Hallo Parviz,

first, Erik is quite right in what he stated in his last reply!

Now, some lines frome me:
To have real galvanic isolation is just an illusion. It would mean, that both sides of barrier can float to any potential. ANY! But that's impossible, of course. Think only about heavy electrostatical charge, which can easily be found in any industrial environment. What, if potential difference caused by electrostatical charge is higher than that what optocoupler can withstand?
No, you need always some earth referencing. For saftey purposes this is even a must in most situations.

So, you must always have some path for electrostatical discharge current to flow to earth, on both sides! For those cases that one or both sides don't have that path, you must have your barrier protected against overvoltages, e.g. by a varistor or transzorb with about 500V threshold voltage connected between both grounds of barrier, perhaps paralleled by 10nF/1kV ceramics capacitor for soft earthing of high frequency EMI.

Let's have some concrete calculation: Assume your galvanic isolation barrier consists of an opto coupler. Breakdown voltage is assumed to be about 1kV, barrier capacitance to be about 1pF.
Now, further assume that ground of the one side (A) is not connected to earth. Ground of the other side (B) shall be connected to earth. What happens now, when ground of A-side is touched by a person, which has stored lots of electrostatical charge, resulting in a electrostatic potential difference realtive to earth of let's say 10kV?
Answer: Body capacitance relative to earth is about 100pF, so, when touching A-ground, barrier capacitance of 1pF is charged while body capacitance is discharged, resulting in a final voltage rise at barrier capacitance of U = Q / C = 100pF x 10kV / (100pF + 1pF) = 9.9kV!

If your cable routing of A-side is extended over large area and does not see any earth referencing, barrier must struggle with extremely high electrostatic charge and overvoltage. Speaking in terms of electrostatical charge and ESD, introducing galvanic isolation barrier consisting of an opto coupler does not make anything safe or safer. It's just the opposite! So, galvanic isolation must always be designed very carefully. And it's never foolproof!

Kai