Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

Journal of Building Material Science

https://journals.bilpubgroup.com/index.php/jbms

ARTICLE

Experimental Study and Fragility Analysis of Effective-Length

Factors in Column Buckling
Brannan Shepherd, Tadeh Zirakian*

Department of Civil Engineering and Construction Management, California State University, Northridge, CA, 91330, USA

ABSTRACT
The design of columns relies heavily on the basis of Leonhard Euler’s Theory of Elastic Buckling. However, to
increase the accuracy in determining the maximum critical load a column can withstand before buckling, a constant
was introduced. This dimensionless coefficient is K, also known as the effective-length factor. This constant is often
found in building design codes and varies in value depending on the type of column support that is applied. This
paper presents experimental and analytical studies on the determination of the effective-length factor in the buckling
stability of columns with partially-fixed support conditions. To this end, the accurate K value of the columns tested by
the Instron Testing Machine (ITM) at California State University, Northridge’s (CSUN’s) Mechanics Laboratory is
determined. The ITM is used in studying the buckling of columns where the supports are neither pinned nor fixed, and
the material cross-section rather rests upon the machine while loading is applied axially. Several column specimens
were tested and the experimental data were analyzed in order to estimation of the accurate effective-length factor. The
calculations from the tested results as well as the conducted probabilistic analysis shed light on how a fragility curve
may aid in predicting the effective-length value of future tests.
Keywords: Column buckling; Fragility curve; Critical loading; Effective length factor; Fixity

1. Introduction imum loading is applied. This occurs mainly in

Column buckling can be defined as the abrupt members with a high slenderness ratio, meaning the
lateral displacement of a column under which max- length of the member is much greater than the width

*CORRESPONDING AUTHOR:
Tadeh Zirakian, Department of Civil Engineering and Construction Management, California State University, Northridge, CA, 91330, USA;
Email: tadeh.zirakian@csun.edu
ARTICLE INFO
Received: 30 January 2023 | Revised: 5 March 2023 | Accepted: 13 March 2023 | Published Online: 20 March 2023
DOI: https://doi.org/10.30564/jbms.v5i1.5432
CITATION
Shepherd, B., Zirakian, T., 2023. Experimental Study and Fragility Analysis of Effective-Length Factors in Column Buckling. Journal of Building
Material Science. 5(1): 9-16. DOI: https://doi.org/10.30564/jbms.v5i1.5432
COPYRIGHT
Copyright © 2023 by the author(s). Published by Bilingual Publishing Group. This is an open access article under the Creative Commons Attribu-
tion-NonCommercial 4.0 International (CC BY-NC 4.0) License. (https://creativecommons.org/licenses/by-nc/4.0/).

9
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

of the cross-section. Shorter-length members will effective length of a column. Hibbeler defines the
not experience buckling but rather crushing under effective length of a column as the distance between
greater loads. Regardless, critical loads for columns points of zero moments [1,2]. Therefore, supports that
of any slenderness ratio can be calculated. Leonhard limit rotation will provide a moment reaction and
Euler’s Theory for Elastic Buckling produced such produce an internal inflection point. This is the case
an equation which is as follows: with completely fixed supports. A moment reaction
2  at both the top and bottom of the member decrease
 =
( )2  (1)
the effective length resulting in an effective length
2 
 =
( )where
2
E is the modulus of elasticity of the material factor of K = 0.5. For pinned supports, there are no
being used, l is the length of the member, and r is the moment reactions at the end conditions therefore,
radius of gyration, which is equivalent to: the effective length results in the full length of the
= member since the closest values of zero moments are

= (2) at the connections. An experimental example of this
2 
 =  2
( )
Here, I is seen as the bending moment of iner- can be cited by Bouras et al. [3]. The experimental
tia and2A is the cross-sectional area of the member. design utilizes a notched cylinder that accepts the
 = column, and a welded angle with a Teflon sheet that
()2
2  With Equation (1), values obtained for various
 = guides the cylinder in rotation emulating a pinned
()2columns with differing support connections yield
= connection. With this setup, the effective length factor
unreliable results.
 By substituting in the effective-
length constant “K” of the column being analyzed, used is K = 1.0 since the resulting zero moments will
the equation then converts to: lie at the connections.
As previously stated, not only is the support
2 
 = (3) condition the only variable that plays a role in the
()2

