Safety factor and assessment limits

What the safety factor compares

The bending stress describes the stress state, the material provides a comparison value. SpanDraft relates the two:

S=Rrefσb,maxS=\frac{R_\mathrm{ref}}{\sigma_{\mathrm{b,max}}}
(1)

Here RrefR_\mathrm{ref} stands for the strength value stored in the material: the yield strength ReR_\mathrm{e} or the 0.2 % proof strength Rp0.2R_{\mathrm{p}0.2}. The material data set states explicitly which one applies, and the report shows the formula with the respective symbol. Influences such as component thickness or temperature must already be taken into account when choosing the strength value, as the chapter Material properties describes. SpanDraft uses no further factors.

What S=1S=1 states depends on the strength value. For materials with a pronounced transition, ReR_\mathrm{e} marks the onset of yielding. Rp0.2R_{\mathrm{p}0.2} is the stress on the stress–strain curve of the material at which a permanent strain of 0.2 % is present. In both cases, S=1S=1 states only that the calculated stress reaches this value. SpanDraft does not calculate permanent strains in the component.

ValueMeaning in the model
S>1S>1The calculated bending stress is below the strength value.
S=1S=1The elastically calculated governing extreme fiber stress reaches the strength value.
S<1S<1The elastically calculated stress is above the strength value.
S=∞S=\inftyThe bending stress is zero. This does not mean “infinitely safe”.
Schematic stress–strain diagram of an idealized elastic–plastic material with a linear rise up to the yield strength and a flat material curve thereafter. Points on the straight line mark S greater than 1 and S equal to 1. The point for S less than 1 lies on the dashed extension above the material curve.
Figure 1: Schematic stress–strain diagram of an idealized elastic–plastic material with yield strength Re and the points S > 1, S = 1 and S < 1. Dashed: extension of the elastic straight line.

For S<1S<1, the strength value is exceeded. σb,max\sigma_{\mathrm{b,max}} then remains a calculated value from the linear-elastic assumption and cannot readily be read as the actual stress in the material. SpanDraft continues calculating on the dashed straight line in Figure 1; the material curve does not follow it. For Rp0.2R_{\mathrm{p}0.2}, a different curve shape applies, as shown in the chapter Material properties.

Which input changes what

The chain from load to safety factor can be read off a calculation run. The starting point is the example from the chapter Bending stress and extreme fibers: single-span beam, L=4 mL=4\,\mathrm{m}, F=1300 NF=1300\,\mathrm{N} at 1.2 m, T-section, S235JR. Each row changes exactly one input; II and WW are changed separately for this purpose as manual cross-section values.

Change∣w∣max|w|_\mathrm{max} in mm∣M∣max|M|_\mathrm{max} in Nmσb,max\sigma_{\mathrm{b,max}} in MPaSS
Baseline case3.677109243.265.432
Load · 27.354218486.532.716
E=69 GPaE=69\,\mathrm{GPa}11.19109243.265.432
I⋅2I\cdot21.839109243.265.432
W⋅2W\cdot23.677109221.6310.86
Re=355 MPaR_\mathrm{e}=355\,\mathrm{MPa}3.677109243.268.205
AlMgSi0,5 T6611.19109243.263.467

Three statements can be read from this:

  • Stiffness and strength are separate. EE and II change the deflection but not the stress. A stiffer material does not increase SS.
  • The moment curve here does not depend on EE and II. SpanDraft uses a constant EE and II over the beam. A change of EE or II therefore scales EIEI equally everywhere. For unchanged loads, geometry and support conditions, support reactions and moment curve remain the same, only displacements and rotations change. This also holds in a statically indeterminate system, for example a continuous beam over three supports.
  • A large SS says nothing about the deformation. The last row uses an aluminum alloy from the material library: SS remains clearly above 1 at 3.467, but the deflection is about three times as large.

For real sections, II and WW change together, since both come from the same geometry. If self-weight is enabled, area and density additionally influence the load.

