If the surge force data given is the (Force-Time) history from a transient simulation software, you might want to employ time history analysis to consider the dynamic effect in your system. Another approach that I have seen is to convert the time history to force spectrum (from DLF generator of Caesar II). But normally, time history will give a more reliable results as long as the timing of the loads is properly considered in the generation of force-time profile.

Normally the long straight pipe length between elbows have the large surge force. I normally put limit stop in the middle of the run and submit the surge force to Structural group.

Regarding the direction of force, just check the source of the surge (i.e, fast closing valve, pump trip). Normally, the high surge pressure will start from the source of surge, due to the kinetic energy (sudden decrease in velocity to zero) being converted to pressure energy (while the opposite elbow will be at normal operating pressure). You can consider that as your initial direction for the surge force (taken from differential pressure between two points). This pressure wave will echo back & forth in your piping system until such time it losses energy due to dampening effect of friction. Time history analysis will capture the dynamic effects at every time step.

Any other opinion is greatly appreciated. Just correct any wrong statement I've made.

Cheers!!