As pointed out in the comments, the question has been asked before in relation to servers, where there is a feeling that powering off a computer at night shortens longevity. But the answers in that question are unsupported and use car analogies, taking about engine wear. In fact, one of the comments on the top answer points out that Google's big investigation into hard drive life gave inconclusive evidence that power cycles had more effect than luck-of-the-draw.
So let's instead turn the question around slightly by asking "what can we do to increase a computer's physical lifespan?", which will ignore the software side of things (malware, degraded performance etc)
Cleaning
The argument goes that if your computer is caked in dust it will significantly reduce the efficiency of heat conduction. A search reveals plenty of places advising this (eg these guys, to pick a random example); but no studies or reports that I saw in the first few pages of results. I guess this is one of these things that are treated as self-evident: overheating is extremely detrimental to lifespan; dirt and dust leads eventually to overheating.
Power Surges
Power surges are not kind to components. Searching reveals common advice is to buy a surge protector; or for even better protection, a UPS (uninterruptible power supply).
If you have no or inadequate protection in place when a power surge occurs, it'll affect the power supply first, and if you're lucky, nothing else. If you're not, the surge can affect anything- motherboard, CPU, graphics card, hard drive, etc. There's an Anandtech forums discussion where someone explains in better detail:
The real severe internal damage comes when the input power causes the power supply to push too much electricity into internal parts [...] Some really delicate components -- mainly processors and memory -- are actually damaged by this, and get bricked. Simple parts can be killed too: You take a resistor and dump too much juice into it, it heats up and burns. You dump too much power into an electrolytic capacitor, and it can burst and spew electrolyte fluid all over the place.
There is a common argument too that poor quality power supplies can decrease the longevity of other components. Certainly a faulty power supply can cause problems - plenty of anecdotes, like pins melting to a motherboard - but even those advocating good quality supplies (another example I picked) tend to rely on citing experience, or are power supply manufacturers themselves.
I've seen the argument that switching on a computer induces a degree of power surge, but I can't find much to back this up.
So is turning an office computer off at night or not likely to be harmful to hardware life?
In the time it's taken me to write this answer, I cannot find good evidence to suggest that turning a computer off once a day would significantly shorten its life. However, the evidence is not good enough to make a recommendation one way or another on the basis of longevity alone. You may want to consider other factors - such as energy savings - instead.
You might find this question on Server Fault useful.
– Deltik – 2015-06-26T12:28:06.353@Deltik thanks, it is really interesting, but here I'm asking about an office PC therefore a different hardware and a 8/9hr per day activity so, if I switch off the PC at evening it is used for only 30% of the day. – Tobia – 2015-06-26T12:31:09.873
Nobody knows.. but note that if you successfully increase the longevity of it, you can end up with a computer that is well below average in performance and hardware generally e.g. because even if it was at the same speed as when you bought it or set it up, technology in the world is increasing in speed and capacity and changing in form, as the years go by. – barlop – 2015-06-26T13:48:29.003
1@barlop this is not as true now as it used to be, hardware from 5 or even 10 years ago will do things like run Office and web browsing just fine, if it has enough RAM and an SSD. – LawrenceC – 2019-05-16T13:27:37.917