This then provides more accurate critical load re- overall effective length. Referring to Tian et al. [4],
sults theoretically as well as for physical applications [1]. columns can contain various stiffnesses, cross-
Specific K values are to be used when calculating sectional areas, distributions in axial loading, and
critical loads with different column support connec- differing ratios between lengths. These variables may
tions. For example, columns that have fixed connec- only take effect when considering stepped columns
tions on both ends will utilize a K value of 0.5, and composed of materials other than wood elements.
columns with simply supported pinned connections However, the theory and determination of critical
will use a K value of 1. The goals established before load behaviors still extend from the foundational
conducting any experimentation were to calculate Elastic Buckling Theory. Expanding on the notion
an accurate K value for the supporting connections that many studies in modern engineering stem from
on California State University, Northridge’s (CSUN) Euler’s Theory, in Falborski et al.’s study of column
Instron Testing Machine (ITM), since it will be used base fixity [5], the base support of a column within a
for this case study specifically, and undergo probabil- frame can be manipulated to obtain a certain degree
istic analysis of results. This is achieved by creating of flexibility. With this flexibility, the inflection
a fragility curve that aids in predicting the probabili- points of the member can be adjusted to lessen the
ty of K values that would be produced in future tests. amount of internal moment received at the upper end
of the column where the structure is more vulnerable,
and result in an increase in the moment at the base
2. Literature review where the foundation can resist greater moment reac-
The primary focus of this case study is on the tions. This enables the structure to withstand higher
effective length constant for columns, however, there amounts of lateral loadings, such as wind or earth-
are far more variables that can change the overall quakes, all while under uniform axial loading.

10
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

In any study involving structural behavior under teration, the behavior of columns will ultimately cre-
extreme loading, the allowable stress and strain of ate effects higher up the chain in structural analysis.
material are vital. In Euler’s Theory, the modulus of By establishing a better understanding of how
elasticity is the primary component relating stress effective length factors and fragility curves function
and strain considering the modulus under axial with a single member, made of wood in this case, the
loading is the quotient of normal stress and strain. more advanced theories and concepts may begin to
Avallone et al. [6] display these relations many times see effects as well within larger systems, structures,
using stress-strain curves. When members are placed and materials. They branch off from the basic
under increasing axial loading, the elements’ materi- understanding of column buckling and its reactions
al properties of stress and strain demonstrate a line- to compressive loading.
ar-proportional relationship which is called the elas-
tic range. Outside this range lies other circumstances 3. Experimental design
with vibrational forces and accelerations that will not
The support connections for the ITM are neither
be covered in this experiment but still are immensely
fully fixed nor pinned, therefore, the end connections
important when pertaining to structural designs in
will be referred to from here on out as partially
earthquake engineering.
fixed. For example, as seen in Figures 1 and 2, the
In the analysis of earthquake effects on struc-
cross-section of the member is rested flush to the
tures, seismic fragility curves have been explored
crossheads of the machine, which only partially
with the goal of determining the vulnerability of
limits rotational and translational displacements.
structures if induced by such type of lateral loading.
Figure 3 shows how the test specimen is positioned
In this paper, the basic use of fragility curves will be in ITM. 15 Douglas Fir specimens roughly 35 inches
explored in determining the probability of effective in length were prepared to undergo compression in
length constants however, studies such as one under the partially fixed machine connections. Calculations
Rajeev and Tesfamariam [7] reveal more in-depth such as weight, volume, density, cross-sectional area,
uses of this tool. In their study, they analyze multiple slenderness ratio, and moment of inertia were then
building failures due to earthquakes and compile taken for each sample as shown in Table 1. For this
several variables such as a number of weak stories, experiment, a Modulus of Elasticity (E) of 1.95×106
year of construction, beam-column connections, and psi is established considering this is the same value
the topology of the site. With these factors applied dedicated to the wood used in CSUN’s laboratory for
they can create relations between the demand of the other experiments. This also maintains the accuracy
earthquake in question and the structural capacity of of results in relation to past and future tests con-
the building to make predictions on other buildings ducted at CSUN. Each specimen was then carefully
similar in design and determine its probability of positioned in the ITM crosshead and testing was
survival or failure with an earthquake of similar conducted through the program TestWorks QTEST.
magnitude. Guevara-Perez [8] also discusses the Information about the specimen being tested such
importance of beam-column connections in the result as length and cross-sectional area was input into the
of weak stories that are a major factor in the ability program and then, at a rate of 0.01 in/min, the cross-
of a structure to successfully endure an earthquake. heads of the ITM compressed the specimens axially
Many factors play a role in the behavior of col- until buckling or complete failure occurred. QTEST
umns and through a review of various studies involv- then provided peak stress (σcr) and critical load (Pcr)
ing advanced concepts in engineering we can see that values from each sample tested. The details of the
at the foundation level, column buckling shows there ITM as well as the length and cross-section dimen-
are key differences in behavior when something as sions of a typical test specimen are illustrated in Fig-
simple as a support condition is altered. With this al- ure 4.

