Truss Deflection by Virtual Work
Put an imaginary unit load where you want the deflection, and it tells you which members matter.
Before this Lesson
- Trusses: Method of Joints · Statics
- Axial Deformation · Mechanics of Materials
A sign hangs from two steel rods, one from each side, each falling 5 ft over a run of 12 ft. Under the sign's weight, each rod stretches 0.02 in. Yet the sign drops 0.052 in., more than two and a half times as far. A truss joint is that sign with more rods, each stretching or shortening a little at its own angle. This Lesson finds one number per member that says how much its stretch moves the joint, using statics alone.
Stretch turns into motion
The rods are pinned at their tops, A and B, and meet at C. When C drops a small distance , each rod swings a hair and gets longer. Only the part of the drop along the rod adds length, and the rod falls 5 for every 13 of its length, so that part is 5/13 of the drop. Call a member's change in length , as the Handbook does:
The part of the drop across the rod only swings it, which doesn't change its length. So a stretch of 0.02 in. lets C sink 0.052 in. The shallower the rod, the farther the joint must fall to use up the same stretch, which is why a clothesline pulled nearly flat sags so far.
A unit load measures leverage
Now forget the sign and hang a load of exactly 1 at C, in any units. Each rod pulls up with 5/13 of its force, so:
The shallow slope that made C move far also makes the rods work hard: each carries more than the whole load. Multiply each rod's by its stretch and add: in., the drop exactly.
Here's why. Keep the unit load at C, and let one rod out by a small with a turnbuckle while the other keeps its length. C sinks by , and the unit load with it. (C also slides a little sideways, which the hanging load doesn't feel, so it sinks only half as far as when both rods stretch.) The rod holds its tension all along, so letting it out is like lowering a bucket on a rope: your hands absorb of work. Nothing else moves against a force, so that's the work the unit load gave up:
So is how far the joint moves, along the unit load, per unit of that member's change in length. Change every member at once, and for motions this small the shares add. That's the Handbook's Truss Deflection by Unit Load Method:
The entry's signs: is positive in tension, is positive when a member gets longer, and is positive in the direction the unit load points. A member the unit load compresses is propping the joint against it, so lengthening that member shoves the joint the other way, and its share is negative.
The unit load is imaginary, a measuring device that never touches the real truss. That's why this is called virtual work, and why can come from anything: the loads, the weather, or a mistake in the shop.
Stretch from the loads
From Axial Deformation: , positive when a bar stretches, with kips, inches, and ksi kept together. Here the force is and the change in length is .
When the loads do the stretching, each member's is , with its real force:
Keep the two forces apart. is what the real loads put in a member, and is what the unit load puts there alone. Both are tension-positive, so a member whose and share a sign moves the joint along the unit load, and one whose signs differ pulls it back.
The formula won't convert units for you: with in ksi, the lengths go in inches. Set it up as a table with a column for , and divide the total by when every member shares one. A calculator's list mode makes the sum quick.
Two passes through the joints
Recall from Trusses: Method of Joints how to solve a truss: find the reactions, then go joint by joint where at most two members are unknown, drawing each pulling away from its joint, so positive is tension. On a truss built out from a wall, start at the free end.
Usually you get only the truss, so you run the method of joints twice:
- Real loads. Solve the truss as loaded for .
- Unit load. Take every real load off, put a single 1 at the joint whose motion you want, pointing the way you want it measured, and solve again for , reactions included.
Often the second pass is free. With one load , at the joint and in the direction asked about, the unit-load pass is the real one divided by , so . With a second load anywhere, run the joints again.
Where the unit load goes, and who sits out
The unit load goes at the joint whose motion you want, pointing along that motion, and it reports only the motion along its own line. If the sign's left rod alone stretched, C would sink and slide to the right, and a vertical unit load sees only the sink. To catch the slide, you hang it sideways, as the table's checkpoint did.
Some members drop out before any arithmetic, because their is zero:
- . The real loads leave the member alone, so it keeps its length, like the hanger BD in the table's truss.
- . Its stretch can't move this joint this way. Hang a load at B of that truss, and BD stretches. But with the unit load at D, joint B has no load and three members, two of them in line, so BD gets . Stretching BD lowers B, and D, resting on its rafters, doesn't care.
The zero-force rules find both kinds in seconds: run them once for the real loads and once for the unit load.
Heat and misfits
Nothing in asks why a member changed length. Warm one member, and it alone grows, by the free expansion from Thermal Strain and Indeterminate Bars, which the Handbook's entry writes as:
So the sum has one term. A member fabricated too long or too short, a member misfit in the Handbook's words, works the same way, with equal to the error, positive if it's too long. All you need is that member's . The loads, , and never enter. And in a determinate truss none of this causes stress: the truss just takes a slightly different shape.
The sign of says which way the joint goes. With a unit load down at C of the triangle truss from the second worked example, the tie AB pulls and the rafters push. A warm tie lets the rafters splay, and C sinks. A rafter that's too long props C up.
If the tie warms 50°F, with steel's per °F from the Handbook's Table 1 - Typical Material Properties:
It's positive, so C sinks 0.023 in.