materials
Article
Experimental Investigation on the Vertical Ductility of Rectangular
CFST Columns Loaded Axially
Bartosz Grzeszykowski *






Citation: Grzeszykowski, B.;
Szmigiera, E.D. Experimental
Investigation on the Vertical Ductility
of Rectangular CFST Columns
Loaded Axially. Materials 2022, 15,
2231. https://doi.org/10.3390/
ma15062231
Academic Editor: Dario De
Domenico
Received: 13 February 2022
Accepted: 14 March 2022
Published: 17 March 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Materials 2022, 15, 2231. https://doi.org/10.3390/ma15062231
and Elżbieta Danuta Szmigiera *
Faculty of Civil Engineering, Warsaw University of Technology, Al. Armii Ludowej 16, 00-637 Warsaw, Poland
* Correspondence: bartosz.grzeszykowski@pw.edu.pl (B.G.); elzbieta.szmigiera@pw.edu.pl (E.D.S.)
Abstract: A total of 5 steel and 21 rectangular composite concrete-filled steel tube (CFST) columns
of moderate slenderness were tested to investigate their ductility under axial compression. The
importance of the vertical ductility of columns was discussed, and a novel ductility measure was
proposed and utilized to examine the ductility of tested specimens. The analyses showed that the
ductility of axially compressed CFST columns highly depends on their failure mode. The key feature
influencing the ductility is their ability to dissipate the energy of imposed loads. The larger the
volume of a material that may permanently deform and consequently dissipate the energy, the greater
this ability. In consequence, the ductility of specimens exhibiting local failure mode was higher
in comparison to the columns that underwent global or mixed global—local failure. It was found
that both steel and composite columns were able to carry axial loads in the post-critical state; but
due to the limitations of local buckling of the steel cross-section in the concrete core and concrete
confinement, all tested composite columns showed greater ductility than their steel counterparts.
Keywords: ductility; pancake type collapse; composite columns; CFST; steel; concrete; experimental
research; axial load
1. Introduction
The vertical ductility of columns is an essential parameter when considering the
responses of structures subjected to accidental events, such as fires, impacts, strong earth-
quakes, and explosions (Figure 1a–c). Such triggering events can, in particular, lead to
scenarios involving collapse. Such scenarios are depicted in Figure 1d,e, in which columns
of the entire floor of a high-rise building collapse, causing the higher floors to impact the
supporting columns of the lower part of the building. In the first case (Figure 1d), the
axial bearing capacity or the vertical ductility of columns is high enough to transfer the
temporary impact loads to the foundations and the structure retains its overall stability. In
the latter case (Figure 1e), the opposite is the issue and a cascade failure occurs, leading to a
progressive collapse of the entire building called a “pancake” type collapse. If the dynamic
impact force is lower than the axial bearing capacity of the columns, the collapse does not
occur. Otherwise, the collapse may be arrested only if the available vertical ductility of
columns is greater than the ductility demand.
https://www.mdpi.com/journal/materials
Materials 2022, 15, 2231 2 of 26
Materials 2022, 15, x FOR PEER REVIEW 2

Figure
Figure 1. Possible 1. Possible
triggering triggering
events events
that can leadthat
to acan lead to a progressive
progressive collapse of acollapse
building:of a(a)
building:
fire; (a) fir
crash with an object or explosion; (c) strong earthquake—consequences of the
(b) crash with an object or explosion; (c) strong earthquake—consequences of the removal of an entire removal of an e
story; (d) local failure does not lead to a progressive collapse due to a sufficient strength or duc
story; (d) local failure does not lead to a progressive collapse due to a sufficient strength or ductility;
(e) progressive collapse, insufficient strength, or ductility of remaining columns, based on [1].
(e) progressive collapse, insufficient strength, or ductility of remaining columns, based on [1].
In the
In the literature, theliterature, the word
word ductility ductility
is often usedistooften usedthe
describe to describe the deformabilit
deformability of
tested specimens and to categorize them in terms of their ductility without using any directusing an
tested specimens and to categorize them in terms of their ductility without
measure [2–9]. rectInmeasure
such cases,[2–9].
theInrate
such
of cases,
change theofrate
the of change
force in theofpost-peak
the force in theofpost-peak
part the pa
the equilibrium path is analyzed in a qualitative manner. The
equilibrium path is analyzed in a qualitative manner. The ductility of structural members ductility of structural m
bers can also be directly calculated using various measures, usually
can also be directly calculated using various measures, usually referred to in the literature referred to in th
eratureMost
as ductility indexes. as ductility indexes.values
often, ductility Most are
often, ductility
defined values
as the are defined
quotients as the quotien
of generalized
generalized displacements, such as deflections, rotations, or curvatures
displacements, such as deflections, rotations, or curvatures [10]. Two main types of analyses [10]. Two m
including thetypes of analyses
vertical ductility including
of columns the vertical
can ductility of columns can be distinguished:
be distinguished:
• Structural Structural
• design design analysis;
analysis;
• •
Comparison Comparison
of of of
the ductility thestructural
ductility elements.
of structural elements.
In the
In the first case, the first case, of
ductility thethe
ductility
structure of the structure in
is calculated is calculated in order
order to check to check its stab
its stability
in an accidentalin an accidental
event. event. The check
The calculations calculations
whether check whether the
the ductility ductility
demand for ademand
given for a g
excitation is lower than the available ductility of a structure.
excitation is lower than the available ductility of a structure. In the latter, the ductility In the latter, the ductili
calculated as one of many parameters in addition to strength,
is calculated as one of many parameters in addition to strength, stiffness, durability, etc., stiffness, durability,
whichthe
which characterize characterize
elements testedthe elements
and enable tested and enable analysis
a comparative a comparative analysis
of a given designof a given
sign solution
solution in relation to otherin types
relation of to other types
structures. of structures.
Depending Depending
on the measureon of the measure of du
ductility
ity adopted for
adopted for calculations, the calculations, the ductility
ductility of elements mayof elements
differ [11]. may differ [11].
Severalofdefinitions
Several definitions of vertical
vertical ductility ductility
measures of measures
columns have of columns
been usedhave inbeen
the used in
literature (seeliterature
Figure 2). (seeAllFigure
of them 2). rely
All of onthem rely on the characteristic
the characteristic points of the points of the equilibr
equilibrium
paths, such aspaths,
xy , xmsuch
, andas xy,, xwhich
x0,85 x0,85, which
m, anddenote yielddenote yield displacement,
displacement, the displacement
the displacement at the a
ultimate
ultimate strength, andstrength, and the
the post-peak post-peak displacement
displacement when the load when the load
carrying carrying
capacity has capacity
undergone a undergone
15% reduction a 15% reduction
in load, in load, respectively.
respectively. The availableThe available
ductility of a ductility
column isofina column
fact the
fact the post peak post peak displacement
displacement for a given
for a given reduction reduction
in load, and xin load,
y or xy ormainly
andused
xm are xm are used ma
for normalization
for normalization purposes. purposes.
 Materials 2022, 15, x FOR PEER REVIEW 3 of 29
Materials 2022, 15, 2231 3 of 26

Ductility
Measure
xm/xy

x0,95/xy
x0,85/xy

x0,85/xm

xy*/xm

Figure 2. Definitions of vertical displacement ductility factors [12–28].

