Roof Thrust

The ancient Roman bridge and aqueduct builders understood it. So did the medieval cathedral builders. But many home builders--and homeowners who alter or add on to their houses and garages--still need to learn about thrust.

Thrust is the outward push that an arch or gable roof produces as a result of the downward loads it carries. Basically, the downward load tries to flatten the arch or gable, causing the ends to push outward.

There are a handful of ways to prevent collapse, for example:

• Give the structure a very high bending strength. This is usually expensive and a waste of material.

• Give the structure a strong, stiff foundation that can handle the thrust without much movement. This is the solution commonly used by the Romans, who were often lucky enough to have solid rock on which to found their arches.

• Make the structure tall and pointy, so that most of the thrust is downward rather than outward. This--along with flying buttresses to share the thrust--was the cathedral builders' solution. It's great if you want a really tall structure, but not if you want to keep a low profile.

• Tie the two ends of the structure together and allow the two outward-thrusting ends to work against each other. This is a good solution if the tie doesn't get in the way. In a tied-arch bridge, the structure of the roadway is used as the tie. And in a gable roof, ceiling joists tie the rafter ends together and prevent the outward thrust from pushing on the walls. At least, that's the intention...

The three case histories summarized below demonstrate what can happen when the effects of thrust are ignored. In each case, Unified was asked to determine the cause of the distress.

Spreading out

Our first example is a two-car garage, either built or designed by the homeowner. It is, as you can see, a concrete block structure with a low-slope gable roof. According to the homeowner, a delivery truck struck the garage shortly after it was built, causing cracks at the point of impact and at the tops of the walls. Unified was asked to check into this claim. Well, there was some impact damage (with which we won't concern ourselves here) and there were cracks at the tops of the walls, but the two weren't in any way related.

The cracks at the tops of the walls were due to thrust. There were no joists tying the rafter ends together, so the block walls on the sides of the garage had to carry the thrust of the roof. Since the walls weren't designed for the task, they began to spread out at the top, cracking the end walls and taking them along for the ride.

Unified also found a pretty strong indication that the homeowner knew the real cause of the cracking. A prop and a collar tie--clearly made from some old wood he found lying around, and just as clearly not part of the original design--had been installed at one end of the garage in a weak and unsuccessful attempt to stop the outward spread.

Spreading out II

In our next example, a house with a gable roof is losing bricks from under its eaves. You can also see a distinctly sagged ridge line. Inside the house, many of the rooms have cracks in the corners between the walls and the ceiling.

In this case, there are ceiling joists connecting the two sides of the house, but they're not well positioned to carry the thrust. In common construction, the rafters, wall studs, and ceiling joists meet at a common point where the load is transferred--the vertical load going into the studs and the horizontal thrust into the joists. In this house, however, the wall studs rise about 18 inches above the ceiling joists, and both types of load have to be carried by the upper portion of the studs.

The studs, unable to handle the thrust, bent outward and cracked. As a result, the ridge line sagged and the bricks popped off the upper portion of the wall.

Despite the homeowner's contention that this was a new problem (caused by a recent hailstorm), it was clear that it had been known about for some time. A series of collar ties had been installed near the middle of the building, but they were too close to the ridge to be very effective at carrying the thrust. Notice, too, in the photo below, that the separation of the rafters at the ridge line is very much like the drawing shown near the top of this page.

Inside out

Now we come to an example where the spreading of the walls has gotten way out of hand. The garage shown below looks normal enough--but where's the roof? You'd have seen the roof had you driven past the garage a week before the photo was taken. For over twenty years, the garage had been covered with a standard rectangular hip roof. But then one nice June day--no snow, no high winds--the roof decided it had had enough and collapsed into the garage, inverting itself. The ridge line at the peak of the roof ended up near the floor, and the hips turned into valleys. The second photo was taken from the center of the garage floor, the photographer's head stuck up through a hole created by the collapse, looking up from the ridge line to the eave.

What happened? As in the previous examples, this roof had no ties connecting the common rafters in the middle of the building. The outward thrust produced by the roof caused the walls to separate from each other at the corners. Now unsupported, the roof popped through the opening like the bottom of an old-fashioned oil can. After the hips bent back on themselves, the walls were pulled back to a nearly vertical position.

Interestingly, a square hip roof, in which all the hips meet at a common point at the peak of the roof, doesn't need ties because the sheathing (the boards, plywood, or OSB that cover the rafters) connects the hips and keeps them from spreading. But a rectangular hip roof has a middle section that acts very much like a gable, and it needs ties to keep the thrust from pushing out the walls.

Apparently, the builders of this roof thought it was close enough to square to get away without providing ties. And for more than two decades, they were right. But over the years, the tops of the walls slowly inched out until eventually the roof snapped though.

These investigations were performed by Mark Lawrence.