I just wanted to close the loop on this thread by saying that putting pitch and roll accelerations in as uniform loads seems to work quite well, while the displacement-based method in the Caesar manual does not. Thanks to DanB for his accurate advice.