11
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

Figure 1. Top crosshead connection. Figure 2. Bottom crosshead connection. Figure 3. Specimen placed in ITM.

Table 1. Properties of the test specimens.

Greatest Least
Cross
Cross Cross Slenderness Slenderness Moment
Wood Length Section Weight Volume Density
Section Section Ratio Ratio of Inertia
Specimen (in) Area (oz) (in3) (oz/in3)
Dimension Dimension (Min) (Max) (in4)
(in2)
(Dmax, in) (Dmin, in)
1 35.000 0.727 0.693 0.504 6.30 17.633 0.357 50.505 48.143 0.0202
2 35.063 0.729 0.716 0.522 5.30 18.301 0.290 48.970 48.097 0.0223
3 35.000 0.734 0.726 0.533 6.70 18.651 0.359 48.209 47.684 0.0234
4 35.063 0.719 0.712 0.512 5.50 17.949 0.306 49.245 48.766 0.0216
5 35.000 0.721 0.714 0.515 5.30 18.018 0.294 49.020 48.544 0.0219
6 35.000 0.724 0.718 0.520 6.70 18.194 0.368 48.747 48.343 0.0223
7 35.000 0.728 0.717 0.522 6.50 18.269 0.356 48.815 48.077 0.0224
8 35.063 0.727 0.715 0.520 6.30 18.226 0.346 49.038 48.229 0.0221
9 35.063 0.717 0.709 0.508 5.40 17.824 0.303 49.453 48.902 0.0213
10 35.000 0.729 0.718 0.523 6.90 18.320 0.377 48.747 48.011 0.0225
11 35.063 0.729 0.725 0.529 6.30 18.531 0.340 48.362 48.097 0.0232
12 35.000 0.733 0.728 0.534 6.50 18.677 0.348 48.077 47.749 0.0236
13 35.063 0.733 0.716 0.525 5.50 18.402 0.299 48.970 47.834 0.0224
14 35.063 0.719 0.727 0.523 5.40 18.328 0.295 48.229 48.766 0.0230
15 35.000 0.738 0.728 0.537 6.90 18.804 0.367 48.077 47.425 0.0237

4. Evaluation of experimental results and a bend in the specimen could be seen. All data
obtained from the testing machine was placed into
After a certain amount of compression, each tabular form and comparisons were made between
column either buckled or completely failed to re- the experimental and theoretical critical load based
sult in fractures, as shown in respective Figures 5 on support connections, which can be seen in Table
and 6. Some members that were not fractured from 2. Average experimental values for peak stress and
testing remained in the elastic range and reverted to critical load were 2,278.69 psi and 1,188.77 lbs,
their original linear state after loading was removed. respectively. We can assume that variations in both
However, there was slight deformation left over values are due to the inhomogeneity of the wood

12
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

Figure 4. Details of the ITM and geometrical dimensions of a typical test specimen.

Figure 5. Specimen #6 buckling under compression. Figure 6. Specimen #3 ruptured after compression.

13
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

Table 2. Summary of theoretical calculations and test data.

