One structure, many books: per train, per pump, per skid

The final book conversation on a project has two dimensions, and both are worth settling early.

The first is what kinds of book. Customers ask for the record by different names with different contents: an MRB, a quality book, a dispatch book, an inspection book, sometimes several of these on one order. These are separate deliverables, not versions of each other; each has its own contents list, and a document that belongs in the inspection book may have no place in the dispatch book.

The second is how the books are cut. One book for the order? One per train? One per pump, per skid, per area of the plant? The equipment supplier shipping a packaged unit often wants the cut their own way regardless: the unit runs as a single piece of equipment, and the book is its record of manufacture, built from the component documents.

Handled by hand, both dimensions are expensive for the same reason: without document control, every document lives in its own folder, so most of the effort is gathering and collating before anything is a book. Finding current revisions, ordering to the specification, building bookmarks and hyperlinks, getting every tag and number correct, then producing the contents to match. And the truly painful event is not a late revised drawing, replacing a file is trivial. It is the structural change: the customer wants the book laid out differently, or one per skid instead of one per order, and the gathering starts again.

The fix is to separate the two things being conflated: the document set, and the structures cut from it. The documents are tracked once, each linked to its equipment. Each book is then a definition, not a pile: this book type contains these document codes, cut at this grouping. The MRB per train and the dispatch book per order are two definitions over the same set. Contents by document code is the default and the easy way; where a section needs documents placed individually, that is a definition too.

Defined that way, everything hard becomes a setting. In DocBoss the split is a grouping on one structure, order, tag, major tag, plant, area, skid, building or a custom grouping, with major tags giving two tiers of equipment, so the machine and its components each sit at the right one. The table of contents can be generated in full at kickoff, real documents, numbers and tags, for the customer to approve the layout before a single file is produced. Books assemble from the latest approved documents, bookmarked and hyperlinked, with the table of contents to the layout, and each regeneration reads from the tracked set, so the book you produce is built from what is current. When the structure changes late, you change the setting and regenerate; nothing is re-gathered.

One tracked set, many definitions, any cut. That is the whole design, and it is the difference between the book being a deliverable and the book being a project of its own.