Figure 2. Definitions of vertical displacement ductility factors [12–28].
The ductility measures defined in Figure 2 are analogous to horizontal ductility mea-
sures used inThe ductility
seismic measures
analysis for thedefined
reduction in Figure
of elastic2 are analogous
horizontal to horizontal
seismic forces [29].ductility
measures used in seismic analysis for the reduction of elastic
However, in the case of vertical ductility of columns, they do not have a clear physical horizontal seismic forces
[29]. However,
interpretation. Relating in thethe case
post of vertical
peak verticalductility
displacement of columns, they do
of a column notdisplacement
to the have a clear physical
interpretation.
at its first Relating to
yield is not essential theverify
post peak verticalof
the stability displacement of a column
columns subjected to the displace-
to accidental
ment
axial loads. at its firsthowever,
Important, yield is isnot
theessential to verify
relationship the stability
of column of columns
load capacity and itssubjected
external to acci-
dental
axial force in theaxial loads.prior
element Important, however,
to the extreme is the
event relationship
(column of column
utilization ratio) inload capacity
relation to and its
external axial force in the element prior to the extreme event (column utilization ratio) in
its ductility.
relation
In this to its
paper, theductility.
importance of the vertical ductility of columns is discussed in the
context of theirIn this paper,under
stability the importance
accidental of the vertical
loads. Based ductility of columns
on the concept of the is ultimate
discussed in the
bearing context
capacityofoftheir stability
columns forunder
extreme accidental
situations,loads. Basedvertical
a novel on the concept
ductilityofmeasure
the ultimate
of bear-
ing capacitycolumns
axially compressed of columns for extreme
is proposed. situations,
A series a novel vertical
of experiments ductility
was carried outmeasure
on axiallyof axially
compressedcompressed columns is proposed. A series of experiments was carried out on axially com-
slender (L/D = 15, where for circular cross-section D is its diameter and for
pressed
rectangular one Dslender (L/D of
is a length = 15,
its where
shorterfor circular
side, L—column length) D
cross-section andis its diameterslender
moderately and for rectan-
(L/D = gular one D is a length of its shorter side, L—column length) and moderately
9.4) steel and composite columns. In total, 21 composite and 5 slender (L/D
steel elements
= 9.4)Asteel
were tested. and composite
comparative columns.
analysis of the In total, 21
vertical composite
ductility and
of the 5 steel
tested elements were
specimens was tested.
conducted.A comparative
The variableanalysis
parameters of the
were vertical ductility
structural of the tested
solution—steel specimens concrete-
or composite was conducted.
The variable parameters were structural solution—steel or composite concrete-filled
filled steel tubes (CFST)—column slenderness, wall thickness, concrete strength, and grade steel
of structural steel. The experimentally obtained equilibrium paths
tubes (CFST)—column slenderness, wall thickness, concrete strength, and gradeand failure modes wereof struc-
analyzed for comparative
tural purposes.
steel. The experimentally obtained equilibrium paths and failure modes were ana-
lyzed for comparative purposes.
2. Ductility of CFST Columns—State-Of-The-Art
Figure 3a presents
2. Ductility the equilibrium
of CFST paths of steel, concrete, and CFST stub columns.
Columns—State-of-the-Art
It can be seen that the ductility of the concrete-filled steel tube is significantly enhanced
Figure 3a presents the equilibrium paths of steel, concrete, and CFST stub columns.
when compared to those of the steel tube and the concrete alone. The combination of two
It can be seen that the ductility of the concrete-filled steel tube is significantly enhanced
materials in CFST columns, steel and concrete, takes advantage of the pros and eliminates
when compared to those of the steel tube and the concrete alone. The combination of two
the cons of both of them. The main factors influencing the increase in ductility of composite
materials in CFST columns, steel and concrete, takes advantage of the pros and eliminates
stub columns are the concrete core counteracting the local buckling of the walls of the steel
the consthe
section towards of both
insideofand
them.theThe main factors
confinement influencing
of the the increase
concrete core in ductility
by the steel jacket. Theof compo-
local buckling of the steel wall towards the center causes the buckled part not to carry of the
site stub columns are the concrete core counteracting the local buckling of the walls
steel
load in the section towards
post-peak theequilibrium
part of the inside and path,
the confinement
which causesofa the concretedecrease
significant core byinthe steel
jacket. The local buckling of the steel wall towards the center
the ductility of the entire element [30]. The increase in ductility of concrete causedcauses the buckled
by itspart not
to carry load in the post-peak part of the equilibrium path, which causes
confinement results in overall improvement of the ductility of the CFST stub column [31,32]. a significant de-
crease in the ductility of the entire element [30]. The increase in
In [33,34], the load capacities of 12 circular and 7 rectangular CFST columns were ana- ductility of concrete
lyzed. Itcaused by its
was found confinement
that for columns results in overall
with ratios L/Dimprovement
< 11, their loadofcapacities
the ductility
wereofgreater
the CFST stub
columnin[31,32].
than expected calculations. In more slender elements, the global buckling determined
In [33,34],
the load bearing theofload
capacities the capacities
elements. Theof 12 circular
reasons andtied
were 7 rectangular CFST
to the triaxial stresscolumns
state were
analyzed. It was found that for columns with ratios L/D
in the concrete resulting from its confinement by the steel jacket, which caused an increase were
< 11, their load capacities
in the load-bearing capacities of the elements. After applying sufficient longitudinal strain
erials 2022, 15, x FOR PEER REVIEW 4 of 29

greater than expected in calculations. In more slender elements, the global buckling de-
Materials 2022, 15, 2231
termined the load bearing capacities of the elements. The reasons were tied to the triaxial 4 of 26

stress state in the concrete resulting from its confinement by the steel jacket, which caused
an increase in the load-bearing capacities of the elements. After applying sufficient longi-
tudinal strain to concrete
to concrete sufficient
sufficient for for microcracks
microcracks to appear,
to appear, the the transverse
transverse dimensions
dimensions of the concrete
of the concrete
corecore increase,
increase, resulting
resulting in aninadditional
an additional confining
confining pressure.
pressure. Significant
Significant in- in the load
increases
creases in the load capacity
capacity are recorded
are recorded only in circular
only in circular sections sections
in whichinthe which
steelthe steel
jacket jacket a uniform
provides
provides a uniform
transverse transverse
pressure.pressure. Thisthe
This limits limits the deformations
deformations of the concrete
of the concrete and ensures a
and thereby
thereby ensures a significant degree of constraint, increasing both the load-bearing capac-the ductility.
significant degree of constraint, increasing both the load-bearing capacity and
ity and the The
ductility. The of
efficiency efficiency
concrete of concrete confinement
confinement in rectangular in CFST
rectangular CFST cross-
cross-sections is less than that
sections is less than that
in circular in [35].
ones circular ones [35]. InCFST
In rectangular rectangular
columns, CFST columns,confinement
the greatest the greatest of concrete
confinement of concrete
occurs occurs at
at the corners ofthe
thecorners of the
profile and in profile and of
the middle in the
the cross-section
middle of theincross-
the area defined
by the four curves shown in Figure 3b [31].
section in the area defined by the four curves shown in Figure 3b [31].

