When computing a transient simulation, no matter what surge protection I try to put in place, it does not seem to help with negative pressures or spikes in pressure.
Pressure "spikes" (large, sudden increases in pressure) tend to occur when a vapor pocket or air pocket collapses. Negative pressures can occur in an event like an emergency pump shut down when flow velocity suddenly decreases, or when a positive pressure wave reflects in such a way that causes a subsequent positive pressure surge.
In general, when modeling a transient analysis, you start by analyzing the system when there is no protection in the system. This allows you to see where the trouble areas are for the system, as well as to determine the severity. This often occurs in the form of vapor forming in areas where vapor pressure is reached. The vapor pocket will collapse and the pressure spike will occur.
Take a close look at transient profile animations of your system. If you're pumping over a hill and the boundary conditions on either end of the system are lower, than it may not be possible to maintain a positive pressure when the pump or pumps remain off. This is because when the pump turns off and the HGL drops, even if you had multiple tanks along the pipeline (such as hydropneumatic tanks or surge tanks), they may help protect the system at first, but will eventually drain out and cause the hydraulic grade to drop to low levels. This can cause vapor pocket formation. When vapor pockets collapse, they can cause severe pressure spikes (or "upsurges"). The vapor pocket collapse or air pocket release may not occur until after the pump has turned back on (which can be modeled using the information here.)
If you only have air valves as protection, it is important to note that they can only limit the pressure from dropping below zero in the immediate vicinity of the air valve. Pressure can still become sub-atmospheric some distance to either side of the air valve. There are a number of factors that come into play, including the physical topology and angle of the surge wave as it approaches the air valve location. In some cases, other protective measures may be necessary, such as a tank or pump flywheel (increased inertia). Another factor to consider is what happens when air is released back out of the air valves. If a controlled air release does not occur (such as with a triple acting air valve or smaller outflow orifice diameter with a double acting air valve) then the adjacent water columns can rejoin too quickly, causing a severe upsurge, which can reflect and combine with other waves, causing severe a downsurge.
The best way to visualize and understand if this is happening is to animate a profile path of the area in question. In your transient solver Calculation Options, make sure you have selected "True" for "Generate Animation data," then open the profile for "Hydraulic Grade and Air/Vapor Volume" for the area of interest. Click the play button at the top or move the time bar to animate the transient simulation and get a better understanding of exactly what's happening. You may notice an air or vapor pocket forming (top graph) and later collapsing with subsequent severe surges forming, reflecting and interacting with each other. You can also generate Time History graphs for particular areas of interest. This will display results over the course of the model run, and can be used in conjunction with the profile animation to get a better understanding of the transient impact.
You may need to consider how long the pumps will be off and size the surge protection device(s) based on that. You can use the "Variable Speed/Torque" transient pump type to simulate the pump turning off and then back on, or consider two runs (one for shutdown and the other for start up). For example, if the pump is shut down for 10 minutes the surge protection device would need to mitigate the transient wave for at least that long. In the case of a surge tank or hydropneumatic tank, the tank will need to be sized appropriately for the water to supply the demands and dampen the transient wave for at least the minimum time the pump is off. There may be concerns with how fast any trapped air is released upon startup.
In some cases you may need to decide if the zero or negative pressures are expected if the system may drain out, and if so, you may need to focus on what happens when the pumps turn back on and expel any air pockets on the downstream end, or at air valves at a high point. If your system ends at an outflow to the atmosphere, use the Discharge to Atmosphere (D2A) element as it supports air inflow upon negative pressure. You could then model a pump shutdown followed by startup to check what happens when the air pocket is fully expelled.
In some cases a system may also experience negative pressure on the suction side (upstream side) during an emergency pump shutdown or startup transient event.
In most cases the pipe length is quite short on the suction side, as there is usually a reservoir connected directly to a pump or set of parallel pumps. When a pump shuts down, the water column on the suction side will suddenly stop, causing an "upsurge", which quickly reflects off the reservoir boundary condition and ultimately becomes a "downsurge" (low pressure) wave that causes negative pressure.
This will be much more likely to occur if the initial pressure on the pump suction side is low in the initial conditions. Meaning, if the starting point is low, it may not take a significant drop in pressure to cause negative pressure to occur. If the negative pressure drops below the vapor pressure limit, vapor pockets can form, introducing a cavitation problem that can damage the pump and piping, and cause subsequent severe upsurge (high pressure) surge effects if the vapor pocket collapses. Typically even in a non-transient simulation the modeler would want to ensure adequate NPSH on the pump suction side, to help reduce the chance of pump cavitation.
It is important to ensure that node elevations are set to the actual pipe elevation, in order to get an accurate calculation of the pressure and if/when vapor pockets occur. If for example you set the pump elevation to ground elevation, the elevation will be higher than it actually is and therefore the pressure will tend to be lower (due to the difference between that elevation and the calculated hydraulic grade). See more here. Also ensure that the upstream boundary reservoir or tank elevation is set to the correct water surface elevation.
Lastly as noted further above, animating the profile path will help you understand how negative pressure may occur on the suction side. To help with this visualization, consider creating a new, "detail" profile view, focused only on the short piping between the upstream boundary and the pump. Also, minimize the artificial increase in reports period using the information here, to ensure you see enough detail in the animated results to be able to understand what is happening. If you still are not able to see enough detail (if the data points in the animation skip around too much), consider temporarily using a shorter simulation duration, just long enough to see the first immediate negative pressure problem on the suction side. This should enable you to prevent excessive report period adjustment, and enable you to potentially use a smaller timestep to see even more detail.
If the elevations and boundary conditions are set correctly, the next factor to examine is pump inertia. A pump shutdown transient simulation assumes that power is disconnected from the pump, and the impeller will slow down based on the pump's inertia. See more explanation here and here. This has a large impact on the change in momentum that occurs from the shutdown and therefore has a large impact on the severely of the subsequent pressure waves (both positive and negative). So, it is important to ensure that you are using the correct inertia value, which should include the inertia of all the rotating equipment as seen here.
See more in this related forum discussion: Negative pressure equal to vapor pressure developing at pump suction head readings only found in extreme pressure report
Transient pressure worse with an air valve added