I was describing my job to a new acquaintance, and he asked me, “When you start a new project, what’s the first thing you do?” Indeed, Monthly Mechanics has explored every part of the design process, but rarely discussed the order. So, here is a flowchart.

**Step 1: Make a model.** What are the components of your structure? Components of a building include the roof, the walls, the floors, and the foundation. Components of a bridge might include the deck, the stringers, the piers, and again the foundation. A rough sketch of the structure helps to identify what parts you need to design, and (importantly!) enables you to define the scope of work with your client.

There are archetypes for the most common components. A beam, such as a deck joist or roof rafter, is basically a line. A column is also a line, but it’s loaded at the top rather than on the side. A floor might be a rectangle. A column foundation is a point. It takes some practice to see the shapes – your best bet is to think about where the loads are coming from and how the loads push, pull, twist, or bend each component. You’ll notice what paths the loads take through your structure while building your model.

This is also the time to identify the failure modes you’ll need to check. Are the beams loaded along one side (like a joist supporting a floor, with loads only from above) or along two sides (like a bridge girder supporting a deck weight and a simultaneous wind load)? Are they continuous over a central support? (If so, you’ll need to check negative bending over the support in addition to positive bending at midspan.) Do they experience any tension or compression, or only flexure?

**Step 2: Determine the loads and distribute them.** Right away you’ll notice that different components experience the loads in different ways. For example, one- and two-family housing is designed with a live load of 40 psf. That weight is a uniform load on the plywood subfloor, and it’s distributed down to the floor joists according to their tributary area. Following the load path, a wall or column receives that same live load as a reaction from the joist ends – a concentrated load.

Once you’ve identified all the loads, apply load combinations. Check building codes to determine how your jurisdiction adds loads together, and calculate all possibilities. It’s prudent to determine the governing load combination for vertical loads (dead, live, snow) as well as for horizontal loads (wind, seismic). Some load combinations emphasize live load; others give extra weight to atmospherics like snow and wind; still others minimize downward forces to test for uplift.

Sometimes while you’re determining the loads, you’ll realize that your model is too simplified. In that case, return to Step 1.

**Step 3: Figure out how to support the loads.** If you’re designing a floor joist, you might start by looking at dimensional lumber. Choose a size and species – maybe a southern pine 2×10 – and calculate the stress for all the failure modes you’re checking – perhaps flexure, shear, and bearing at the support. (This is called the stress required.) Compare it with the beam’s strength – found in material-specific manuals such as the National Design Specification for Wood Construction. (The strength is also called the stress provided.) If the stress required exceeds the stress provided, then you’ll need to choose a stronger beam – maybe a 2×12, or an LVL, or a steel beam. If the stress provided exceeds the stress required by a long shot, then you can economize by choosing a smaller beam.

Design is an iterative process. Repeat Step 3 until you close in on the Goldilocks beam – the shallowest, lightest, or least expensive member whose strength still exceeds the stress required. Then proceed down the load path.

Anyone can build a bridge that stands, but it takes an engineer to build a bridge that barely stands.

-Colin Chapman (maybe)

I am sure you are checking everything twice for accuracy but don’t mention it. My experience is that there are a couple of things that prevent errors in the final solution. One is gut level check of magnitudes. Two it is nice to have a second set of eyes reviewing the solution.

Meanwhile, I appreciate your posts as an active professional engineer.

Thanks for your insight, Gerald. I agree that it’s indispensable to have another person check your work – in professional engineering as in many other fields. I’ll file away that idea for a future post!