In short
- End restraint lowers the mid-span moment and the deflection, but it moves moment to the supports, where the connection has to resist it.
- Real connections are rarely perfectly pinned or perfectly fixed; when in doubt, design the span for the pinned case and the support for the fixed case.
- For continuous beams, live load on alternate spans gives larger span moments than full loading.
The four idealisations
- Pinned (simple) support: prevents vertical movement, allows rotation. A steel beam on a fin plate, a timber joist on a wall plate or a precast plank on a ledge is close to this.
- Fixed support: prevents both movement and rotation. A beam cast monolithically into a very stiff wall, or welded to a much stiffer member, approaches it.
- Free end: nothing restrains it — the tip of a cantilever.
- Continuity: a beam running over intermediate supports. The supports themselves may be simple, but the beam's own stiffness ties the spans together.
Uniform-load coefficients
For a prismatic beam of span L under a uniform load w, the results can be written as a coefficient times wL (reactions), wL2 (moments) and wL4/EI (deflection):
| Case | Span moment | Support moment | Max deflection | Reactions |
|---|---|---|---|---|
| Simply supported | 1/8 = 0.125 | 0 | 5/384 = 0.0130 | 0.5 and 0.5 |
| Propped cantilever (fixed–pinned) | 9/128 = 0.0703 | 1/8 = 0.125 | 1/185 = 0.0054 | 0.625 fixed end, 0.375 pinned end |
| Fixed at both ends | 1/24 = 0.0417 | 1/12 = 0.0833 | 1/384 = 0.0026 | 0.5 and 0.5 |
| Cantilever | — | 1/2 = 0.5 | 1/8 = 0.125 | 1.0 at the fixed end |
| Two equal continuous spans, both loaded | 9/128 = 0.0703 | 1/8 = 0.125 | 1/185 = 0.0054 | 0.375, 1.25, 0.375 |
Two details are worth noticing. A propped cantilever and one span of a two-span continuous beam give identical results, because by symmetry the middle support of the continuous beam does not rotate — it acts like a fixed end. And the middle support of a two-span beam carries 1.25 wL, 25 % more than the simple-beam assumption would give it.
A worked comparison
Span 6 m, w = 20 kN/m and EI = 16 000 kN·m² (a steel section with I = 8000 cm⁴):
| Case | Span moment (kN·m) | Support moment (kN·m) | Deflection (mm) |
|---|---|---|---|
| Simply supported | 90 | 0 | 21.1 |
| Propped cantilever | 50.6 | 90 | 8.8 |
| Fixed at both ends | 30 | 60 | 4.2 |
| Cantilever, 6 m long | — | 360 | 202.5 |
Fixing both ends cuts the deflection by a factor of five. But the fixed-ended beam needs connections that can transfer 60 kN·m and a supporting structure stiff enough not to rotate under that moment. If either is missing, the beam behaves closer to the simple case and the mid-span moment rises back towards 90 kN·m.
When fixity is uncertain
Most practical connections are semi-rigid: they transfer some moment, but rotate a little. A cautious and common approach is to bracket the behaviour:
- design the mid-span section for the simply supported moment, which is the upper bound for the span;
- design the support and its connection for the fixed-end moment, which is the upper bound for the support;
- check the deflection with the model you can actually justify, usually the simple one.
A beam built into masonry, for example, is often assumed simply supported for its span and given top reinforcement near the supports to control the cracks that partial fixity causes.
Continuous beams and pattern loading
Dead load acts on every span at once, but live load can be present on some spans and absent on others. For two equal spans, loading only one span raises its maximum span moment from 0.0703 wL2 to 49/512 = 0.0957 wL2, about 36 % more, while the moment over the middle support drops to wL2/16. Design codes therefore ask for the worst arrangement of live load for each effect: alternate spans loaded for the span moments, adjacent spans loaded for the support moment.
Choosing the model in practice
- Look at what the support can physically do: can it resist rotation, and is it stiff enough compared with the beam?
- Pick the idealisation that is safe for the quantity you are checking, or bracket it with two models.
- Make sure the details match the assumption: a beam designed as continuous needs top reinforcement or a moment splice over the supports.
- Write the assumption on the calculation — it is the first thing a checker will ask about.