S as a load factor

In the linear model, internal forces and moments and stresses are proportional to the loads. If all loads are scaled by the same factor λ\lambda, the bending stress changes by the same factor. If it is different from zero, it reaches the strength value exactly at λ=S\lambda=S:

σb(λ)=λ σb,max⇒σb(S)=Rref\sigma_\mathrm{b}(\lambda)=\lambda\,\sigma_{\mathrm{b,max}}\quad\Rightarrow\quad\sigma_\mathrm{b}(S)=R_\mathrm{ref}
(2)
Straight line through the origin in the diagram of bending stress and load factor. At load factor 1 lies the calculated value, at load factor S the line reaches the strength value Re.
Figure 2: Bending stress over the load factor λ, to scale for the example with S = 5.432.

Figure 2 shows this for the example with S=5.432S=5.432. A SpanDraft calculation run with F⋅SF\cdot S yields S=1.000S=1.000 there.

This reading presupposes that all loads are scaled by the same factor. If a portion such as the self-weight is not scaled along, a different factor applies. For S>1S>1, λ=S\lambda=S means an increase of the loads, for S<1S<1 a decrease. For S=∞S=\infty, there is no such factor. The statement concerns only the bending stress.

What S = 1 means

S=1S=1 initially means only that the elastically calculated governing extreme fiber stress reaches the stored strength value. It does not follow in general that the component fails exactly at this point or that its load-bearing capacity is exhausted.

A deliberately idealized counterexample is a rectangular cross-section made of an elastic–perfectly plastic material. Elastically, W=bh2/6W=bh^2/6 holds. In the fully plastic state, the stress ReR_\mathrm{e} acts on each half of the cross-section. Each half yields the resultant Re⋅bh/2R_\mathrm{e}\cdot bh/2, which acts at the centroid of its half at ±h/4h/4. The distance between the two resultants is h/2h/2:1

Mel=Re bh26,Mpl=Re bh24=1.5 MelM_\mathrm{el}=R_\mathrm{e}\,\frac{bh^2}{6},\qquad M_\mathrm{pl}=R_\mathrm{e}\,\frac{bh^2}{4}=1.5\,M_\mathrm{el}
(3)
Rectangular cross-section with width b and height h. Next to it, the elastic stress distribution, which reaches the yield strength Re only at the edges, and the fully plastic distribution with the magnitude Re over the entire height.
Figure 3: Elastic and idealized fully plastic stress distribution in the rectangular cross-section.

SpanDraft does not take this plastic reserve into account and uses exclusively the elastic WW. Conversely, S>1S>1 means only that the calculated elastic bending stress is below the strength value. It says nothing about stability or fatigue.

What the assessment does not cover

SS compares exactly one stress with exactly one strength value. The following effects lie outside this evaluation. They are not decisive in every case but must be considered separately depending on the component.

EffectWhy it can change the statement
Shear and equivalent stressShear force produces shear stresses. Where they coincide with a large bending stress, the equivalent stress is higher than the bending stress there.2,3
Axial forceA real axial force superposes as N/AN/A with the bending stress. The model contains no user-defined axial loads.
StabilityCompressed regions can buckle before the strength value is reached. Thin-walled open sections are sensitive to lateral-torsional buckling because of their low torsional stiffness, and the critical load also depends on the lateral support.4
FatigueRepeated loads can accumulate damage even far below the yield strength. A single static stress calculation does not show this.5
Notches and local detailsHoles, shoulders and other geometric discontinuities produce local stress peaks that beam theory does not contain.6 SpanDraft does not assess welds either.
DeformationNo allowable deflection is stored. ∣w∣max|w|_\mathrm{max} is reported but not assessed.
Normative designStandards define their own actions, factors and verification formats. SS is neither such a code check nor an approval.

Analytical

A more comprehensive static stress assessment takes further stress components into account and an equivalent or interaction criterion that matches the material and the verification. Stability, fatigue and local details are verification types of their own.

SpanDraft

SpanDraft evaluates exclusively the elastic bending stress. The calculated shear force is available in the results but does not enter σb,max\sigma_{\mathrm{b,max}} or SS.

Deep dive: example on shear stress

The example shows only one thing: SS takes no shear stress into account. It is no SpanDraft evaluation and no complete assessment of a short beam; shear and equivalent stress are hand calculations.