(a) (b)
Figure 3. (a) Figure
Axial compressive
3. (a) Axial behavior of CFST
compressive steel, of
behavior concrete and CFST
CFST steel, stuband
concrete columns,
CFST based on
stub columns, based
[36]; (b) effectively confined concrete in a rectangular column, based on [31].
on [36]; (b) effectively confined concrete in a rectangular column, based on [31].

In [37], test In
results of almost
[37], test results270 of CFST
almostcolumns
270 CFST with circular,
columns octagonal,
with circular,and rectan- and rectan-
octagonal,
gular cross-sections with 2 ≤ L/Dwith
gular cross-sections ≤ 9 and
2 ≤ 19 L/D ≤ D/t
≤ 9≤ and
75 were
19 ≤presented.
D/t ≤ 75 As werea result of the As a result
presented.
analyses, three types of behavior of stub CFST columns were distinguished:
of the analyses, three types of behavior of stub CFST columns were distinguished: 1—with hard- 1—
ening, 2—nearly perfectly plastic behavior, and 3—with softening.
with hardening, 2—nearly perfectly plastic behavior, and 3—with softening. All columns with cir-All columns
cular cross-sections and most
with circular of those with
cross-sections and octagonal
most of those cross-sections
with octagonal, due to the confine- due to the
cross-sections,
ment of concrete, were classified
confinement of concrete, intowerecategories
classified1 andinto2.categories
Some octagonal1 and 2. andSomeall rectan-
octagonal and all
gular columns were classified
rectangular columns as were
type 3.classified
In [13], the vertical
as type equilibrium
3. In paths ofequilibrium
[13], the vertical 14 CFST paths of
circular and1411CFST
rectangular
circularstubsand with 4 ≤ L/D ≤ 4.8
11 rectangular stubs 17 ≤ D/t
and with 4 ≤≤L/D
47 were≤ 4.8analyzed.
and 17 ≤ It was
D/t ≤ 47 were
found that the ductility
analyzed. It of
wascircular
foundcolumns, defined as
that the ductility ofμcircular
= xm/xy (see Figuredefined
columns, 2), is greater
as µ = xm /xy (see
than the ductility
Figure of 2),columns
is greaterwith thanrectangular
the ductilitycross-sections.
of columns with Forrectangular
circular stubs, the duc- For circu-
cross-sections.
tility is μ ≥ 10, and for
lar stubs, the ductility is 2µ<≥μ10,
rectangular, < 8.and
Allfor
tested circular columns
rectangular, 2 < µ < 8.exhibited
All testedtype 1 columns
circular
behavior (even with type
exhibited a relatively
1 behavior thin(even
steelwithjacket), whereasthin
a relatively columns with rectangular
steel jacket), whereas columns with
cross-sections showed the
rectangular same behavior
cross-sections showed only with
the samethick steel walls
behavior (D/t <thick
only with 20). steel
In [13], it (D/t < 20).
walls
was also found that
In [13], for circular
it was also found cross-sections
that for circular the cross-sections
confinement of theconcrete
confinement usually be-
of concrete usually
comes significant only after reaching approximately 90% of the load capacities of tested
becomes significant only after reaching approximately 90% of the load capacities of tested
elements, and the local buckling of steel profiles is equivalent to reaching their ultimate
elements, and the local buckling of steel profiles is equivalent to reaching their ultimate
load capacities.
load In [6] the behavior
capacities. In [6] the ofbehavior
circular and rectangular
of circular CFST stub columns
and rectangular CFST stub was
columns was
analyzed. Itanalyzed.
was found It that
was anfound increase
that an in increase
steel ratioincan change
steel the change
ratio can failure oftheconcrete
failure of concrete
from being from
localized
being into more orinto
localized lessmore
uniform.
or less This changeThis
uniform. corresponds to the transition
change corresponds to the transition
of the axial response of the specimen from softening into a ductile response; that is, the that is, the
of the axial response of the specimen from softening into a ductile response;
axial
axial ductility of ductility
the member of the is member
increasediswith increased
by an with by an
increase in increase
the steelin the steel contribution
contribution
ratio: =
ratio: ξ = Aafy/Acfc, where
ξ A f /A f
a y As candc , where A
Ac are the s and A
areas are the areas of the steel and concrete
c of the steel and concrete cross-sections, cross-sections,
respectively, and fy and fc are the yield strength of the steel and concrete, respectively. The
use of a thicker steel casing caused greater confinement of the concrete, which resulted
in crushing of the concrete over a greater length of the stub column, which increased its
ductility.
 Materials 2022, 15, x FOR PEER REVIEW 5 of 29

Materials 2022, 15, 2231 5 of 26

respectively, and fy and fc are the yield strength of the steel and concrete, respectively. The
use of a thicker steel casing caused greater confinement of the concrete, which resulted in
crushing of the concrete over a greater length of the stub column, which increased its duc-
The use of a higher class of concrete increases the load capacities of CFST stub columns
tility.
but decreases
The use their
of aductility.
higher class In [7], it was found
of concrete increases thatthethe usecapacities
load of high-strength
of CFST stub concrete
col-
173 ÷but
(fc =umns 184decreases
MPa) in their circular
ductility. In [7], it was found that the use of high-strengthincon-
CFST columns resulted in a significant decrease the
vertical ductility of the tested elements. This phenomenon was attributed
crete (fc = 173 ÷ 184 MPa) in circular CFST columns resulted in a significant decrease in the to the inadequate
ability of steel
vertical casingof
ductility (fythe
= 420 MPa)
tested to confine
elements. Thishigh strength concrete.
phenomenon It was stated
was attributed that in
to the inade-
orderquate ability of steel casing (fy = 420 MPa) to confine high strength concrete. It was stated
to increase the efficiency of confinement, thicker steel walls should be used and the
steelthat in order toratio
contribution increase
in thethe efficiency ofshould
cross-section confinement, thicker
be at least 30%.steel walls
Similar should be used
conclusions were
drawnandfrom
the steel contribution
the analyses ratioout
carried in the cross-section
in [37], where itshould be at least 30%.
was additionally Similar conclu-
recommended that
sions were value
the maximum drawnbe from
D/tthe analyses carried out in [37], where it was additionally recom-
= 30.
mended
In [38] thethatductility
the maximum value be CFST
of rectangular D/t = 30.
stub columns was analyzed. Figure 4a shows
In [38]
the ductility the ductility
index of rectangular
as a function of the steel CFSTcontribution
stub columnsratio,was analyzed.
and Figure Figure 4a shows
4b shows DI
the ductility
depending on the index as a ratio
aspect functionof theof the steel contribution
rectangular section βratio, andDuctility
= h/b. Figure 4bindex showswas DI
depending
defined based on on the aspect
strains DIratio
= ε85%of /εtheu ,rectangular
where εu issection β = h/b.
the strain at Ductility
the ultimate indexload,
was andde-
fined based on strains DI = ε /ε , where ε is the strain at the
ε85% is the strain when the load falls to 85% of the ultimate load. The increase in the steel
85% u u ultimate load, and ε 85% is
the strain when the load falls to 85% of the ultimate load. The
contribution ratio had a positive effect on the ductility of rectangular CFST stub columns. increase in the steel contri-
The bution
inverseratio had a positive
relationship can beeffect
noticed on thewhen ductility
analyzing of rectangular
the influenceCFSTof stub columns. The
the cross-sectional
shapeinverse
factorrelationship
β. Columnscan withbesquare
noticedcross-sections
when analyzing the influence
exhibited of the
the greatest cross-sectional
ductility, and the
shape factor
elongation of one β. of
Columns with square
the cross-section cross-sections
sided has a negativeexhibited
effecttheongreatest ductility,
its ductility. and
Similar
the elongation
observations wereofpresented
one of thein cross-section
[39]. sided has a negative effect on its ductility. Similar
observations were presented in [39].

