University of Arizona 355 



taken after a cross-section of the pipe had been broken out by the 

 pipe layer. Patching of these breaks was not possible and the long 

 sections were renio\ ed from the trench and replaced. 



The cause of the breaks was not apparent and inquiry among 

 cement pipe men did not throw any light upon the problem. 



Additional trouble was being had with the gate pits which occur 

 at intervals of about a thousand feet along the line, and this trouble 

 was not confined to the. line of 20-inch pipe. In Fig. 6 is shown 

 a broken gate pit. Examination showed that the gate pits were 

 being destroyed by the thrust of the pipe lines caused by longi- 

 tudinal expansion. Most of the pipe had been allowed to become 

 very dry in the stack yard, as is recommended by pipe men. This 

 caused a considerable shrinkage. When water was admitted to the 

 completed pipe line, the pipe walls absorbed it slowly and ex- 

 panded, crushing the gate pits. In experimenting with expansion 

 joints, it was found that these joints must be placed closer than 

 200 feet in order to absorb expansion. 



Some laboratory tests were made to determine the nature and 

 rate of expansion, and its relation to the absorption of water. It 

 was found that the absorption of water under no head was rapid on 

 the outside of the pipe specimens but extremely slow on the inside, 

 the difference being due to the glaze left on the inside in the 

 process of manufacture. Under considerable head the absorption 

 on the inside would be more rapid. The expansion lagged some- 

 what behind the absorption. 



The first longitudinal break occurred on a curve and it was 

 thought therefore that the cause might be longitudinal shear. 

 Mathematical analysis of the problem demonstrated that while this 

 might be true for pipe lines laid on sharp curves, the deviations 

 from a straight line made by a careless pipe layer could not account 

 for the cracks. 



Two 16-inch pipe which had been broken in the internal press- 

 ure testing machine and had subsequently become dry, were further 

 tested in the laboratory by being placed horizontally and immersed 

 to cover the lower one-fourth and lower one-half, respectively. The 

 pipes were placed so that the open cracks were at the top. Absorp- 

 tion from the outside caused the cracks first to close and then to re- 

 open in part. In a similar test on a 14-inch unbroken pipe, the 

 lower side was found to expand almost normally while no expan- 

 sion occurred on the top, and a slight crack opened on the inside 

 at the top at one end. 



While making percolation and internal pressure tests of pipe, 



