Question about cylindrical Arc-Length method by M.A. Crisfield?

[Simulate]

I want to know how to compute the load increment in the first
analysis
step of the cylindrical Arc-Length method ?

I firstly use Newton-Raphson method then swich to the cylindrical
Arc-
Length method if has convergence problem. Usually(also in M.A.
Crisfield's paper and book), the arc-length in the the first analysis
step can be compute by deltaL=deltaLamda1*sqrt[(dis1,dis1)], where
dis1 is the displacement increment obtained by one iteration of
Newton-
Raphson method and deltaLamda1 is the guessed load increment factor
in
the first analysis step. However, the guessed load increment factor
is very large which lead to unconvergence in my program.

The rational load increment factor in the first analysis step should
be control by the Arc-Length radius, however, Crisfield's method give
the load increment factor in the first analysis step exceed the Arc-
Length radius.

Why?

Am I miss something????

Please help me.