Difference in solution between FLAC3D9 and 7

I’m doing a simple test on FLAC3D version 9 and 7. It consist of a slice (2D plane strain) of elastic cylindrical zones (tunnel lining) under gravity. V9 takes a lot of time to solve (for only 180 zones) and the mechanical ratios (local and average) reduce very slowly, which is different with V7. The data file and the plots are below. What is the reason for this ?

model new
model large-strain off

[rad_int = 4.92]
[rad_ext = 5.34]
[width_ = 2.0]
[zn_rv = 10]
[zn_or = 45]
[zs_or = math.pi*rad_ext / (2.*zn_or)

model gravity 0. 0. -9.81
zone create cylindrical-shell point 0 (0,0,0) point 1 (@rad_ext,0,0) …
point 2 (0,@width_,0) point 3 (0,0,@rad_ext) …
dim @rad_int ratio 1.0 size @zn_rv 1 @zn_or
group ‘mat_=voussoirs’

zone reflect or 0. 0. 0. dip 90. d-d 90.
zone reflect or 0. 0. 0. dip 0. d-d 90.

zone cmodel assign elastic
;zone property dens 2.4e-3 young 35e3 poisson 0.2
zone property dens 2.4e3 young 35e9 poisson 0.2

zone gridpoint fix v-y range p-y 0.
zone gridpoint fix v-y range p-y [width_]
zone gridpoint fix v-z range p-x 0. tol [zs_or] p-z -1e3 0.

history interval 100
model history mechanical ratio-average
model history mechanical ratio-local

model solve ratio-local 1e-3 cyc 100 and

Hi!, there is a big difference between F3D9 and F3D7. The gridpoints in F3D9 has less inertial mass than in F3D7, so the gps are more sensitives to unbalanced forces, but when you have a “really stable” problem, it takes less cycles to converge.

Thanks !
as when solved to a ratio-local of 1e-3, the max displacements from the 2 versions show about 20% of difference, what is the “right” method ?
By cycling further F3D9, it seems that the difference in max displacement with F3D7 tends to reduce. May be it will converge to the F3D7 solution, but by alot cycling. Is there anyway to make F3D9 works like F3D7 in this type of “unstable” equilibrium ?

Where can we find the complete list of difference between F3D7 and 9 ?

after model new, you can use “zone mechanical safety-factor” to increase the gps inertial mass

1 Like


Thanks, it seems to work now, but with multiple tries for the safety-factor. Any tips to determine this factor more quickly ?