It seems that SeismoArtif constantly produces acceleration histories which doesn't fit into +-10 % allowable range when compared to target spectrum, but still states that convergence has been archived. In practice this is probably not an issue, but for example ASCE 4-98 states that any point of the spectrum should not lie below 10 % of the target spectrum. That makes the acceleration histories not usable in the analysis where strict following of the codes is required. Of course this could be managed by increasing the acceleration using spectrum factor. Another problem is that we have been using quite accurate acceleration histories for long time and now down grading is probably not an option. I'm not an expert to say what causes the acceleration histories to be inaccurate, but the software limit for 3 iterations seems a bit small for me (our old software did something like 5 to 20 iterations, described in the: http://www.kolumbus.fi/pentti.varpasuo/ ... ra_v02.htm).
It would be highly appreciated if you could slightly improve the accuracy of the acceleration histories in the next version. Thank you!