(a) (b)
Figure 4. Ductility index DI versus: (a) steel contribution ratio ξ with β = 1.3; (b) the ratio of the
Figure 4. Ductility index DI versus: (a) steel contribution ratio ξ with β = 1.3; (b) the ratio of the
cross-sectional dimensions β, with ξ = 0.7, based on [38].
cross-sectional dimensions β, with ξ = 0.7, based on [38].
The influence of the aspect ratio of the cross-section on the ductility of a CFST stub
The influence of the aspect ratio of the cross-section on the ductility of a CFST stub
column was also noticed in [20], where elliptical cross-sections were analyzed. As a result
column was also noticed in [20], where elliptical cross-sections were analyzed. As a result
of the parametric analysis, it was found that the ductility of the columns decreases with
of the parametric analysis, it was found that the ductility of the columns decreases with
the increase in the eccentricity of the ellipse. It is worth noting that the standard [40] states
the increase in the eccentricity of the ellipse. It is worth noting that the standard [40] states
that the height-to-width ratio of the composite cross-section should be between 0.2 and
that the height-to-width ratio of the composite cross-section should be between 0.2 and 5.0,
5.0, which excludes the use of very elongated sections. In [20] it was confirmed that the
which excludes
increase the concrete
in the use of very elongated
strength classsections.
resulted In
in [20] it was confirmed
a decrease that the
in the ductility of increase
the test
in the concrete
elements. strength class resulted in a decrease in the ductility of the test elements.
Figure 5 presents
Figure typical
5 presents locallocal
typical failure modes
failure of axially
modes compressed
of axially stub columns:
compressed steel,
stub columns:
concrete, and CFST. The steel element (Figure 5a) undergoes local inward
steel, concrete, and CFST. The steel element (Figure 5a) undergoes local inward and out- and outward
buckling
ward of the wall,
buckling of whereas
the wall,in CFST stub
whereas columns
in CFST stub that exhibit
columns thatsymmetrical elephant ele-
exhibit symmetrical foot
phant
failure foot failure
(Figure 5c) only(Figure 5c) onlybuckling
the outward the outward buckling
of the of the steel
steel occurs, alongoccurs, along
with the with the
crushing of
crushing
concrete. of concrete.
In CFST In CFSTnon-symmetrical
stub columns, stub columns, non-symmetrical
shear and sliding shear and(Figure
failure sliding5d)
failure
may
also(Figure 5d) maytoalso
occur similar the occur
typical similar to mode
failure the typical
of thefailure
plain mode
concreteof the plain
stub concrete
(Figure 5b). stub
The
difference is that the steel casing prevents the brittle shear failure usual for concrete stub
columns and allows solid wedge movement along a shear frictional surface, increasing the
ductility of CFST elements.
Materials 2022, 15, x FOR PEER REVIEW 6 of 29

(Figure 5b). The difference is that the steel casing prevents the brittle shear failure usual
Materials 2022, 15, 2231 for concrete stub columns and allows solid wedge movement along a shear frictional6sur- of 26
face, increasing the ductility of CFST elements.

(a) (b) (c) (d)

Figure 5. Local failure modes of axially compressed stub columns: (a) steel—local buckling, (b) con-
Figure 5. Local failure modes of axially compressed stub columns: (a) steel—local buckling, (b) con-
crete without reinforcement—shear (c) CFST—elephant foot, (d) CFST—shear and sliding, based on
crete without reinforcement—shear (c) CFST—elephant foot, (d) CFST—shear and sliding, based
[36,41].
on [36,41].
The above considerations concern mainly the CFST stub columns—eliminating slen-
The above considerations concern mainly the CFST stub columns—eliminating slen-
derness
dernessfrom
fromthe theanalysis
analysisisishelpful
helpfulto toidentify
identifythe
thebehavior
behaviorof ofsteel
steeland
andconcrete
concreteand andthethe
mutual
mutual interaction between the two materials at the cross-sectional level. In [42]ititwas
interaction between the two materials at the cross-sectional level. In [42] was
found
foundthatthatslender
slenderCFST
CFSTcolumns
columnswith withrelative
relativeslenderness
slendernesslevels levelsgreater
greaterthanthan 𝜆𝜆λ == 0.2
0.2
exhibited insignificant confinement of their concrete cores by their steel
exhibited insignificant confinement of their concrete cores by their steel jackets. However, jackets. However,
such
suchelements
elementsare arerarely
rarelyused
usedin inpractice,
practice,since
sincethis
thistype
typeof ofcolumn
columnisisdesigned
designedto toreduce
reduce
cross-sectional dimensions. Due to global imperfections in the more
cross-sectional dimensions. Due to global imperfections in the more slender CFST columns slender CFST col-
umns often used in practice, the local failure modes presented in Figure
often used in practice, the local failure modes presented in Figure 5 are not likely to occur. 5 are not likely to
occur.Figure 6a,b shows the typical failure modes of very slender steel and composite CFST
Figure
circular 6a,b shows
columns, the typical
as presented infailure
[42]. Inmodes of very
both cases, theslender
ultimate steel and composite
bearing capacity CFST
of the
circular
columnscolumns,
is relatedastopresented
the global in [42]. In However,
buckling. both cases, inthe
caseultimate
of the steelbearing
column,capacity
in theof the
center
columns is related to the global buckling. However, in case of the
of the column the additional local buckling of the more compressed wall towards the inside steel column, in the
center
of the of the columnoccurs.
cross-section the additional
In very local
slenderbuckling of the more
CFST columns compressed
(Figure 6b), this wall towards
phenomenon
Materials 2022, 15, x FOR PEER REVIEW 7 of 29
the
doesinside of thebecause
not occur cross-section occurs.core
the concrete In very slender
prevents theCFST
inward columns (Figureof6b),
deformation this
steel. phe-
Similar
nomenon
conclusions does notdrawn
were occur in
because
the study the of
concrete
slendercore prevents
elliptical the inward
columns [43]. deformation of
steel. Similar conclusions were drawn in the study of slender elliptical columns [43].
In [27] the behavior of slender and moderately slender RCFST (round-ended con-
crete-filled steel tube) columns was investigated. The shortest sides of a rectangular sec-
tion were converted into half circles, creating ovular, rounded ends. It was found that in
the case of moderately slender columns, failure occurred as combination of global and
local failure, as shown in (Figure 6b). The more compressed side of the cross-section un-
derwent local buckling, but the concrete filling prevented its deformation towards the in-
side. For very slender elements, the local buckling did not occur, and the failure was dic-
tated by an elastic flexural global buckling, similar to that shown in Figure 6c. The ductil-
ity analysis showed that with the increases in B/t and B/D, where B and D are the larger
and smaller cross-section dimensions, respectively, the ductility of the elements de-
creased. The decrease in ductility also occurred with an increase in the class of concrete
used.
In the literature, there is lack of investigations considering the ductility of rectangular
CFST columns with moderate slenderness that are most used in practice. Either stub or
very slender CFST columns were tested, and their behavior is described in the literature.
Therefore, it was necessary to carry out the investigation on the vertical ductility of axially
compressed, moderately slender, rectangular CFST columns.

