Determinacy and Stability

Count first, then look: a structure can have enough supports and still collapse.

Before this Lesson

Set a plank across three skateboards, all pointing the same way. Each one pushes up, so you have three unknown forces, and statics gives you three equations. The arithmetic balances. Now nudge the plank with a finger, and it rolls across the garage. Add a fourth skateboard, and the arithmetic says you have a support to spare. It still rolls. This Lesson teaches the counting for beams, frames, and trusses, then what the counting can't see.

Three equations, and how many unknowns

Take a straight beam with no hinges, and call its number of reaction components rr. Statics offers three equations. Compare the two, and only three things can happen.

r = 2: unstable r = 3: determinate r = 4: indeterminate, degree 1
The same beam with too few, just enough, and extra reactions
  • Unstable, r<3r < 3. There are too few restraints. A beam on one pin swings like a gate.
  • Statically determinate, r=3r = 3. Unknowns and equations match, so statics alone finds every reaction.
  • Statically indeterminate, r>3r > 3. There are more unknowns than equations. The surplus, r−3r - 3, is the degree of indeterminacy, and each extra restraint is a redundant.

Don't let that last word worry you. "Indeterminate" is the analyst's complaint, not the structure's. Take the redundant away, and the beam still stands. What's missing is only your ability to finish with statics. Compare a three-legged stool with a four-legged table. Statics settles the load in each leg of the stool. On the table, slip a shim under one leg, and the load shifts among the legs with every equilibrium equation still satisfied. The split depends on how things bend, which is the business of Beams with One Redundant.

A hinge hands you an equation

An internal hinge is a pin built into a beam's length. The pieces on either side can swing relative to each other, like a door on its frame, and they can pass a force across, but not a moment. So before any analysis, you know one fact about the beam: M=0M = 0 at the hinge. That fact is an extra equation, called a condition equation, and cc counts them.

One free body gets only three equations, so where does a fourth come from? A hinged beam isn't one body. Cut it at the hinge, and you have two bodies and six equations. The cut exposes a force with two components, two new unknowns, and no moment. That leaves a net of four: 3+13 + 1.

r = 4 c = 1 A B H C A B C Ax Ay By Cy Hx Hy Hx Hy 3 equations 3 equations
A hinged beam, whole and cut at the hinge

So a beam compares its reactions with 3+c3 + c:

r<3+cunstabler=3+cdeterminater>3+cindeterminate, with degree r−(3+c)\begin{aligned} r &< 3 + c && \text{unstable} \\ r &= 3 + c && \text{determinate} \\ r &> 3 + c && \text{indeterminate, with degree } r - (3 + c) \end{aligned}

The Handbook's Stability, Determinacy, and Classification of Structures section defines rr and cc, and prints no line for beams. This one comes from its frame line, as you'll see.

Worked Example A beam with a fixed end and a hinge
1 of 3

Find the degree of indeterminacy of this beam.

A B H C D
A beam fixed at A, on rollers at B, C, and D, with a hinge at H

How many hinges make a beam determinate

Hinges can bring a continuous beam down to determinate on purpose, so that a settling pier doesn't build up moment in a long bridge girder. Each hinge lowers the degree by one, so the number you need is the degree itself:

c=r−3c = r - 3

Over five supports, a pin and four rollers, r=6r = 6, so it takes three hinges.

The count says how many, not where. To test a layout, walk the beam piece by piece, where a piece is a stretch between hinges. Every piece needs two holds, and a pin or a roller is one. A piece with two supports, or a fixed end, stands on its own. Once a piece stands, its hinge is a solid hold for the piece next door, which then needs just one more. If every piece gets its two holds, the layout works.

Stands A B C D E Folds A B C D E
The same beam with three hinges, placed two ways

In the top layout, the pieces over A and B and over C and D stand, and the others hang from them. In the bottom layout, the first piece has only A, and the next has no support at all. Each waits on the other, so they fold like a knee, while the piece over B, C, and D has a support to spare. That's how the total comes out even: a surplus in one place covers a shortage in another.

