumping had, however, shown that about 1000 gallons of water a minute was coming through the leak. With the increased pumping plant the tunnel was finally laid dry Feb. 13, 1892. Upon examination the head of the drift was found to be in the same undisturbed condition in which it was left when the water broke in three months before.

A brick bulkhead was built into the end of the brickwork of the tunnel, and provided with a timber door for passage, and two 10-in. pipes for the outlet of the water. With these openings closed, the flow was checked sufficiently to allow the placing of pumps at the bottom of the shore shaft. Meanwhile the pressure of the water against the bulkhead caused dangerous leakage, and so after the pumps were in position the 10-in. pipes were opened, relieving the pressure and allowing the water its normal rate of flow. Trouble with the pumps now arose, and after various stoppages and breaks the discharge pipe finally fell, disabling the whole plant. It became necessary to close the 10-in. pipes in the bulkhead and draw up the pumps. This allowed the tunnel to again fill with water.

After thoroughly overhauling the pumping machinery, the contractor again laid the tunnel dry on March 19; and after the pumps had been permanently placed so as to take care of the water, an examination of the work was made. It was found that the water was coming from the north, and with the hope of avoiding the difficulties of the old heading, it was decided to make a dÉtour of the south. On April 16 work was begun at a point about 90 ft. back from the face, and deflecting the line about 38° toward the south. About 38 ft. from the angle of junction a brick bulkhead with two 8-in. openings was built into the new bore. The work progressed successfully for about 75 ft., when water was again encountered; and upon pushing forward the heading, gravel and sand came in such quantities that it was found impracticable to continue the work further. On June 1 the bulkhead was permanently closed, and the work in this direction was abandoned.

A further and closer examination was now made of the heading first abandoned. Upon breaking through the rock-like clay it was found that the water came from an underground stream flowing from the north through a well defined channel in red clay. This channel was about 13 ft. above the grade of the tunnel; and above it in every direction visible was a bed of hard, dry, red clay, while immediately in front of the face of the work was a bank of coarse gravel. Fig. 122 is a sketch of the channel and stream where they entered the work. In this last drawing the photograph has been followed exactly, no particular being exaggerated in the slightest. The water from this stream was clear and pure; and a chemical analysis showed that it was not lake water, but must come from some separate source.

Fig. 122.—Sketch Showing Underground Stream, Milwaukee Water-Works Tunnel.

While the engineer did not consider the difficulty of proceeding along the old line insurmountable, it was decided to be less difficult on the whole to go back from 150 ft. to 175 ft. and deflect the line to the north and upward, so as to pass over the underground entrance. Instead of allowing the water to flow at its normal rate and take care of it by pumping, the contractors desired to reduce the pumping, and to this end they constructed a bulkhead just west of the deflection toward the south with a view of shutting off the water. The water, however, accumulated with a pressure of some 50 lbs. per sq. in. and penetrated the filling around the brick lining of the tunnel, preventing the cutting through of the lining for the new line. A second bulkhead was then built about 20 ft. west of the first, but with not much better results, for upon closing it the water was found to leak through the brickwork for a long distance west. Finally on Aug. 2, 1892, the contractors lifted their pumps and allowed the tunnel to fill again with water.

No further work was done on the tunnel by the contractors, although they continued work on the lake shaft for some months. Difficulties had, however, arisen here, which will be described further on; and finally a disagreement arose between the contractors and the city over the delay in prosecuting the tunnel work and over one or two other questions, which resulted in the City Council suspending their contract and ordering the Board of Public Works to go ahead with the work.

The first step to be taken by the engineer was to purchase adequate pumping machinery and empty the tunnel. This was effected Jan. 17, 1894; and as soon as practicable thereafter the two bulkheads were removed and the tunnel cleaned, tram-car tracks laid, and everything prepared for work. It was now determined to go ahead on the original line of the tunnel if possible, and the bulkhead here was removed and work begun. Meanwhile, a safety bulkhead had been built to replace the first one torn away. This was provided with a door and drainage pipes. Work was begun on the original heading, but had proceeded only a little way when the water broke in, driving out the workmen. This was removed three or four times, when the flow suddenly increased to 3000 gallons per minute. An examination of the lake bottom above the break showed that it had settled down, indicating that the new break connected with the lake bottom, and making further work along the original line out of the question.

The question now arose what it was best to do. It was impracticable to use a shield, as the material ahead of the break required blasting, and the pressure from above was enormous. On account of its expense and difficulty of application the freezing process did not seem advisable, and the plenum process was likewise out of the question on account of the great pressure which would be required at this depth. The dÉtour to the south which had been made by the contractor had been unsuccessful, and had left the ground in a treacherous condition. To depress the tunnel was not advisable, for it was not by any means certain that the bed of gravel could be avoided in that way; and, moreover, it would be necessary to ascend again further on, and thus leave a trap which would effectually cut off escape to those at work on the face if water again broke into the tunnel.

