Usually with this kind of mining, the water is drawn locally, usually from the same formation that is being mined. It is then pumped down a hole into the sand layer, where it picks up sand along the way (in a slurry – imagine a water+sand slurpee – the most Winnipeg analogy ever) and comes up the same hole through a second pipe, or an adjacent hole. Of course, when the water stops moving for a minute or an hour, all the sand sinks to the bottom, and the water can be returned to the hole it came from.

The principal concern of opponents is that the water will become contaminated somehow in this process, and will ruin their drinking water miles (sometimes hundreds of miles) away. This is so completely overblown. Unless this company is doing something weird (which I doubt), they’re using the local water and putting it back, after simply using it to move the sand. Sand itself tends to be near perfectly inert (as chemically inert as glass, essentially). Furthermore, this sand has such high purity that there are no expected contaminated riders that should be disturbed in the process. Clean water mixed with clean sand, and returned as clean water.

This isn’t anything like a gold mine where there’s naturally occuring arsenic, or processing related cyanide, or similar type issues. Or people are conflating it with fracking somehow. The issue here is purely “mining bad, mkay”.

But people forget: everything we use in our day-to-day lives either had to come from a mine or a field (except ammonia, which is made from the atmosphere).

So to answer your questions specifically:
(1) the water is usually returned to the same reservoir it came from.
(2) there is likely to be small impacts to groundwater levels – some water will evaporate from ponds on the surface while the sand is settling, for example, and recovery will be close to, but never 100%. It is likely to be measurable, but not noticeable.
(3) The voids will likely fill with water. It is possible there is some minor subsidence where the ground above slowly settles. This is also the case for any aquifer where farmers are drawing irrigation water. These millimetre scale changes are likely measureable over time with very precise instruments, like satellite based dInSAR (differential interferometric synthetic aperture radar – I actually studied this in grad school, so feel free to add questions). The effect on people will be negligible.

Never dumb questions. :)