For a rectangular cross-section and a point force at midspan, τmax/σ=h/(2L)\tau_\mathrm{max}/\sigma=h/(2L) holds, so 2.5 % for L/h=20L/h=20. In slender beams, shear is therefore usually of secondary importance.2 For T and I sections, the transition from web to flange deserves particular attention, because both stresses can occur there clearly at the same time.2

T-section with point A at the transition from flange to web. Next to it, the linear bending stress with the maximum at the edge and the shear stress with the maximum in the centroidal axis and a jump at point A.
Figure 4: Bending and shear stress over the height of the T-section for a positive moment, each scaled on its own.

At point AA, both stresses occur at the same time. At this point, the shear stress and the von Mises equivalent stress can be determined by hand:2,3

τ=V QI t,σv=σ2+3τ2\tau=\frac{V\,Q}{I\,t},\qquad \sigma_\mathrm{v}=\sqrt{\sigma^2+3\tau^2}
(4)

Here QQ is the first moment of area of the partial cross-section above the location under consideration and tt the width there. The table compares the example beam with a beam ten times shorter under ten times the force. Both are statically determinate, so MM and VV follow from equilibrium alone: the largest moment remains the same, the shear force becomes ten times as large. The evaluated location is to the left of the force, with σ=11.33 MPa\sigma=11.33\,\mathrm{MPa} at point AA, Q=23 684 mm3Q=23\,684\,\mathrm{mm^3} and t=10 mmt=10\,\mathrm{mm}.

QuantityL=4 mL=4\,\mathrm{m}L=0.4 mL=0.4\,\mathrm{m}
SS (SpanDraft)5.4325.432
VV in N (SpanDraft)9109100
τ\tau at AA in MPa (hand calculation)1.19711.97
σv\sigma_\mathrm{v} at AA in MPa (hand calculation)11.5223.63

SS remains the same although shear stress and equivalent stress at point AA change markedly. The short beam with L/h=4L/h=4 also lies in a range in which the shear deformation neglected in the Euler–Bernoulli model is no longer readily small. Here this concerns the deflection, not MM and VV.

Display in application and report

Application and report show ∣w∣max|w|_\mathrm{max}, ∣M∣max|M|_\mathrm{max}, σb,max\sigma_{\mathrm{b,max}} and SS. For deflection and moment, the report additionally gives the location and, if applicable, the side. The corresponding operation is described in the help under Bending stress and safety factor.

Calculation is always done with unrounded values. Rounding is applied only for display, by default to three significant digits. The safety factor is never rounded up: a value below 1 never appears as 1, for example 0.9995 as 0.999. A displayed 1, on the other hand, can also stand for 1.004. For this range, the number of digits can be increased in the settings; the calculation does not change as a result.

Deep dive: edge cases of the evaluation

If the bending stress is exactly zero, SpanDraft explicitly returns S=∞S=\infty, without dividing by zero. NaN does not arise in the evaluation. Small stresses are not artificially set to zero, so there is no zero threshold.

In addition, the limits of floating-point numbers apply: if the division for SS overflows, ∞\infty also results. If the stress division underflows, the stress can become zero. A stress that is no longer representable is rejected and not output as an infinite value.

Sources

  1. Laboratory Module 3: Sheet metal forming: elastic-plastic beam bending. 2.002 Mechanics and Materials II, Massachusetts Institute of Technology / MIT OpenCourseWare. 2004. Accessed .
  2. David Roylance. Stresses in Beams. Mechanics of Materials, Massachusetts Institute of Technology / MIT OpenCourseWare. 2000. Accessed .
  3. David Roylance. Yield and Plastic Flow. Mechanics of Materials, Massachusetts Institute of Technology / MIT OpenCourseWare. Accessed .
  4. Rafał Piotrowski; Andrzej Szychowski. Evaluation of the Influence of Bolt Fastener Spacing on the Elastic Critical Load from the Lateral-Torsional Buckling Condition of Built-Up Bending Members. Materials 17 (14), 3392. 2024. DOI 10.3390/ma17143392. Accessed .
  5. David Roylance. Fatigue. Mechanics of Materials, Massachusetts Institute of Technology. 2001. Accessed .
  6. David Roylance. Closed-Form Solutions. Mechanics of Materials, section Stresses around a circular hole, Massachusetts Institute of Technology via Engineering LibreTexts. Accessed .