Lets assume you are backed up.

remove all drives, but the failing and new disk.

restart

Identify their disk#s (diskutil list, or using SoftRAID disk tiles.)

You have to be very very careful here:

lets say the old disk is disk2 and the new is disk3

You can use the dd command in terminal to clone the old to new, using a command like this:
sudo dd bs=1m conv=sync,noerror if=/dev/rdisk2 of=/dev/rdisk3

sudo gives admin capability
dd is the command
bs= specifies the block size to be transferred, 1mb will go faster than the default 512 byte transfer
conv= means if an unreadable block is encountered, it will fill the remaining data with zeros, so the two disks remain in sync. with 1MB block sizes, this means a minimum of 1MB would be corrupted on a bad source, with each bad section. You can use a smaller size, but it takes much longer, and it does not seem your disk is that bad off.
if= is the path to the input (old) disk
of= is the path to the output (new) disk

If you do it backwards, all data will be lost, so backup first.

there is no feedback, but you can guesstimate the time by size of the disk divided by the throughput (on Thunderbolt should be 200-225MB/s)

When you get a command prompt, Immediately remove the old disk and restart. Both disks will have the same SoftRAID Identifier, which can confuse the IO registry. this is very important!

put the rest of the disks in and it should be good.