Checkpoint
1 of 2

A bridge girder runs continuously over six supports, with a pin at one end and rollers at the other five. How many internal hinges make it statically determinate?

hinges

Frames: a ledger for any shape

A frame's members are rigidly joined, so its corners carry moment, and it can branch and close back on itself. It takes a ledger that works for any shape.

Unknowns. Cut a member anywhere, and the cut exposes three internal forces: axial force, shear, and moment. Know those at one section, and statics gives them everywhere along the member. So each of the mm members brings three unknowns, and the supports add rr, for 3m+r3m + r in all.

Equations. Cut a joint free, and it's a tiny body with three equations. Hinges add cc, for 3j+c3j + c in all, with jj the number of joints. The Handbook doesn't say what counts as a joint, so count one wherever members meet, and at each support and free end.

The same Handbook section prints this comparison for a plane frame, with too few unknowns as Unstable, a match as Stable and statically determinate, and a surplus as Stable and statically indeterminate. It stops there, and the degree is the difference, (3m+r)−(3j+c)(3m + r) - (3j + c). A beam is a row of members, so m=j−1m = j - 1. Put that in, the 3j3j cancels, and you're back to rr against 3+c3 + c.

Take a portal frame with fixed bases: m=3m = 3, j=4j = 4, and r=6r = 6, so 15 against 12, for degree 3. You can see those three. Cut the girder, and you have two cantilevers, which statics handles easily. All it can't find is the three forces at the cut.

A B C D A B C D N V M N V M
A fixed-base portal frame, and the same frame with its girder cut

Now the hinges. A hinge partway along a member would split it, adding one member and one joint, so 3 to each side, which cancels. Leave mm and jj alone and add 1 to cc.

Where members meet at a hinge, count the swings. Hold one member still, and each of the others can swing on its own. Each swing is a moment that must be zero, so a hinge joining nn members gives n−1n - 1 equations. The Handbook leaves this count to you. A girder pinned into a column that runs straight through counts the same way: only the girder swings, so each pinned girder end, often called a release, gives one equation, even though three members meet there.

n = 2, c = 1 n = 3, c = 2 one pinned girder end, c = 1 two pinned girder ends, c = 2
Each dashed member is a free swing, and each swing is one condition equation

A pin at a support gets no cc. It's already in rr, as two reactions instead of three.

Worked Example A gable frame with a crown hinge
1 of 3

Find the degree of indeterminacy of this frame.

A B C D E
A gable frame with fixed bases and a hinge at the crown

More stories, more bays

A bigger frame uses the same ledger, with more to count. Count columns one story at a time, girders one floor at a time, and joints one level at a time. Then add the condition equations: n−1n - 1 at each internal pin, and one at each pinned girder end.

A second story does change one thing. The upper story is a closed ring of members, and cutting a ring exposes three more unknowns with no support anywhere near. That's why rr against 3+c3 + c stops working here, and why the ledger is worth the trouble.

Worked Example A two-story frame with pinned bases
1 of 3

Classify this frame and give its degree.

A B C D E F H1 H2
A two-story, one-bay frame with pinned bases and a pin in each girder

When the count lies

The Handbook's stability tables end with one sentence: "Stability also requires an appropriate arrangement of members and reaction components." That's its only warning. Passing the count is necessary but not enough, and a structure can pass it and still collapse in three ways.

The reactions are all parallel. That's the plank. Sum forces perpendicular to the reactions, and none of them appears in the equation, so any load in that direction has nothing to balance it. Under purely vertical loads the plank looks fine, and that doesn't rescue it. Stability means holding under any load, not only the one in the sketch.

The reactions are all concurrent. Pin a beam at A, and let its other end roll against a vertical wall. The roller's line runs along the beam, straight through A. Take moments about A, and every reaction has zero arm, so any load with an arm about A turns the beam.