(a) (b) (c)

Figure 6. Typical failure modes of slender and moderately slender circular columns: (a) steel col-
Figure 6. Typical failure modes of slender and moderately slender circular columns: (a) steel column,
umn, global + local buckling; (b) CFST column, global and local buckling; (c) CFST column, only
global
elastic + buckling,
global local buckling; (b)[27,44].
based on CFST column, global and local buckling; (c) CFST column, only elastic
global buckling, based on [27,44].
3. Proposal for the Vertical Ductility Assessment of Columns
Figure 7 shows two equilibrium paths of columns with the same ultimate bearing
capacity that exhibit sufficient (Figure 7a) and insufficient (Figure 7b) ductility under dy-
namic axial loads. The same scale of axes was kept for Figure 7a,b; therefore, it can be seen
Materials 2022, 15, 2231 7 of 26

In [27] the behavior of slender and moderately slender RCFST (round-ended concrete-
filled steel tube) columns was investigated. The shortest sides of a rectangular section
were converted into half circles, creating ovular, rounded ends. It was found that in the
case of moderately slender columns, failure occurred as combination of global and local
failure, as shown in (Figure 6b). The more compressed side of the cross-section underwent
local buckling, but the concrete filling prevented its deformation towards the inside. For
very slender elements, the local buckling did not occur, and the failure was dictated by an
elastic flexural global buckling, similar to that shown in Figure 6c. The ductility analysis
showed that with the increases in B/t and B/D, where B and D are the larger and smaller
cross-section dimensions, respectively, the ductility of the elements decreased. The decrease
in ductility also occurred with an increase in the class of concrete used.
In the literature, there is lack of investigations considering the ductility of rectangular
CFST columns with moderate slenderness that are most used in practice. Either stub or
very slender CFST columns were tested, and their behavior is described in the literature.
Therefore, it was necessary to carry out the investigation on the vertical ductility of axially
compressed, moderately slender, rectangular CFST columns.

3. Proposal for the Vertical Ductility Assessment of Columns

Figure 7 shows two equilibrium paths of columns with the same ultimate bearing
capacity that exhibit sufficient (Figure 7a) and insufficient (Figure 7b) ductility under
dynamic axial loads. The same scale of axes was kept for Figure 7a,b; therefore, it can be
seen that the ductility of the column whose equilibrium path is depicted in Figure 7a is
greater in relation to Figure 7b. If the element is sufficiently ductile, after the imposed
loads are withdrawn, there is a new stable equilibrium point (marked as X in Figure 7a).
In this case the column maintains its residual load bearing capacity and does not lose
its stability after the extreme event. On the other hand, if the ductility of the column is
insufficient (Figure 7b), then after the accidental loading the next point of equilibrium does
not exist and the column collapses. On analyzing the data in Figure 7, it can be concluded
that the preloading force prior to the imposed loading is not less important. The smaller
Materials 2022, 15, x FOR PEER REVIEW 8 of 29
the utilization ratio of the column, the lower the vertical ductility demand because a new
equilibrium in the post-peak may occur for a column with less residual strength.

(a) (b)
Figure 7.
Figure 7. Equilibrium
Equilibriumpath
pathfor
fora acolumn
columnthat exhibits:
that (a)(a)
exhibits: sufficient andand
sufficient (b) (b)
insufficient ductility.
insufficient 1—
ductility.
displacement and force in the column just before the excitation; 2—maximum instantaneous
1—displacement and force in the column just before the excitation; 2—maximum instantaneous dis-
placement; 3—oscillation around the equilibrium position; 4—permanent final displacement.
displacement; 3—oscillation around the equilibrium position; 4—permanent final displacement.

AnFigure
In energetic
7, themeasure
ultimateofaxial
the bearing μω of axially
ductilitycapacity of thecompressed columns
column is denoted as based on the
NR , whereas
displacements of the column top in the pre- and post-peak periods, x 0 and x ω ,
N0 = ωNR is an equivalent axial force in the column prior to the extreme event. A reductionrespectively,
is proposedω(see
parameter Figure
is used to 8). The values
account of these
for the displacements
influences depend
of various on the shape
combinations of the
of actions’
equilibrium path and on the value of the ω coefficient. The ductility measures based only
on the quotients of displacements do not consider the full shape of the equilibrium path,
but only the chosen points. The proposed energy measure has the advantage of consider-
ing the entire shape of the equilibrium path, both in the pre- and post-peak up to the dis-
placement xω. E0 is the energy (mostly elastic) accumulated in the column at N0 load, and
Materials 2022, 15, 2231 8 of 26

