Thanks Aziz

Did you select an appropriate cross sectional area at reference ? I think you might be doing mistake to consider cross sectional area. Please check it.

Is your computational domain large enough, and did you apply appropriate boundary conditions?

Try to enlarge your computational domain in the direction of lift force.

Remember that Cd= Inline-Force/(1/2*denisty*velocity^2*ProjectedArea)...

where Projected Area= Depth * Diameter

so in Fluent, before starting the solution, Go to Reference values and check the value of density, velocity and Area and length...

In case of 2D,  

Depth is assumed unity= 1m, so ProjArea= Diameter*1=Diameter

Length= Diameter

after writing the correct values, now solve the problem

In case of 3D

Only thing which will be change is 

Projected Area=Depth *Diameter

Length=Diameter

and then solve

Make the computational domain larger enough (say 10D ) in both directions of the lift. Then use a slip boundary condition on the wall. I hope it helps.

As Aziz told, there should be no lift for a symmetric body. However, since you are doing unsteady simulations, you can see lift varying in time, due to vortex shedding, but the average value should be zero.