Dear Peter Thanks for your immediate answer. Thanks for your clarification about shell elements!
Well, I have used the real dimensions which I obtained from the laboratory test data, but the variation can , occur as you mentioned. The result of the flange-induced web buckling, in reality, is the depth to thickness ratio of the web which is 1270(mm)/3.27(mm). So, I expect that failure when modeling, too. In that case, I considered the web thickness to height ratio to be my "help" for getting the buckling failure, since that is the main reason. Do you think that is not enough, still?