coefficients and material safety factors in the permanent and accidental design situation,
the reserve factor, and the strain rate behavior of the steel and concrete. Assuming that the
utilization of the element in the permanent design situation is equal to 100% and that the
typical safety factors for concrete and steel are equal to 1.4 and 1.0, respectively, an upper
limit of this coefficient for composite steel-concrete columns is equal to approximately 0.8.
Increasing the column’s reserve factor in permanent design situation results in a decrease
in the ω coefficient.
An energetic measure of the ductility µω of axially compressed columns based on the
displacements of the column top in the pre- and post-peak periods, x0 and xω , respectively,
is proposed (see Figure 8). The values of these displacements depend on the shape of
the equilibrium path and on the value of the ω coefficient. The ductility measures based
only on the quotients of displacements do not consider the full shape of the equilibrium
path, but only the chosen points. The proposed energy measure has the advantage of
considering the entire shape of the equilibrium path, both in the pre- and post-peak up to
the displacement xω . E0 is the energy (mostly elastic) accumulated in the column at N0 load,
and Eω is the energy in the column (accumulated and dissipated) when the force drops
to N0 in the post-peak part of the equilibrium path. In fact, the energy Eω is the physical
measure of the vertical ductility of a given column, and the energy E0 is introduced for
normalization purposes. The advantage of the proposed measure of ductility is also the
fact that for a given equilibrium path, E0 and Eω can be easily determined, since both x0
and xω are explicitly defined as points of intersection of the equilibrium path with the
ordinate N0 . This avoids imprecise definition of the yield and failure points used in the
conventional vertical ductility measures adopted from horizontal ductility factors widely
used in seismic analysis [29]. Those have a physical interpretation for horizontal ductility,
but not necessarily for the vertical one. The failure point in the proposed vertical measure
Materials 2022, 15, x FOR PEER REVIEW
is defined as the point on the equilibrium path for which for a given force N0 , failure 9ofofthe
29
column will occur.

Figure 8.
Figure 8. Proposed
Proposed measure
measure of
of vertical
vertical ductility
ductility based
based on
on the
the energy.
energy.

4. Experimental
The number of Research—Methodology
publications on modelling of the behavior of composite CFST columns con-
tinuesThe experimental
to rapidly research
rise. Fiber elementwas conducted
models on 26
[45,46] and FEMcold-formed
simulationssteel rectangular hol-
[3,6,9,11,20,27,47–51]
lowproven
are sectiontocolumns
be capable made of S235JRpredicting
of accurately and S355JR-grade
the shapessteel. The length
of equilibrium of each
paths element
of composite
was 750 mm. The analysis included 3 types of composite steel–concrete column. For
CFST columns, including their post-peak behavior. Recently, artificial neural network com-
(ANN)
parative studies, steel columns that had not been filled with concrete were also tested.the
models [52] and machine learning algorithms [53] were employed to accurately predict A,
B, and Dstrength
ultimate describeofcolumns made of
CFST columns. RHS100
Further × 50 × 5,ANN-based
improved RHS100 × 50 × 3, and
models RHS120
were then used× 80to×
3 profiles,
predict andrespectively. S1 and
plot the complete S2load-shortening
axial indicate the steel grades
curves S235 and S355,
of concentrically respectively,
loaded rectangu-
andand
lar C1 circular
and C2 indicate the concrete
CFST columns classes C20/25
that reasonably and the
matched C50/60, respectively.
experimental In the
results exper-
[54]. This
imental analysis, the variable parameters were the type of column (steel, composite), slen-
approach will hopefully allow determining the vertical ductility of CFST columns without
derness, wall
performing thickness,
costly and time concrete strength,
consuming and steel
experimental grade.
and FEM Standardized
analyses. steel and con-
crete samples were tested to obtain the material properties. A detailed description of the
material and geometrical properties is provided in Table 1 and depicted in Figure 9. Table
1 columns contain, respectively, the notation of the specimen; steel cross-section; steel
grade; yield strength of steel; concrete strength; quotients D/t and L/D, where D—height
Materials 2022, 15, 2231 9 of 26

4. Experimental Research—Methodology
The experimental research was conducted on 26 cold-formed steel rectangular hol-
low section columns made of S235JR and S355JR-grade steel. The length of each ele-
ment was 750 mm. The analysis included 3 types of composite steel–concrete column.
For comparative studies, steel columns that had not been filled with concrete were also
tested. A, B, and D describe columns made of RHS100 × 50 × 5, RHS100 × 50 × 3, and
RHS120 × 80 × 3 profiles, respectively. S1 and S2 indicate the steel grades S235 and S355,
respectively, and C1 and C2 indicate the concrete classes C20/25 and C50/60, respectively.
In the experimental analysis, the variable parameters were the type of column (steel, com-
posite), slenderness, wall thickness, concrete strength, and steel grade. Standardized steel
and concrete samples were tested to obtain the material properties. A detailed description
of the material and geometrical properties is provided in Table 1 and depicted in Figure 9.
Table 1 columns contain, respectively, the notation of the specimen; steel cross-section; steel
grade; yield strength of steel; concrete strength; quotients D/t and L/D, where D—height
of the steel profile, t—wall thickness, L—total length of the column; λ—relative slenderness;
ξ—steel contribution ratio; the number of elements tested in each series. The columns from
series D were of moderate slenderness, and elements from series A and B were considered
slender in relation to columns from series D.

Table 1. Material and geometric properties of the investigated columns—experimental research.

Steel Steel Yield Strength Concrete Strength

Series D/t/L/D λ ξ Quantity
Cross-Section Grade fy (MPa) fc (MPa)
DS1 S235 317.0 - 0.29 - 1
DS1C1 S235 317.0 37.0 0.32 1.16 3
RHS
DS1C2 S235 317.0 47.9 26.7/9.4 0.34 0.90 3
120 × 80 × 3
DS2 S355 414.0 - 0.33 - 1
DS2C1 S355 414.0 37.0 0.35 1.51 2
AS1 S235 408.3 - 0.54 - 1
AS1C1 S235 408.3 37.0 0.56 4.12 3
RHS
AS1C2 S235 408.3 47.9 10.0/15 0.57 3.18 3
100 × 50 × 5
AS2 S355 426.2 - 0.55 - 1
AS2C1 S355 426.2 37.0 0.57 4.30 2
BS1 S235 357.4 - 0.48 - 1
RHS
BS1C1 S235 357.4 37.0 16.7/15 0.53 1.97 3
100 × 50 × 3
BS1C2 S235 357.4 47.9 0.54 1.52 2

The columns were compressed in a hydraulic press. The applied force vs. column
shortening was measured, along with the vertical strains of the steel profiles using strain
gauges placed mid-length. In addition, the digital image correlation method (DIC) [45]
was used to measure the displacement in composite members. A random in position
but uniform in size speckle pattern was applied to the surface of each CFST sample (see
Figure 10a). In order to maximize the contrast of images white primer and black car paint
were used. In order to ensure proper adhesion of the paint, the samples were sandblasted
prior to the application of the coating. The displacements were measured using two
cameras taking pictures of a certain rectangular area of each tested composite column, and
the coordinates of selected points located in the measuring area were determined. The
measured coordinates, which were related to the coordinates of the same points measured
in the first image of the unloaded sample, were then used to calculate the components of
the displacement and strain field. The calibration was performed in accordance with [55]
using a calibration panel with two scale bars of MV 175 × 140. The calibration deviation
was equal to 0.019 pixels which corresponds to an accuracy of measurements equal to
approximately 5 µm.
Materials 2022, 15, 2231 10 of 26
Materials 2022, 15, x FOR PEER REVIEW 10 of 29