Experimental
Peak Effective
Wood Fixity Theoretical Pcr Theoretical Pcr Experimental Pcr Normal
Stress Length
Specimen (%) (SS) (FF) (PF) Distribution
(psi) Factor
(K)
1 77.115 316.775 1,267.101 839.10 1,665.60 0.614 0.962
2 90.104 349.087 1,396.349 1,156.20 2,245.10 0.549 0.521
3 91.834 367.725 1,470.899 1,257.20 2,359.30 0.541 0.429
4 89.931 338.560 1,354.242 1,117.80 2,183.60 0.550 0.530
5 75.990 343.595 1,374.380 893.70 1,736.00 0.620 0.973
6 93.781 350.856 1,403.424 1,243.90 2,393.00 0.531 0.331
7 101.037 351.322 1,405.290 1,434.90 2,749.00 0.495 0.081
8 92.223 346.673 1,386.692 1,193.80 2,296.60 0.539 0.409
9 97.101 333.369 1,333.476 1,259.40 2,477.50 0.514 0.190
10 98.631 353.279 1,413.116 1,375.20 2,627.30 0.507 0.140
11 79.898 362.417 1,449.669 1,005.00 1,901.60 0.601 0.920
12 84.400 370.267 1,481.068 1,108.30 2,076.90 0.578 0.791
13 96.486 351.003 1,404.011 1,310.30 2,496.60 0.518 0.213
14 92.249 360.412 1,441.649 1,241.70 2,375.50 0.539 0.408
15 96.614 372.793 1,491.171 1,395.10 2,596.70 0.517 0.208
Average 90.493 351.209 1,404.836 1,188.77 2,278.69 0.548
Std. Dev. 0.038

members themselves. Some members had less or in-between fixed and pinned connections but resem-
greater densities as well as internal defects, seen bles fixed circumstances more. This is due to the pre-
in most wooden materials. For example, a material viously stated fact that the partially fixed crossheads
such as steel or any other metal alloy is pure and of the ITM do not completely limit rotation, so the
homogeneous throughout the entire length of the columns are able to, just slightly, have an increased
member, whereas wood may have internal knots in effective length. Figure 8 shows this slight rotation-
the wood or initial fractures that may not be seen on al freedom when subjected to the compressive load.
the surface. Since we were given the experimental To further analyze the degree to which the partially
Pcr, the experimental K-value for each specimen was fixed supports constrain rotations and translations, its
calculated and an average effective length constant fixity percentage was calculated.
of 0.548 was found. Effective length constants from
each member can be seen in Table 2. Figure 7 al-
lows a visual representation of the variability in ef-
fective length values. Clear observations were made
for the tested columns and a general effective length
ratio lies somewhere between 0.5 and 0.6.
Using Equation (3), theoretical critical loads with
fixed and pinned connections were calculated and the
results show that the experimental critical load lies Figure 7. Effective length constants of test specimens.

14
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

the generation of a fragility curve. Fragility curves

are S-shape plots primarily used in earthquake
engineering to assess the damage state of structures
or to aid in the prediction of how much damage a
structure will encounter when earthquake conditions
are applied. This is done through a cumulative
probability distribution plotted with earthquake
intensity and damage grades. This method of analysis
can also be applied to column buckling. To do this,
the mean and standard deviation of the effective
length values from the entire column set is applied
to a cumulative distribution function. These values
Figure 8. Upper crosshead support under loading.
can also be seen in Table 2. Then, the probability
distribution was plotted against the original
5. Probabilistic analysis respective K-values obtained from experimental
testing. The plot is shown in Figure 9.
Columns contain a certain level of fixity in the
supports. What influences this level of fixity is the
degrees of freedom allowed in the support condition.
For example, a fixed-end condition restricts transla-
tion in the x, y, and z directions as well as rotation
therefore it contains fixity in all six degrees of free-
dom. To further analyze the behavior of columns
placed in the ITM, calculations of fixity percentage
were obtained. Taking the average K-value of 0.548
from the experimental data, interpolations between
the fixed and pinned supports were made. The equa- Figure 9. Fragility curve.
tion to determine this value is as follows:
Along the ordinate, we can see the range is from
0.548−0.50 −100 0.548−0.50
1−0.50
= 0−100
 = − 0.50
∙ 100zero
+ 100
to one. This is the probability percentage that
(4) a given effective length value will show. In other
−100 0.548−0.50
 = − ∙ 100 + 100
0−100 0.50 words, by taking any reference point along the curve
Allowing fixed supports to have 100 percent fix- the area underneath the curve represents the proba-
ity and a K-value of 0.50, pin-supported conditions bility that the respective effective length ratio will
having zero percent fixity and a K-value of 1.0, occur. For example, at the effective length value of
interpolation between the two conditions concluded 0.55 the cumulative probability outputs a value of
that the columns tested under partially fixed obtained about 53 percent. This means that for future testing
an average fixity percentage of 90.49 percent. All of these columns, there is a 53 percent probability
values for individual members can be seen in Table 2. that the resulting effective lengths between 0.495
Again, this supports the observation that a partially and 0.55 will be produced. The same analysis is
fixed support such as the one seen in CSUN’s ITM done from the higher and lower ends of the curve.
lies closer in relation to fixed-end conditions than At the effective length reference point of 0.6 the
simply supported, but still results in displaying its corresponding probability is around 10 percent that a
own distinct column behavior. greater effective length factor is produced and at an
Further probabilistic analysis was done through effective length factor of 0.5 the probability is less

