As an alternative approach:

Break problem into per/consolidation station sized chunks..

• Read rows into an hash where the key is the customer number and the values are the rest of record values(address, city, etc...)
• Use a regex to determine if the first field is "CUSTOMER CONSOLIDATION STATION" and if so, assign the previous hashes to a hash with a key of conslidation station number.

  You now have a hash of hashes where the outer hash has station numbers and the inner hash has all the individual customers as keys with the rest of their values

• From there is it a matter of outputting to csv from the hash-of-hashes, which you can find examples of in perldsc.

Your program gets split into two parts. Open and read the file(s) followed by outputting the h-o-h to csv.

Assuming your records resemble your output, you will be able to read customer records until you find a consolidation station record at which point the customer records are assigned to the consolidation stations record.

psudo-code:

until m/consolidation/ {
 $cust_hash->{'city'} = "city";
 $cust_hash->{'cust_num'} = "cust_num";
  ....etc...
}
$outer-hash->{$consolidation_station} = $cust_hash;
#end read loop
#output to csv is clearly within your capabilities...
[download]

Hope that is helpful...