A partially prestressed member will be cracked, almost by definition.
For calculating stresses, yes an elastic cracked analysis is OK, assuming the concrete compression is within code limits for elastic analysis, and the prestress steel is well below yield.
MotorCity - you don't do an elastic analysis ignoring concrete in tension then?
The initial question is too general for a simple answer. Yes, elastic stresses can be used to decide if there is sufficient tension to crack the section. Then you have to use cracked section analysis.
Most codes nominate the same tensile stress at transfer as at service below which the section is assumed to be uncracked. I think this is incorrect. The rules on this for bonded prestress are based on having bonded tendons at the tension face. For unbonded prestress, there should always be bonded reinforcement at the tension face. This does not normally happen at transfer so a lower limit should be adopted. I think this limit should be zero, so as soon as there is tension, a cracked section analysis should be done. Eurocode has a method of calculating a crack width for this situation. When using it for one case, I found that the crack width became uncontrollable at a hypothetical tensile stress of just over 1MPa, much lower than most design code limits to decide on cracked/uncracked.
