Hi!
While Abaqus does have some specific fracture-related features for grey iron, it may be enough to just input the stress-strain curves point-by-point; authors such as Angus and Roark regard the material as non-linear past about 25% UTS and some tests I've carried out confirm this (others with different strain-gauges are ongoing). As with many "awkward" materials, I guess it depends what data you've got and whether you're doing cyclic loading or just monotonic. There's some good stuff on this in a PhD thesis by Downing (all authors should be pretty easy to Google!).

It may be worth just doing a couple of load points elastically first to see which bits are in tension/compression (ie how much it matters!) and considering how conservative you want/have to be (eg is treating it all as elastic OK if you assume a low elastic modulus ie very stiff?)...

Good luck!

Thanks for replying MToft. After thinking about my problem for a while I´ve found that I don´t need to model this behaviour at the moment. As usual if you let a few days pass you´ll cool down and go back to reality....
My experience based on my companys data from grey Iron (tons and tons of data) is that it isn´t linear at all. 25% of uts as linear is probably a good approximation though.
Good to hear opinons from someone else so thanks again for your answer.