I am designing a sewage force main which is about 5 km length and 500mm diameter. The maximum flow is 440 l/s (raw sewage) for an operating pumping head of 60m. I want to model a transient scenario for pump station power failure followed by emergency startup of pumps. I am aiming to run the model for different lag times between power failure and pumps startup to determine the minimum required lag that produces reasonable transient pressures.

I have an intermediate high point along the line (not high to make a siphon), so I placed a combination air valve at that location (air release/vacuum breaker valve). I also have a hyrdopneumatic tank (air chamber) installed at the pump station. The problem is that sub-atmospheric pressure is formed at the high point after power failure. This makes the installed valve ingress huge amounts of air into the pipe through its large orifice to limit sub-atmospheric pressures.

Now, with column separation at the high point, pumps starts operating again increasing the velocity and pressure of the water column upstream the high point. At this point, the air valve expels air to the atmosphere from the large orifice first. This is done at a rate almost equivalent to pipe filling rate till the water columns rejoin and water begins to fill the valve body. After that the mechanism of air release is switched to the small orifice. Very high transient pressures are formed when the two columns rejoin.

The transient pressures of air pocket collapse can be decreased if the air valve expels air at slow rate to allow for air inside the pipe to compress and slow down the water column before collapse. So instead of a combination air valve, I decided to place a vacuum breaker valve next to a single air release valve (with small outflow orifice) at the high point. The vacuum breaker will let air into the pipe through its large orifice to limit sub-atmospheric pressures but will not allow air to escape. When I ran the model, I got the required effect but the vacuum breaker ingressed about 5 cubic meters of air into the pipe. When pressure rises again, the model indicates that pressures and flow returns to normal after a while but the huge air volume is still trapped at the high point.  It seems that the transient model doesn't allow for air to exit the pipe from a point other than the source point.

I think that the following hydraulic behavior will happen, so please correct me if I am wrong.

Most of the the trapped air pocket at the vacuum breaker valve should be swept away with the flow after pump startup and only a small amount of air will remain trapped at the high point without blocking the whole pipe section. Even when air is swept away, it will provide the required effect of slowing down the water columns before collapse. The single air release valve can release the trapped air volume slowly during line operation.

Increasing the lag time between power failure and pumps startup has nothing to do with this situation. I have also tried to increase the size of the hydropneumatic tank at the pump station; doubling its size decreased the trapped air volume to 3 cubic meters instead of 5. I have searched for higher performance combination air valves and found the "D-060-C HF NS" (Combination Air Valve for High Flow