I do not clearly understand what you are trying to do, but it looks as if you want to extract the values of column 8 in each file individually and find the minimum and maximum of these values per file and compute some other value or values.

In the OPed code, the array  @column8 is automatically created and values from all files are pushed to it. The array is never cleared from file to file. The problem seems to me to be with this array and not with the  $ZMAX and  $ZMIN scalar variables.

I think I would write something along the lines of the following untested code. Lexical variables are used everywhere. Warnings and strictures are enabled. A new array is created for each file.

use warnings;
use strict;

use List::Util qw/ min max sum /;

my $a = -55;
my $b = -54;
my $c = -55;
my $d = -54;

FILE:
for (1..110) {

    my $file = "x_$a\_to_$b\_y$c\_to_$d";

    open(my $in_fh, '<', $file) or die "opening '$file': $!";

    my @column8;  # FRESH, NEW, EMPTY ARRAY for each file!

    while(<$in_fh>){
        push @column8, (split(/\s+/, $_))[7];
        }

    next FILE if @column8 < 2;  # are min/max valid?
    
    my $ZMAX = max @column8;
    my $ZMIN = min @column8;
    my $thick = $ZMAX - $ZMIN;

    print "$thick\n";

    $c++;
    $d++;

    }
[download]

BTW: Another way to make an array 'forget' its previous contents is to clear the array by assigning the empty list to it:
    @column8 = ();