Figure 9. Design schemes and descriptions of the measurements performed on the investigated
Materials 2022, 15, x FOR PEER REVIEW 11 of 29 col-
umns.9. Design schemes and descriptions of the measurements performed on the investigated
Figure
columns.
The columns were compressed in a hydraulic press. The applied force vs. column
shortening was measured, along with the vertical strains of the steel profiles using strain
gauges placed mid-length. In addition, the digital image correlation method (DIC) [45]
was used to measure the displacement in composite members. A random in position but
uniform in size speckle pattern was applied to the surface of each CFST sample
(see Figure 10a). In order to maximize the contrast of images white primer and black car
paint were used. In order to ensure proper adhesion of the paint, the samples were sand-
blasted prior to the application of the coating. The displacements were measured using
two cameras taking pictures of a certain rectangular area of each tested composite column,
and the coordinates of selected points located in the measuring area were determined. The
measured coordinates, which were related to the coordinates of the same points measured
in the first image of the unloaded sample, were then used to calculate the components of
the displacement and strain field. The calibration was performed in accordance with [55]
using a calibration panel with two scale bars of MV 175 × 140. The calibration deviation
was equal to 0.019 pixels which corresponds to an accuracy of measurements equal to
approximately 5 μm.

(a) (b)
Figure 10. (a) Speckle pattern application, (b) experimental setup.
Figure 10. (a) Speckle pattern application, (b) experimental setup.
In order to obtain equilibrium paths in the post-peak parts of the equilibrium paths
of columns, a displacement control method was used in the analysis. For technical rea-
sons, due to expected exceptionally large deformations of investigated elements and
therefore a possibility of uncontrolled ejection of columns from the hydraulic press after
Materials 2022, 15, 2231 11 of 26

In order to obtain equilibrium paths in the post-peak parts of the equilibrium paths of
columns, a displacement control method was used in the analysis. For technical reasons,
due to expected exceptionally large deformations of investigated elements and therefore a
possibility of uncontrolled ejection of columns from the hydraulic press after buckling, the
knife edged boundary conditions simulating pinned-pinned support boundary conditions
were not used. The test elements were each placed on a rigid, non-deformable steel plate
and compressed by a spherical traverse, embedded in a spherical seat; hence, columns’ ends
were partially fixed due to friction. The experimental setup was as shown in Figure 10b.

5. Results of the Experimental Research and Discussion

5.1. Failure Modes
The typical failure modes of the analyzed columns are shown in Figure 11. Three types
were observed: a combination of global and local buckling (from this point called only
the global buckling), local shear and sliding, and elephant foot (splitting) failure. Global
buckling occurred in all slender columns from series A (Figure 11a) and in one specimen,
BS1C2_1, from series B (Figure 11b). In all remaining slender columns from series B, the
failure occurred due to the local shear and sliding of the concrete core (Figure 11c). In only
one moderately slender specimen, DS2C1_1, was elephant foot (splitting) failure observed
(Figure 11d). The failure of all remaining specimens from series D occurred due to shear
and sliding of inner concrete (Figure 11e).
In moderately slender columns from series D, all failure modes were of the local type,
and in slender columns with 5 mm wall thickness (series A), global type. This was due to
the relative slenderness of the tested elements, which in the case of columns from series
D was on average λ = 0.33, and in the case of columns from series A, λ = 0.57. It is worth
noting that in the case of columns from series B with similar slenderness to elements from
series A (λ = 0.53), but with thinner walls (t = 3 mm), the failure mode of most test elements
is local. Only in one specimen BS1C2_1 is the failure mode global.
The steel contribution ratio ξ is used in the literature to estimate the effectiveness of
concrete confinement by the steel jacket in a composite column [38,49]. The specimens from
series D, in which mostly the local shear and sliding failure occurred, exhibited the smallest
values of this coefficient, 0.90 ≤ ξ ≤ 1.51, (see Table 1). The symmetric “elephant foot”
failure type occurred only in specimen DS2C1_1, which was characterized by the highest
value of steel contribution ratio among all elements from series D (ξ = 1.51). This suggests
the influence of concrete confinement on the types of local failure modes of moderately
slender composite columns. The greater this phenomenon, the greater the probability of a
symmetrical failure mode such as “elephant’s foot.” On the other hand, in another column
of this series, DS2C1_2, the “elephant’s foot” failure did not occur, which suggests the
high complexity of phenomena occurring in rectangular CFST columns and their random
nature.
In slender elements from series A (t = 5 mm) in which the steel contribution ratio
had of the highest values, 3.18 ≤ ξ ≤ 4.30, only global buckling was observed. As a result
of significant confinement of concrete by the steel jacket, the concrete core was neither
crushed (“elephant foot” failure) nor sheared (“shear and slip” failure). In these elements,
the local load-bearing capacity and ductility of concrete increased by its confinement are
high enough that they do not determine the load-bearing capacity of the entire element.
Therefore, global sway buckling occurs, and in the middle of the column a plastic hinge
is formed. Only after the buckling of the entire element, in the compression zone of the
middle section, does the concrete undergo one-sided crushing, and the compressed part of
the steel jacket is subject to local buckling (see Figure 11a).
In slender elements from series B (t = 3 mm), the confinement of concrete (1.52 ≤ ξ ≤ 1.97)
was not high enough to trigger the global buckling failure mode before the local shear of
concrete. This was the case only in one specimen marked BS1C2_1 (see Figure 11b), in which
the occurrence of global buckling before local concrete core shear could have been caused by a
Materials 2022, 15, 2231 12 of 26

Materials 2022, 15, x FOR PEER REVIEW

larger arc imperfection of the element or additional imperfections caused by the gusset12plates
of 29
welded not ideally perpendicular to the axis of the column.

(a) (b) (c) (d) (e)

