Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 9

Security classification#01

TECHNICAL REPORT
Discipline or report series#10 Document number#12 Report number#52 Issue#13
Aero Stress Report 1
#15
Title
Analysis of Short Column Buckling.

Authors#20 Telephone: Date#40


S H Watson 45572

Summary#60

This report has been written in response to a comment in the Transmissions and Structures Lessons Learnt Log
to the effect that there are no methods published within Rolls-Royce to calculate the allowable load on a short
column.

This report explains why the classic Euler column buckling equation is inappropriate when the column, or strut,
is ‘short’, ie. has a low slenderness ratio.

The report then gives methods which can be applied in this situation, one for the preliminary sizing of the
column at the design stage, and two which may be used for final analysis.

Additional keywords#90 Retention category


Strut, Column, Euler, Secant, Tangent Modulus.
A or B
Circulation: Approved
Technical Library, by:

Project Sub-system Process Part number

Engine types#24 Engine marks/vehicle name Vehicle number/build#28 Module

Material Design definition Design def. issue CTP#32 ATP#36 Dept#44 File/folder#48

Security classification#01
Page 1 of 9#56
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

1 INTRODUCTION

This report gives guidance on calculating the allowable compressive load for columns (struts) of
different slenderness ratio, particularly when the strut is ‘short’ and the well-known Euler equation no
longer applies. Many ways of calculating a safe allowable compressive load for a short strut have been
proposed which can be confusing. This report limits itself to the methods recommended for use by
Civil Aerospace, Derby. Note, this report only deals with struts which fail by bending of the strut as a
whole, ie the cross section of the strut is stable. The report does not consider struts which fail by
localised buckling of the strut wall.

2 THE EULER (LONG STRUT) EQUATION

Euler is the most well-known strut equation, the derivation for which is in most strength of material
textbooks; the elastic buckling load for a long strut is given by :-

C 2 EI
Peuler  ........(a)
L2

 2 EI
or Peuler  ........(b)
L '2

Where

Peuler = Euler buckling load

C = End fixity constant

E = Young’s modulus

I = 2nd moment of area of strut cross-section

L = Length of strut

L
L’ = Equivalent length of strut L' 
C

Theoretical values of C and L’ are given in the following table for different end conditions.

End Conditions Pinned Pinned Fixed Fixed

Pinned Fixed Fixed Free

C= 1.0 2.0 4.0 0.5

L’ = 1.0 x L 0.7 x L 0.5 x L 2.0 x L

Notes.

(a) C and consequently L’ can be different in different planes.

Document number Issue Security classification


Page 2 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

(b) Many aerospace structures are insufficiently stiff to provide a fixed end condition to the strut.
For instance, in a truss composed of welded tubes, the C=2 and C=4 given in the table above
may, in some design codes, be reduced to 1.5 and 2 respectively.

The Euler equation is not applicable to struts below a certain slenderness ratio because it calculates
the buckling load entirely as a function of the elastic properties of the material and makes no reference
to the material’s strength. The effect the material’s strength can have is shown by :-

Substitute I  Ak 2 into equation (b) and divide through by A gives :-

Peuler  2E
 allowcomp   ............(c)
A  L' k  2
Where :-

A = Cross sectional area of strut

k = Radius of gyration of strut

 allow-comp = Allowable, nominal compressive stress


A typical plot of allow-comp versus L’/k for a given value of E is shown below in fig 1. Also plotted on the
graph are the material’s ultimate and 0.1% proof strengths (strictly speaking these should be for
compressive yield but the RR database usually only has tensile values). If tests were carried out on
real struts then, typically, the failure stress would be expected to lie on a line similar to the one marked
‘test’. The ‘test’ line is below the Euler and proof stress curves because, even before the proof stress
level is reached, the slope of the material’s stress-strain curve is reduced from the E value used in the
Euler calculation. In addition manufacturing tolerances may induce bending of the strut before collapse
and the bending compressive stress adds to the nominal compressive stress further reducing the
effective E value of the strut. A strut has to have an extremely low L’/k ratio before it is capable of
carrying a load such that the compressive stress is close to the ultimate.

Fig 1. Example of Allowable Stress for Strut

1400

1200
Crompressive Stress

1000
Euler
800 Ult

600 Proof
Test
400

200

0
0 20 40 60 80 100 120
L'/k

Document number Issue Security classification


Page 3 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

Many methods have been created to calculate the inelastic buckling load of short struts but they sort
into three categories :-

(a) By assuming an empirical curve can be drawn between L’/k=0 and the Euler curve at large values of
L’/k.

(b) By assuming the strut will fail when the maximum compressive stress (nominal plus bending stress)
reaches a critical level.

