To:Al Smith <ajs@Cs.Nott.AC.UK>
cc:d.w.corne@reading.ac.uk, emma@dai.ed.ac.uk,mcs@essex.ac.uk,wbl
Subject:South Wales Scheduling Data
--------
Dear Al,
	To confirm our telephone conversation.

1) The line flow capacities given in sw.cost are for the
   non-contingency version of the south wales problem and so are 50% of
   the real maximum post-fault contingency ratings. The capacities for
   non-contingency maintenance planning are 85% of maximum post-fault
   rating. I.e. the data should be multiplied by .85 before being
   used on the non-contingency problem. 

   On the contingency version of the should Wales problem the line
   capacity data should be multiplied by 2.0 first. In the first part
   of the calculation they should then be further multiplied by 0.85
   before use.

2) fitness.c rev 1.7 was used to create the schedule of cost 616 given
   in Figure C.7 page 263 of "Genetic Programming and Data
   Structures". This version, in any one week either provides the
   isolated_node (or network split) penalty or the line overload
   penalty. The description given by equation C.1 on page 247 refers
   to a later improved version which performs a loadflow calculation
   on the largest interconnected part of the network. 

3) The schedule given in Figure C.12 page 263 of "Genetic Programming
   and Data Structures" has a cost of 12025

I hope this helps.

Bill