Hi

What your experiment is, is not entirely clear from your description but, when using tools based on first principles, like FE, inherent assumptions are that you do describe geometry and material properties as closely as possible to your experiment, to get as good a match as possible. In short, you want to get the right answer for the right reason.

If you by self weight imply preload from the static g-field and you notice a substantial difference, you should then be able to examine this in your experiment by reorienting it with respect to the g-field as loads or initial deformation then matter.

If you by self weight to load imply the weight of transducers and other equipment that is part of the measurement equipment, these may matter. You can then either incorporate them into your model or design your tests such that such weights have negligible influence. The latter is necessary when you have to move transducers to get test data that you wich to collate, e.g. in experimental modal analysis.

More rambling on this topic can be found here if you are interested: http://qringtech.com/learnmore/why-simulate-measure-correlate-automate/

Sincerely

Claes

My experiment is four-point felxural loading.

I measured the deflection due to two points load.

 I meant by the self-weight is the self-weight of specimen itself 

Hi

A static test?

Simply supported is easy in theory and not so easy to mimic in test as it involves mechanisms and offset away from the neutral bending center.

Influence from self weight would then imply a bending weak test object. Is this the case?

I mean, have you checked your actual geometry before loading when located as in test and when rotated such that the length direction is parallel with the g-field? For self weight truly to matter, there should then be a measurable change in the geometry.

/C

Yes, I suggest you to take into account the self-weight also.

In Midas FEA, I found type of load called Body Force. Does it represent the self-weight only or self-weight with other parameters? 

Because when I removed it there was a big difference in deflection results in comparison to experimental and calculation results. 

Unless it is a failure experiment, you need not consider the self weight. This is because, all your strains and deflection u will measure from the datum, after you reached the equilibrium due to self weights. That means, the deflection measurements u made are excluding deflections (strains) due to self weight