(c) By assuming the effective E value will decrease as the effects of plasticity are encountered until a
critical level is reached.

What the engineer requires are methods that; (i) creates a strut which is strong enough to carry the
required load without imposing too great a weight penalty; (ii) is easy to apply; (iii) is acceptable to the
certifying authorities.

3 JOHNSON’S PARABOLIC FORMULA.

This is an empirical curve fit formula and, as such, has largely fallen out of favour in the aerospace
industry. However it is easy to apply and hence useful for the initial sizing of inelastic struts at the
preliminary design stage. The formula states that the allowable stress curve, in the inelastic region, is
given by :-

 proof ( L' / k ) 2
2

 allowcomp   proof  ........(d)


4 2 E

Where:-  proof = 0.1% Proof strength of the material.

 proof
This curve joins, and is tangential to, the Euler curve when  allowcomp  see fig2.
2

Fig 2. Exam ple of Johnson's Parabolic Equation

140
%age Compressive Proof Stress

120

100

80 Euler

60 Johnson

40

20

0
0 50 100 150
L'/k

Document number Issue Security classification


Page 4 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

hy
The effect of eccentric loading can be approximated by dividing the above allowable stress by 1 
k2

where :-

h = load eccentricity.

y = distance from the neutral axis to the extreme fibre.

4 THE SECANT FORMULA.

This well known formula is derived in many text books (see, for instance, ref. 1 or 2). It is in common
usage in Britain, and is the method given in DRA 69. The method assumes that a strut’s allowable load
is determined by the sum of the bending and nominal compressive stress reaching some critical value,
usually taken to be the material’s proof stress. A ‘perfectly straight’ strut will not bend before failure so
it is necessary to have, or artificially introduce, some eccentricity of loading to induce bending.

In its usual form, and using the proof stress as the allowable stress, the allowable load, P allow , is given
by :-

Pallow  hy  L' Pallow 


 proof  1  2 sec 
 .....(e)
A  k  2k EA 

Where :-

Pallow = Allowable compressive load

h = load eccentricity (DRA69 suggests 0.015 inch (=0.38mm) for engine mount links).

y = distance from the neutral axis to the extreme fibre on the concave side of the strut.

Equation (e) is not easily solved because Pallow appears in two places, one of which is inside the secant
term, thus requiring solution by trial and error. To get around this problem ref 1 makes an
approximation for the secant term which leads to the quadratic equation given in DRA69, ie :-

 hy    hy  
1  0.26 2   Pallow  Peuler 1  2    proof A   proof APeuler  0
2
Pallow .....(f)
 k    k  

Notes.

(1) The equation given in DRA69 includes factors of 2240 to convert proof stress in tons/inch 2 to
lbf/inch2. Equation (f) will work in any consistent set of units.

(2) Some early editions of ref. 1 are in error; the equation given in (f) and DRA69 is correct!

(3) When the eccentricity of loading is zero (ie h=0) secant formula no longer applies. The secant
formula says the allowable load follows the Euler curve for high values of L’/k and tracks along a
load=proof.A for low values of L’/k.

(4) Fig 3 shows how the allowable stress varies with eccentricity ratio according to the secant formula.
Document number Issue Security classification
Page 5 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

Fig 3 Typical Plot of Allowable Strut Stress for Differing


Eccentricity Ratios using the Secant Formula

120
%age of Compressive Proof Stress

100

80 hy/(k^2)=0.0
hy/(k^2)=0.2
60 hy/(k^2)=0.4
hy/(k^2)=0.6
40 hy/(k^2)=1.0

20

0
0 50 100 150 200
L'/k

5 THE TANGENT MODULUS METHOD.

This method for inelastic buckling is in general use in the American aerospace industry so should be
acceptable to the FAA and DoD. It uses the Euler formula but replaces E with the local tangent slope
of the stress-strain curve, also known as the tangent modulus and denoted E t. The justification for this
modification to the Euler formula is given in many texts, see ref 2.

Pt  2 Et
t   .....(g)
A  L' k  2
Where :-

t = Critical buckling stress


Pt = Critical buckling load

Et = Tangent Modulus

Unfortunately Et is a function of applied stress so equation (g) would has to be solved iteratively.
However Mil-Hdbk-5 (The Dept of Defense materials handbook) contains plots of t versus Et for
many materials which makes the calculation very simple. If such a plot is available then equation (g) is
rearranged :-

Document number Issue Security classification


Page 6 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

t 2
 .......(h)
Et  L' k  2

For a strut with a given L’/k ratio calculate the right-hand side of (h). A straight line with this value of
slope, and passing through the origin, is plotted on the t vs Et graph. Where the two lines intersect
gives the value of t and consequently Pt.