15
 Journal of Building Material Science | Volume 05 | Issue 01 | June 2023

than 10 percent to produce a value less than or equal appreciation for funding made possible in support
to 0.5. This coincides with the behavior of the col- of this research endeavor through the CSU-LSAMP
umns and concludes that through further testing, 80 (California State University Louis Stokes Alliance
percent of columns will produce an effective length for Minority Participation) program via the NSF
between 0.5 and 0.6 with a larger majority resulting (National Science Foundation) grant #HRD-1302873
closer to 0.55. and the Chancellor’s Office of the California State
University.
6. Conclusions
References
In this paper, the determination of the effec-
tive-length factors of columns with partially-fixed [1] Hibbeler, R.C., 2023. Mechanics of materials
end supports and subjected to axial loading was in- (9th edition). Upper Saddle River: Prentice Hall.
vestigated through the adoption of experimental and [2] Crandall, S.H., Dahl, N.C., Lardner, T.J., et al.,
analytical approaches. To this end, a case study on 2012. An introduction to the mechanics of solids
the testing of columns by CSUN’s ITM was conduct- (3rd Edition). McGraw Hill: New York.
ed and discussed. The support conditions of CSUN’s [3] Bouras, F., Chaplain, M., Nafa, Z., 2010. Exper-
ITM allow the columns to produce an alternate be- imental and modelling buckling of wood-based
havior due to external compression that is not seen columns under repeated loading. EPJ Web of
under traditional support connections such as fixed Conferences. 6, 28003.
and pinned conditions. A partially-fixed connection DOI: https://doi.org/10.1051/epjconf/20100628003
with an average effective length of about 0.55, al- [4] Tian, W., Sun, J., Hao, J., 2021. Effective length
though not far from fixed conditions, still reveals factors of three- and two-segment stepped col-
umns. Journal of Constructional Steel Research.
there is slightly different behavior and is significant
181, 106585.
enough to be classified as its own support condition.
DOI: https://doi.org/10.1016/j.jcsr.2021.106585
Creating a fragility curve of the column behavior
[5] Falborski, T., Hassan, A.S., Kanvinde, A.M., et
also provided a predictability tool to determine how
al., 2020. Column base fixity in steel moment
the column will behave under future testing. The ef-
frames: Observations from instrumented build-
fectiveness of the adopted experimental, analytical,
ings. Journal of Constructional Steel Research.
and probabilistic approaches in the determination of
168, 105993.
the accurate effective-length factor in columns with
[6] Avallone, E., Baumeister, T., Sadegh, A., 2006.
partially-fixed supports was demonstrated in this re-
Marks’ standard handbook for mechanical engi-
search endeavor. In particular, the application of the
neers (11th Edition). McGraw Hill: New York.
fragility methodology in efficient buckling stability
[7] Rajeev, P., Tesfamariam, S., 2012. Seismic fra-
assessment of structures seems to be quite promis-
gilities for reinforced concrete buildings with
ing.
consideration of irregularities. Structural Safety.
39, 1-13.
Conflict of Interest DOI: https://doi.org/10.1016/j.strusafe.2012.06.001
There is no conflict of interest. [8] Guevara-Perez, T., 2012. “Story” and “Weak Sto-
ry” in Earthquake Design: A Multidisciplinary
Approach. Indian Institute of Technology Kanpur
Acknowledgement [INternet]. Available from: https://www.iitk.ac.in/
The authors would like to express their great nicee/wcee/article/WCEE2012_0183.pdf

16