It was finally decided that the old plan of deflecting the line toward the north and upward so as to pass over the underground stream should be tried. A hole was therefore cut through the tunnel lining 1433 ft. from the shore, and work was begun on a dÉtour of 20° toward the north and an upward grade of 10%. Fair progress was made on this new line, gradually ascending into solid rock, until May 10, when the test borings, which were constantly made in every direction from the face, showed that sand was being approached. A brick bulkhead was therefore built into the masonry as a safeguard, should it happen that water was encountered in large quantities. As the borings seemed to indicate that the top surface of the rock underlying the sand was nearly level, the lower half of the tunnel was first excavated, leaving about 18 ins. of the rock to serve as a roof (Sketch a, Fig. 123), and the brick invert was built for a distance of 52 ft. The rock roof was then carefully broken through for short distances at a time, and short sheeting driven ahead into the sand, which proved to be a very fine quicksand flowing through the smallest openings. Extreme care had to be taken in this work, but little by little the brickwork was pushed ahead until at a distance of 90 ft. from the point where the sand was first met, and 208 ft. from the old tunnel, the sand stopped and the heading entered a hard clay.

All this work had been done on an ascending grade, and the ascent was continued about 40 ft. farther in the clay. By this time a sufficient elevation was gained to pass over the underground stream, and the tunnel line was changed to head toward the lake shaft, and the grade reduced to a level. The underground stream was passed without trouble and the tunnel continued for a distance of 54 ft. without difficulty. On July 10 the clay in the heading suddenly softened, and before the miners could secure it by bracing, the water rushed in, followed by gravel, filling up solidly some 34 ft. of the tunnel before it was stopped by a timber bulkhead hastily built.

Upon examining the lake bottom a cavity over 60 ft. deep and 10 ft. in diameter was found directly over the end of the tunnel, which had been caused by the gravel breaking into the tunnel. Having now reached an elevation where it was possible to use compressed air, it was determined to put in double air-locks and use the plenum process. The locks were built, and some 670 cu. yds. of clay were dumped into the hole in the lake bottom. On Aug. 4 the air-locks were tried with 26 lbs. air pressure; but, upon a temporary release of the pressure, the water passed around the locks and back of the tunnel lining for some distance, and even forced through the lining, carrying considerable clay and fine sand with it. Upon sounding the lake bottom it was found that the cavity had again increased to a depth of 65 ft., whereupon an additional 600 cu. yds. of clay were dumped into it.

On account of the water leaking through the brickwork, the only dry place to cut through the brickwork and build in an air-lock was just ahead of the brick bulkhead. This lock was completed Aug. 27, and to avoid encountering the danger of the direct connection with the lake at the end of the drift, it was decided to make another dÉtour to the north. On Aug. 28, therefore, the brick on the north side of the tunnel 12 ft. back from the end of the brickwork was cut through under 25 lbs. air pressure, and work proceeded in good, hard clay. The original air-lock was cut out and a new lock built into this clay about 34 ft. from the last dÉtour, to be used in case of further difficulties. After building the tunnel for about 80 ft. from the dÉtour, the soundings again indicated the approach to gravel and water, and on Oct. 14 the water broke through from the bottom in such volume and with such force that the men ran out, closing every air-lock and the valves of every drain in their haste to escape, until the brick bulkhead was reached. It was with great difficulty that the portion of the tunnel up to the last air-lock was recovered and cleaned out.

It was now recognized that a pressure of from 38 to 40 lbs. of air would be needed to hold this water, and accordingly another compressor was added to the plant. With a pressure of 36 lbs. the water was driven out and the work again started. At this time also an additional 350 cu. yds. of clay were dumped into the hole in the lake bottom. Altogether, 1620 cu. yds. of clay had been put into this hole.

Loose gravel and boulders, some of immense size, were now encountered, and the work became exceedingly difficult on account of the great escape of air. The interstices between the gravel and boulders were not filled with silt or sand, but contained water. Moreover, this material extended upward to the lake bottom, as was shown by the escape of air at the surface of the lake. For an area of several hundred square feet the surface of the water resembled a pot of boiling water. At times the air would escape very rapidly; and again only a few bubbles would show.

It need hardly be said that the work in this gravel was very slow. It was impossible to blast or to tear out the large boulders whole, as so much surface would be exposed that an inrush of water would take place despite the air pressure. The method of procedure was to excavate a heading and build the brick roof arch first, and then to take out the bench and build the invert. Fig. 123 gives a number of sketches showing how the work was done. A short piece of heading was taken out, the top and face of the bench being meanwhile plastered with clay (Sketches b and c, Fig. 123) to reduce the escape of air, and then the roof arch was built and supported on side sills resting on the bench. Bit by bit the roof arch was pushed forward until some little distance had been completed, then the heading was plastered with clay and the bench taken out little by little and the invert built. All the gravel except the small area upon which work was actually in progress was kept thoroughly plastered with clay; and as the air escaped through the completed brickwork very rapidly, water was allowed to cover a portion of the invert (see Sketch c, Fig. 123), so as to reduce the area of escape.

