EDIT: Nevermind. There is no inconsistency below. One of the two answers I got is wrong because I assumed that the voltage drop across the current source was zero. As klouchis has pointed out, this is incorrect. One of my calculations (the second one I did) agrees with his suggested method for finding Voc. So, ignore this first paragraph below:
I think that the given values might be mutually inconsistent (i.e. the current demanded by the current source is not the same as the current demanded by the other branch given the resistors voltage sources present). Here are my arguments:
I think the easiest way to find the open circuit voltage is to take note of two things:
1. The voltage at both ends of R3 is the same, because there is no current across R3. Therefore, if you consider the "node" that is at the other end of R3 (the end opposite to point a), you will see that this "node" must also be at voltage Voc.
2. The voltage at this node can easily be determined by noticing that the branch with R2 must have current I going through it. The current source demands it. It follows from Ohm's law for that branch that:
Voc = IR2 = (2 A)(4 Ω) = 8 V (EDIT: WRONG because the current source has some non-zero voltage drop across it! See method below instead!)
YET, if you look at what the current must be in the other branch (the one with the two voltage sources), you get an inconsistency. KCL would seem to demand that the current in this branch is also I. Ohm's law says that the voltage across R1 is given by the current times the resistance. The voltage across R1 can be calculated as follows. Call the "downstream" side of R1 point 'c.' Then:
Vc + V2 = Voc
OR
Vc = Voc - V2
Ohm's law says:
I = (V1 - Vc)/(R1)
Substituting the expression for Vc:
I = (V1 - Voc + V2)/(R1)
(2 A)(3 Ω) = 15 V - Voc + 4 V
Voc = 15 V + 4 V - 6 V = 13 V.
This is clearly inconsistent with the above answer. (EDIT: That's because the above answer was WRONG).