- How many PLY do you recommend to let the engine run on a single move
This will depend on many factors such as the actual position, how you use the engine for opening analysis, etc
If the engine gives a decent advantage (let's say +-2 or so), there is usually not much point in having it calculate down to mate. Other than that opening analysis with an engine is tricky and should certainly not be automated. Engines are useful for pointing out tactical mistakes in openings and can be useful in evaluating midgame/endgame positions that result from some opening lines.
However at least currently you cannot expect engines to give you the absolute best line from say move 5 in any opening. In many openings engines give an evaluation of +0.6 or so even though experience shows that positions are perfectly playable for both sides.
Also human interaction is needed to evaluate how playable the position is for a human: e.g. you might have an equal position where you are required to find 10 only moves or so; this is not what you want to get into.
- How long does it take you time-wise to reach this?
Depends on position, computing power and perhaps other factors.