Revit families

Why I'm building my own Revit fitting content

Making a two-45 offset in Revit takes a small sequence of steps in my current workflow. Model the two fittings with a piece of pipe between them, delete that pipe, then align the offset run. After repeating it, I started wondering whether that behaviour could be built into the family itself.

The idea is to place the assembly through routing preferences while still counting the individual fittings correctly. There's no giant saving in one offset. But these small, repeated tasks are worth a closer look, especially if the content can make them simpler.

Start with a lighter family

Lightweight family architecture matters just as much as placement. For a model with thousands of fittings, I want to start with only the geometry, parameters and behaviour the work actually needs.

Some content I've used has been difficult to modify with confidence. Changing one part means understanding what else depends on it. Building our own families is an opportunity to make those relationships clear from the beginning.

A bare-bones family still has a job to do. It needs to connect properly, respond to its intended inputs and communicate what the fitting is. Anything removed has to leave those functions intact.

I'm looking at both file size and the complexity inside each family. The hope is that lighter content will help the model run more smoothly. That needs testing in the model; a smaller family file alone doesn't establish a performance improvement.

Make the information consistent

The same deliberate approach should apply to the information each fitting carries. Descriptions, naming, weight, cost, labour and counts need consistent standards that fit the way we schedule and spool the model.

Defining those standards early gives us a basis for the rest of the library. It also means considering fabrication information while developing the content, rather than reconciling it afterward.

That's where nesting gets interesting. Along with the two-45 offset, I'd like to place a tapping, nipple and gauge together, with separate counts for each component. One placement could handle the assembly, provided the schedules still report it correctly.

Both ideas need placement and scheduling checks before we can rely on them. A convenient assembly isn't much help if it makes the quantities harder to trust.

Find a method worth repeating

I've seen some placement work become quicker, but haven't measured the overall effect across a project. For now, the work is to identify useful changes and make them repeatable.

AI and automation may help carry a consistent family-building method into other fitting systems. First, we need standards and a way to test the results so that we're repeating something we understand.

That's what gets me interested in building our own content. If we can do this with one assembly, where else can we make the work simpler?