As an example. Assume that a graph of t vs Et is available (see fig 4) and it is required to establish
the buckling load of a strut with L’/k=30. From (h) the slope of the straight line=0.01097 The
intersection of this straight line with the tangent modulus curve gives an allowable column stress of
900 MPa.

Fig 4 Use of the Tangent Modulus Curve

1600

1400

1200

1000
Stress (MPa)

Stress vs Et
800
Slope=0.01097
600

400

200

0
0 50 100 150 200 250
Et (GPa)

A problem with this method is the lack of stress versus E t curves in the RR materials database. The
Materials Laboratory may be able to supply stress-strain data which will allow the tangent modulus
curve to be plotted or give an equivalent material code which is in Mil-Hdbk-5. An alternative is if the
Ramberg-Osgood parameter for the material is available then an approximate stress () versus Et
graph can be drawn from the following relationship given in ref 3.

1
Et 
1 E   0.002n  proof   proof  n 1

An appreciation of the Tangent Modulus method, and the effect of the Ramberg-Osgood parameter,
can be gained from fig 5 which shows, in non-dimensional form, the relationship between allowable
stress and slenderness ratio. The Ramberg-Osgood parameter ‘n’ changes the shape of the stress-
strain curve in the region of the onset of plasticity. n=7 represents a normalised low alloy steel whilst
n=50 represents a similar steel heat treated to an ultimate strength of 1240 MPa (both at room
temperature). Table B1.1 of ref 4 gives n for numerous materials.

Document number Issue Security classification


Page 7 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

Fig 5 Nondimensional Tangent Modulus Column Curves


using the Ramberg-Osgood Parameter

140.00
%age of Compressive Proof Stress

120.00

100.00

Euler
80.00
n=10
n=20
60.00
n=50

40.00

20.00

0.00
0 1 2 3 4 5 6 7
sqrt(Proof Stress/E).L'/k

6 FINITE ELEMENT METHODS.

The modern method of establishing a strut’s buckling load is to use a finite element program and this
could be deemed the only method in the case of non-uniform cross-section struts or assemblies of
struts. In order to calculate anything other than the equivalent to the Euler elastic stability load it is
necessary to do a non-linear analysis switching on (a) geometric non-linearity to account for large
deflections or eccentric loading and (b) material property non-linearity to account for the change in E
value beyond the proportional limit. Note that in the analysis of a ‘perfectly straight’ column it may be
necessary artificially to introduce an extremely small amount of load eccentricity in order to calculate
the correct buckling load. If you are unfamiliar with this type of analysis it is strongly recommended
that some simple test cases are performed to ensure the software is being used correctly.

7 FACTOR OF SAFTEY

In the absence of any test evidence it would be usual to apply a factor of safety of 1.5 to the above
results to establish the safe working load of a strut.

8 ADDITIONAL COMMENTS

All the above formulae assume that the strut has a stable cross-section, ie. the strut will buckle by
overall bending. For thin wall compression members there is the possibility of the strut failing by
localised buckling of the wall. This secondary type of failure is outside the scope of this report but a
lot of data is available in books on aircraft structures and the reader is directed to references 3 & 4 as
a good starting points.
Document number Issue Security classification
Page 8 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.
Security classification

The methods given in this report refer only to columns with cross-sections that have two axes of
symmetry, ie circular, rectangular, elliptical, I beams etc. In these cases flexural instability can be
considered about the two axes independently. If the cross-section has one or no axis of symmetry eg.
a C, T or L section then there is the possibility of buckling in a combined bend and torsion mode. A
special case is the cruciform + cross-section which may also fail in a torsion mode. Ref 4 would be a
good starting point for these cases.

If the column is carrying lateral loads as well as the compressive load then it should be analysed as a
beam-column because the two loadings interact; ie. the bending moment due to the lateral load will
be increased by the presence of the compressive load. Usually this would be analysed by non-linear
finite element analysis although ref 4 has some analytical solutions.

Ref 4 also has some information on columns whose cross-section changes along the column’s length
although usually these would be analysed using finite elements.

9 REFERENCES
1 Strength of Materials, J Case & A H Chilver :- Edward Arnold, 1959.

2 Mechanics of Materials, J M Gere & S P Timoshenko :- PWS, 1985.

3 Airplane Structures, A S Niles & J S Newell :- Chapman & Hall, 1954.

4 Analysis & Design of Flight Vehicle Structures, E F Bruhn :- Jacobs, 1973.

Document number Issue Security classification


Page 9 of 9
©2018 Rolls-Royce plc
The information in this document is the property of Rolls-Royce plc and may not be copied, or communicated to a third party, or
used, for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc.

You might also like