A part forms a mechanism. The supports are fine, but some piece can move anyway. Put three hinges in a row along the girder of a fixed-base portal. The columns stand as cantilevers, and m=3m = 3, j=4j = 4, r=6r = 6, and c=3c = 3 give 15=1515 = 15. Yet the middle hinge drops, like a knee folding. Any three pivots in a straight line, with nothing holding the middle one, will do this, and a pin at a support counts as a pivot.

All parallel r = 3 free to slide All concurrent r = 3 A free to turn A mechanism 3m + r = 15, 3j + c = 15 free to drop
Three structures whose count is satisfied, and what each is still free to do

So after the count, look. Draw each reaction's line: are they all parallel, or do they all pass through one point? Then look for a piece short of holds, or three pivots in a row.

Checkpoint
1 of 2

Each of these beams has three reactions and no hinges, so each passes the count. Which are geometrically unstable?

A.
B.
C.
D.

Change one thing

Questions often change one thing and ask what follows, and you don't need to recount. Each change moves rr or cc by a known step, and the degree moves with it.

What one change does to the count
ChangeWhat movesDegree
Fixed support to pinr down 1Down 1
Pin to rollerr down 1Down 1
Add a rollerr up 1Up 1
Add an internal hingec up 1Down 1

A degree below zero means unstable, because a determinate structure has nothing to spare. An indeterminate one can lose its redundants one at a time and keep standing. Either way, the table only tracks the count, so keep looking.

Checkpoint
1 of 2

Which of these statements are true?

A.
Adding a roller at midspan of a simply supported beam makes it unstable.
B.
Replacing a cantilever's fixed support with a pin makes it unstable.
C.
Inserting an internal hinge in the span of a propped cantilever makes it determinate.
D.
Adding a fourth roller to a beam on three level rollers makes it stable.

Trusses: the same ledger, smaller entries

A truss runs the same ledger with smaller entries. From Trusses: Method of Joints, every member is a two-force member, so it brings one unknown instead of three. Every joint is a particle, so it gets two equations instead of three. Every joint is already a pin, so there's no cc. Compare m+rm + r with 2j2j, as the Handbook's Plane Truss table does, with the same three classifications as its frame table.

This is why trusses are made of triangles. A triangle on a pin and a roller balances at 6 against 6, and each new joint with two new bars raises both sides by 2.

A rectangle doesn't get that deal. Four bars on a pin and a roller give 7 against 8, one short, and it leans over into a parallelogram. A diagonal brings it to 8 against 8. A second, crossing diagonal makes 9 against 8, for degree 1, and so does swapping the roller for a second pin.

The count can lie here too. Take a three-panel truss, and move the middle panel's diagonal into the first panel. The count stays at 16 against 16, but the middle panel now has nothing to stop it racking, and the extra diagonal next door can't help. So after counting a truss, check every rectangular panel for a diagonal.

Stands m + r = 16 2j = 16 Racks m + r = 16 2j = 16
Two trusses with the same count: one stands, and one racks
Checkpoint
1 of 3
A two-panel truss on two pins, with crossing diagonals in one panel

The crossing diagonals aren't connected where they cross. What is the degree of indeterminacy?

Recap

  • Compare unknowns with equations: rr with 3+c3 + c for a beam, 3m+r3m + r with 3j+c3j + c for a frame, and m+rm + r with 2j2j for a truss. Too few is unstable, and the surplus is the degree of indeterminacy.
  • A hinge joining nn members gives n−1n - 1 condition equations, and a pinned girder end gives one. A beam needs c=r−3c = r - 3 hinges to be determinate, with two holds for every piece.
  • Indeterminate means extra restraint, not trouble. Statics alone just can't finish the job.
  • Passing the count isn't enough. Look for reactions all parallel or all concurrent, three pivots in a row, and truss panels without a diagonal.
  • One change moves r or c by a known step, and the degree moves with it.