PiPK Laving axu Pipe Line Failures 113 



and the other was laid on its side in shallow water so that the lower 

 quarter of the pipe was immersed. The pipes were withdrawn fre- 

 quently, well wiped with towels, and weighed and measured. The 

 results are shown in the graphs of Fig. 22. The pipe which was im- 

 mersed entirely gained about five percent of its weight by absorp- 

 tion of water, and gained it very rapidly. The increase in weight 

 was practically completed in one and one-half hours. The second 

 pipe gained about one-quarter as much in the same period of time. 

 This indicates that the water does not creep upward through the con- 

 crete by capillarity except at a very slow rate. The expansion curves 

 for the first pipe and for the lower side of the second pipe are almost 

 identical. They exhibit a remarkable lag behind the saturation-time 

 curve, suggesting that the expansion is due to recrystallization. 

 The percentage of elongation is .0005. If this figure is multiplied 

 by the modulus of elasticity for concrete it will give the internal 

 stress which would result in a pipe line with immovable ends. The 

 pipe being of a rich mixture and well aged, the modulus should be 

 about three million, and, therefore, the internal stress equivalent 

 should be at least 1500 pounds per square inch. 



The saturation-time curve for internal wetting would be con- 

 siderably different from that shown in Fig. 22. On account of the 

 dense, hard interior surface of the pipe the absorption of water 

 would be very slow. On the other hand, an internal pressure head 

 of several pounds would tend to increase the rapidity of absorption. 

 The pipe that was immersed over one-fourth of its surface was 

 observed to show a hair crack at the tongue end on the hiside top, 

 extending inward about 5 inches. Another experiment was made, 

 therefore, by taking a 16-inch pipe that had been broken by internal 

 pressure (see Fig. 25) and immersing it over one-fourth of its sur- 

 face. The pipe was laid in the water with the open crack on top. 

 Another broken pipe was immersed over one-half of its surface. 

 These pipes had been provided with inlaid metal measuring points 

 spaced 8 inches apart symmetrically across the top, so that the open- 

 ing or closing of the crack could be measured. The results of these 

 tests are shown in Fig. 23. The saturation of the lower part of the 

 pipes caused the cracks first to close and then to reopen in part. 

 This proves that the absorption of water was largely from the out- 

 side and, as the water penetrated toward the center, the pipe, like a 

 compound spring, closed. When the penetration passed the center 

 of the pipe wall, the spring began to open. 



In the case of a dry pipe line in which water is admitted, the 

 absorption is from the inside, though slow ; and the expansion of the 



