It is caused when a moving fluid is stopped rapidly (perhaps a valve closed very rapidly). A moving fluid has momentum and if a valve closes very quickly, the momentum of the fluid is transferred to the pipe. The pipe or components connected to the pipe can react as if being hit by a hammer and may recoil causing the shock wave to bounce back and forth a few times. You can minimize water hammer by reducing sudden changes in fluid velocity or by providing some sort of cushion where flow can be redirected and the fluid's kinetic energy to disperse more slowly. Plumbers will often leave a section of pipe filled with air (although over time the air will dissolve into the water). An air bladder would work much better and allow the energy to be dissipated over a longer period of time instead of rocking the pipe and fittings.

The idea is to increase the period of the wave being generated. If there is no path for the fluid to follow once a valve closes, the fluid will compress and recoil - which you will interpret to be water hammer.

Modeling it can be done with basic fluid flow mechanics. The underlying mechanism is simple. That said, solving the problem mathematically will not help you solve the problem in the field. Solving the problem in the field is designing a damped path for the fluid to go when the valve closes. Or close the valve very slowly if you can.