When a large boulder was reached, which lay partly within and partly without the tunnel section, the lining was built out and around it, as shown in Sketch d, Fig. 123. The boulder was then broken and taken out. All through this gravel bed the cross-section of the lining is made irregular by the construction of these pockets in the lining to get around boulders. Sometimes they were on one side and sometimes on the other, or on both, or at the top or bottom. In fact, there was no regularity. Despite the hazard and danger of this work, continual progress was made, though sometimes it was only 4 ft. of completed tunnel per week, working night and day; and, if some cases of caisson disease be excepted, the only mishap occurring was a fire which got into the timber packing behind the lining and caused some trouble. From the gravel the tunnel ran into clay and quicksand, and then into hard, dry clay similar to that encountered near the shore. Some difficulty was had with the quicksand, but it was successfully overcome; and when the hard clay was struck, the trouble, as far as the work from the shore shaft was concerned, was virtually over.

Meanwhile, a different set of afflictions had come upon the engineer and contractors in sinking the lake shaft and driving the heading toward shore. This shaft was intended to be built by sinking a cast-iron cylinder 10 ft. in diameter, made up of sections bolted together. Work was begun July 5, 1892, and the sinking was accomplished first by weighting the cylinder, and afterwards by pumping out the sand and water within it until the pressure from the outside broke through under the cutting edge and forced the sand into the cylinder, allowing it to sink a little. From 10 to 30 cu. yds. of sand were carried into the cylinder each time, and finally it was feared that if the process continued, the crib, which had been previously erected, would be undermined. On Sept. 6, therefore, the contractors were ordered to discontinue this method of work. No change was made, however, until Oct. 1, when the cylinder had reached a depth of 68 ft., and by this time there was quite a large cavity underneath the crib. This was refilled, and the cylinder pumped out, and excavation begun inside of it. On Oct. 11 a 2¹/₂-ft. deep ring of brickwork was laid underneath the cutting edge; but in trying to put in another ring beneath the first, two days later, the sand and water broke through the bottom, driving the men out, and filling the cylinder to a depth of 16 ft. with sand. The pumps were started, but the water could not be lowered to a greater depth than 60 ft.

At the request of the contractors, the city engineer had a boring made at the center of the shaft to determine the character of the material to be further penetrated. This boring showed that sand mixed with loam and gravel would be found for a depth of 26 ft., then would come 15 ft. of red clay, and finally a layer of hard clay like that penetrated by the shore end of the tunnel. About the middle of December the contractors made another attempt to pump the shaft, but finding that the water came in at the rate of 25 gallons a minute, abandoned the attempt. In the latter part of February preparations were made to put an air-lock in the shaft and use compressed air. Hardly had the work been begun by this system when, on April 20, 1893, a terrific easterly storm swept the top of the crib bare of the buildings and machinery, and drowned all but one of the 15 men at work there.

This disaster delayed the work for some time, but in June the contractors erected a new building and new machinery, and resumed work. Very little progress was made; and the air escaped so rapidly that it loosened the sand surrounding the shaft and reduced the friction to such an extent that on July 28 the entire cylinder lifted bodily about 6 ft., and sand rushed in, filling the lower part of the cylinder to within 45 ft. of the lake surface. No further work was done by the contractors although they submitted a proposition to sink a steel cylinder inside the cast-iron cylinder and extending from 5 ft. above datum to 100 ft. below datum for $300 per ft. This proposition was refused by the city; and since work on the tunnel proper had been abandoned by the contractors some time before, as had already been described, the city suspended their contract on Oct. 19.

On Oct. 30 a contract was made with Mr. Thos. Murphy of Milwaukee, Wis., to sink a steel cylinder inside the old iron cylinder. The water was first pumped out of the old cylinder, and a timber bulkhead built at the bottom. On this the steel cylinder was built, and then the bulkhead was removed. Air pressure was put on, and the excavation proceeded successfully until the bottom layer of clay was met with, when all chances for trouble ceased.

The cylinder, as it was completed, penetrated 9 ft. into the hard clay, and was underpinned with brickwork for a depth of 29 ft. or more, to a point 4 ft. below the grade line of the tunnel. At the lower end, the section of the shaft was changed from a circle to a square. Later the steel cylinder was lined with brick.

On March 28, 1894, an agreement was made with Mr. Thos. Murphy to construct the tunnel from the lake shaft toward the shore. Except that considerable water was encountered, which, owing to inadequate pumping machinery, filled the tunnel and shaft at two different times, and had to be removed, no very great difficulty was had with this part of the work.

On July 28, 1895, the headings from the lake and shore shafts met. Meanwhile the cast-iron pipe intake, the intake crib, etc., had been completed, and practically all that remained to be done was to clean the tunnel and lift the pumping machinery at the shore shaft. During the cleaning, the air pressure had been kept up on account of the leakage through the brick lining, and, indeed, the pressure was kept up until the last possible moment, and everything made ready for removing the air-locks, bulkheads, pumps, etc., in the least possible time. The pumps were the last to come out.