An html archive consists out of a set of html pages with associated media (read: images and simple javascript).
A CHM has a indication which page is the "main" page, which is usually some overview page. Besides that, it has a Table of Contents (TOC) which is a tree of nodes that point to html files. Walking the tree would give a more or less linear order.
But the default page might not be the first page of the TOC, or in the TOC at all, and not all pages might be in the TOC. In that case there is no order that can be detected by automated means.
Extracting a CHM with a general decompilation tool will yield you a bunch of htmls, a .hhk and a .hhc. The .hhc is the TOC in XML form. The hhk is the index, but you don't need it now. The default page is in an internal file, and generally not visible after extraction (use properties of chm tools)
Besides the determining of the order, there is the actual merging itself. This can be hard, but practical workarounds might be importing them into office by some scripted means.
I think an able scripter might pull it off, but it is not trivial.