CFD Online Discussion Forums - need favor in icoFsiElastciNonLinSolidFoam

Dear all
With respect.
I using icoFsiElastciNonLinSolidFoam to simulate Hron Turek benchmark. the default tutorial is not consider. I use other test with fluid density 1000, v=1e-3 and solid density 10000, E=5e5, v=0.4.
My problem is that after 4.48 second, model has error, after changing fvscheme and delete some results from [4 4.48], then run again in time 4s, the solver need grad(Du_0) and grad(Usolid). I choice Guass linear for them. the simulation will start from last time (4s), but pointhistory give different number.
For example I uses

HTML Code:

  pointHistory

   {

       type pointHistory;

       functionObjectLibs ("libpointHistory.so");



       region solid;



       refHistoryPoint (0.6 0.2 0.025334);

   }

In first simulaton I have (time:4second -0.290069214667 -0.0081072673056 -3.07099999979e-08) for x,y,z
In second time=4second ( -0.30060291207 0.0094709455648 0)
sometimes, I need to stop running and start again, but the problem is same.
my question is: why results different? why solver need Du_0 and Usolid? how run solver again and continue from previous time?
Thanks