McManus Engineering found evidence of forcemain leaking over the span of several weeks to months.

1. There was no appreciable "spike" in pressure recorded at the pumping station prior to either failure. This indicates that failure was not the result of a sudden increase or decrease of pressure at the pumping station. The delay in the failure time for the two "Y"s may be attributable to a deterioration in the condition of the pipe material through fatigue. Alternatively the failure may have been caused by a local catastrophic increase in pressure created by a transient surge wave, possibly exacerbated by trapped air. Incorrect operating of the air release valves may have contributed to this possible failure scenario.

2. The pressures in the system are broadly in line with the pressures indicated in the 1998 forcemain study and within the limits of the design criteria.

3. The soft start motor installed on the 75 HP pump has appreciably reduced the pressure envelope under which the forcemain operates. This can be expected to significantly reduce surge pressures and contribute to the extended life of the forcemain.

4. The pumps have been operating on 12 minute cycles; i.e. 5 per hour. This short pump shut down/start up time may be insufficient to allow transient surge waves to dampen and dissipate completely. The initial transient analysis undertaken by Proctor and Redfem recommended a minimum of 10 minutes before pump start up follows shutdown.

The pressure/flow data from Richmond Pumping Station indicate that the forcemain system has been operating well within the design parameters of the polyethylene and PVC pipes including the PVC "Y"' s. The_reaspns for the failure of the "Y"' s are therefore not clear at this stage. A number of theories have been postulated:-

2. The thrust blocks used to support the fittings were incorrectly sized or in the wrong location.

pump shut down and start up. I 4. The change in geometry of the forcemain has introduced more fittings

where air may be drawn into the system 5. The operation of the pumps without a "soft starter" increased the

Several methods of repairing the system have been proposed including replacing the PVC "Y"'s with a higher strength material such as cast iron, hyprescon or higher specification PVC. It is our opinion that to design a repair is premature until more work has been done to identify precisely why failure occurred in the "Y'"s.

number is in line with Trow's estimate, based on a simpler formula, of approximately 10,000 Ibs (based on 80 psig steady pressure).

Since transient pressure cycles were to be expected in this system, the basis for designing the thrust block must consider hydraulic transient pressures (which requires a hydraulic transient analysis as presented in Section 4 and which not all designers do in standard practice) and, ideally, transient forces (which most designers seldom consider). EHG's experience in other jurisdictions is that smaller diameter forcemains are often split to two inside a MH structure prior to a crossing, which provides the required rigidity and anchorage but is not as feasible for larger forcemains such as the subject system. In other cases, a vault is provided for the split and an air release valve and MH is provided at the change in profile-just prior to going under the river.

The transient thrusts applied in the longitudinal direction (x on the figures) immediately decrease
(along with system pressure) after a power failure but then increase and reach maximum values
which far exceed 10,000 Ibs and which, in the two 200 HP pump case, change direction very
rapidly several times (x switches from positive to negative on the bottom graph). EHG
understands that a "filter cloth" of some kind was installed between the concrete thrust block and
the PVC Y- fitting. Further, the thrust block did not enclose the fitting but only provided contact
for about half of its circumference. Finally, it does not appear that thrust block restraint was,
provided to prevent longitudinal motion of the pipe.

In wet soil conditions and with repeated pressure (expansion-contraction) and thrust (longitudinal
motion) cycling, it is possible the pipe could have moved a few mm back and forth over time.
This remains true even if one allows for some restraint by friction at the compression gaskets and
due to surrounding soils. Indeed, IPEX correspondence (after the break) indicates the spigot was
found pushed all the way into contact with the PVC bell, not separated by some clearance as
installation standards followed by CAGE construction require. Since the lateral branch can not
move in a longitudinal direction as readily as the straight-through branch of the Y, this motion
provides a mechanism for crack initiation and growth and, ultimately, liquid leaks and the
uncontrolled release of air. EHG and DMEL staff visited with CAGE 'Construction and
inspected one of the broken Y- fittings kept in their construction yard. Small hairline cracks
were observed and photographed along the dividing point on the outside of the pipe fitting
(Figure 4, top).

A photograph showing what appears to be external erosion of the PVC plastic pipe wall (due to a leak) was also examined. It was apparent that pressurized liquid and sand/silt escaping the system through a crack eroded the plastic material in a fan shape, likely over the span of several weeks or months (Figure 4, bottom). With time, the crack likely grew bigger gradually and the PVC material eroded by the leak weakened the Y- fitting (as the pipe wall became thinner). This evidence confirms that a crack occurred on the downstream Y- fitting.