Unread post Sat Dec 10, 2016 9:28 pm

N Four-bar issues

Has anyone been having trouble with getting the N Four-bar mechanism to run correctly using the specified 400 time points? When I use 400 time points I can't get the solver (using QN, MN, or full NR) to converge with low enough velocity level constraint violations to predict the correct time history. Each revolution some of the velocity leaks out through the violations and eventually the mechanism doesn't make it back to the top. If I decrease the allowable violation it just never converges with the given step size. Decreasing the step size to 1e-2 works perfectly, but doesn't match the problem specification. I've tried using BDF 2-6 and tried using a full NR method but it still won't converge for 400 time points. I was wondering if anyone else was having a similar problem or if anyone has any other ideas for increasing the ability of the solver to converge. Plots for the time history with h = 0.01 and the specified h = 0.05 are below.
Quasi-Newton BDF 6 h=1e-2.png
Quasi-Newton BDF 6 h=1e-2.png (22.13 KiB) Viewed 389 times
Newton Raphson BDF 4 h=5e-2.png
BDF 4 seemed to work for the longest amount of time
Newton Raphson BDF 4 h=5e-2.png (22.32 KiB) Viewed 389 times