Difference in experimental result and the analytical result from Abaqus

[Jekei]

Hi,
I am modelling a bolted connection in double shear. Force is applied, on central plate, as a displacement using a BC at a reference point. And history output is also requested at this reference point for both the displacement and force(for the plot).When I compare my analytical result(force Vs displacement) from Abaqus with experimental result ,the plot from my analytical result is quite below the experimental plot. My material property data is correct and also I tried many time changing the meshing of the model just as to get the desired result but still the result is below the experimental value. I would appreciate if some one could suggest what might be the reason behind the difference between these plots.
( I am specially concerned about the differences I get towards the yielding point and after that)
The plot attached:

 

[FEA way]

What are the boundary conditions ? Can you attach some pictures showing this model with visible mesh ? Which material model do you use ?

 

[Jekei]

I fixed my upper and lower place on the left using a reference point and applied displacement load on the right of central plate using another reference point. I am using A325 bolt and A572-50 plates.

 

[SWComposites]

What material model and failure criteria are you using?
What are the materials?
Where did you get the material properties? How do you know they are accurate?
Were the bolts torqued or loose in the test? Have you modelled the bolt preload?

 

[FEA way]

Also, what are the contact settings in this case ? And can you attach a pocture showing stress plot on the deformed model ?

 

[Jekei]

@SWComposites (Aerospace)
I am using Non-Linear material model. And I am not concern about failure mode right now because I just need the behavior of the connection before the failure actually occur. According to my understanding its is not necessary to define the failure criteria for that .Right?
I am using A325 bolt and A572-50 plates.
All the material properties are taken from the lab experiment conducted here in my university and it is accurate.
I have preloaded the bolt.
I apply the load to a reference point and couple the point to the end cross-sectional surface. Then request the force displacement in output history. Does it capture the behavior well? Is there any better way to get the Force VS displacement plot?

 

[Jekei]

@FEA way
I used find contact pair option to get the pairs and make sure the master and slave fulfil its requirements. Contact property is same for all surface with coefficient of friction 0.3.