I believe this calls for an application of the GRT (untested):

Actually, what you have there is a ST (http://en.wikipedia.org/wiki/Schwartzian_transform). A GRT (http://www.sysarch.com/Perl/sort_paper.html) is characterised by having a simple lexical sort rather than requiring a routine. Like this:

$ perl -le '
> @arr = qw{ TEXT1 TEXT3 TEXT11 TEXT2 TEXT13 };
> print for
>    map  { substr $_, 9 }
>    sort
>    map  { sprintf q{%09d%s}, ( split m{(?<=[A-Z])(?=\d)} )[ 1 ], $_ 
+}
> @arr;'
TEXT1
TEXT2
TEXT3
TEXT11
TEXT13
$
[download]

I hope this is of interest.

Update: Added references.