Figure 11. Typical failure modes of composite columns observed during the experiment: (a) global
Figure 11. Typical failure modes of composite columns observed during the experiment: (a) global
buckling of slender columns from series A; (b) global buckling of single specimen BS1C2_1; (c) local
buckling
shear andof slender
sliding columns
failure from series
of slender A; (b)
columns global
from buckling
series of single
B (except specimen
BS1C2_1); BS1C2_1;
(d) elephant foot(c)(split-
local
ting)
shearfailure of single
and sliding specimen
failure DS2C1_1;
of slender (e) local
columns fromshear
seriesand sliding BS1C2_1);
B (except failure of moderate slender
(d) elephant foot
columns
(splitting)from series
failure D (except
of single DS2C1_1).
specimen DS2C1_1; (e) local shear and sliding failure of moderate slender
columns from series D (except DS2C1_1).
In moderately slender columns from series D, all failure modes were of the local type,
5.2. Qualitative Analysis of Equilibrium Paths
and in slender columns with 5 mm wall thickness (series A), global type. This was due to
the relative slenderness
Experimentally of the equilibrium
obtained tested elements, which
paths of theinanalyzed
the case of columns
columns arefrom series
presented
in Figures 12 and 13, in which the vertical axes have been normalized with respect
D was on average 𝜆𝜆 = 0.33, and in the case of columns from series A, 𝜆𝜆 = 0.57. It is worth to the
ultimate bearing capacity of each column.
noting that in the case of columns from series B with similar slenderness to elements from
Regardless of the slenderness of the columns, the class of concrete, or the strength of
series A (𝜆𝜆 = 0.53), but with thinner walls (t = 3 mm), the failure mode of most test elements
steel, in all test series, the rate of change of the force in the post-peak part of the equilibrium
is local. Only in one specimen BS1C2_1 is the failure mode global.
paths is fairly consistent in composite columns in relation to their steel counterparts.
The steel contribution ratio ξ is used in the literature to estimate the effectiveness of
Indeed, normalized equilibrium paths of composite columns were always above the paths
concrete confinement by the steel jacket in a composite column [38,49]. The specimens
of their steel counterparts in terms of strength, and this difference increased with increasing
from series D, in which mostly the local shear and sliding failure occurred, exhibited the
displacement. This was due to the concrete core counteracting the local buckling of the
smallest values of this coefficient, 0.90 ≤ ξ ≤ 1.51, (see Table 1). The symmetric “elephant
walls of the steel section towards the inside. In steel columns, local buckling causes the
foot” failure type occurred only in specimen DS2C1_1, which was characterized by the
buckled part of the cross-section to effectively stop carrying the loads. Consequently,
highest value of steel contribution ratio among all elements from series D (ξ = 1.51). This
this reduces the ductility of the entire column. When the concrete core counteracts the
suggests the influence of concrete confinement on the types of local failure modes of mod-
erately slender composite columns. The greater this phenomenon, the greater the proba-
bility of a symmetrical failure mode such as “elephant’s foot.” On the other hand, in an-
other column of this series, DS2C1_2, the “elephant’s foot” failure did not occur, which
Materials 2022, 15, 2231 13 of 26

deformation of the steel jacket towards the center, the deformability of such a cross-section
increases, as the greater part of the steel section is still able to bear loads. Moreover, the steel
Materials 2022, 15, x FOR PEER REVIEW
jacket provides concrete confinement, which increases the ductility of the inner concrete, 14 of 29
and therefore, the ductility of the entire element.

(a) (b)

(c) (d)

(e)
Figure 12. Normalized equilibrium paths of slender elements from series: (a) AS1C1; (b) AS1C2; (c)
Figure 12. Normalized equilibrium paths of slender elements from series: (a) AS1C1; (b) AS1C2;
AS2C1; (d) BS1C1; (e) BS1C2.
(c) AS2C1; (d) BS1C1; (e) BS1C2.
Materials 2022, 15, 2231 14 of 26
Materials 2022, 15, x FOR PEER REVIEW 15 of 29

(a) (b)

(c)
Figure 13. Normalized
Figure 13. Normalized equilibrium
equilibrium paths
paths of
of moderately
moderately slender
slender elements
elements from
from series:
series: (a)
(a) DS1C1,
DS1C1,
(b) DS1C2, (c) DS2C1.
(b) DS1C2, (c) DS2C1.

A comparison
Figure 12e depictsof normalized
the normalized equilibrium paths paths
equilibrium of steel ofcolumns from allfrom
two specimens series can
series
be seen in Figure 14a. By analyzing the presented data, it can
BS1C2 whose failure modes were different. Specimen BS1C2_1 exhibited global bucklingbe concluded that the influ-
ence of (see
failure the steel
Figure grade
11b),(S235 or S355)
whereas on the ductility
in specimen BS1C2_2, of steel
localcolumns
shear and was insignificant
sliding failure
for both 3(see
occurred mmFigure
(series11c).D) and
The5implications
mm wall thickness (series
of this can A), regardless
be seen in Figure 12e.of their slender-
The ductility
ness. It can also
of specimen be seen
BS1C2_2, that the columns
considering made greater
displacements of thicker than plates
15 mm, exhibited
is muchsignificantly
larger than
greater ductility.
the ductility Moreover,
of BS1C2_1. thecolumns
In all equilibrium
which paths of columns
exhibited from mode,
local failure series A as showed
a result
greater flattening after reaching their limit load capacities.
of additional shear caused by friction and jamming of the crushed concrete within Before reaching the vertical
the
displacements
steel jacket, thereequal
wasto approximately
a cyclic increase and 8 mm, the presented
decrease relationships
in force. Elements whereresemble
this type anof
elastic-ideally plastic response
deformation occurred exhibitedby the structures.
significantly The horizontal
smaller equilibrium paths, and therefore
displacements. the
Therefore,
ductility of columns
after reaching from series
the ultimate bearingB and D, areof
capacity similar.
the weakest cross-section of a member, in
The influence
subsequent of the wall
cross-sections, thickness
further of the steel
deformation mayjacket
occur,on incorporating
the behavior ofmore slender com-
material
posite columns
along the lengthcan be seen
of the columnin Figure
in the14b, which depicts
dissipation the normalized equilibrium paths
of energy.
of elements from series
A comparison AS1C1 and
of normalized BS1C1. Itpaths
equilibrium can be of noticed that up
steel columns to the
from vertical
all series candis-
be
placements
seen in Figure equal
14a.to Byapproximately 15 mm, the
analyzing the presented equilibrium
data, paths of the
it can be concluded columns
that made
the influence
from
of theplates 3 mm(S235
steel grade thickor lieS355)
below onthe
thepaths for columns
ductility made ofwas
of steel columns thicker steel sheets,
insignificant and
for both
therefore,
3 mm (series exhibit
D) and smaller
5 mmductility. After reaching
wall thickness (series A),approximately
regardless of their 10–15 mm verticalIt dis-
slenderness. can
placement,
also be seenthethatequilibrium
the columnspaths madeofofcolumns from series
thicker plates B start
exhibited to increase,greater
significantly and therefore,
ductility.
those elements
Moreover, exhibited greater
the equilibrium paths post peak strength
of columns (and ductility)
from series A showedfor largerflattening
greater displace-
ments in comparison to columns from series A. This was caused by the local failure mode
after reaching their limit load capacities. Before reaching the vertical displacements equal
of all elements from series BS1C1 (see Figure 11c).
to approximately 8 mm, the presented relationships resemble an elastic-ideally plastic
Materials 2022, 15, 2231 15 of 26

Materials 2022, 15, x FOR PEER REVIEW 16 of 29

response by the structures. The equilibrium paths, and therefore the ductility of columns
from series B and D, are similar.

(a) (b)
Figure 14. Comparison of normalized equilibrium paths of: (a) steel columns from all series; (b)
Figure 14. Comparison of normalized equilibrium paths of: (a) steel columns from all series; (b) slen-
slender composite columns from series AS1C1 and BS1C1.
der composite columns from series AS1C1 and BS1C1.
The influence of the concrete strength on the ductility of moderately slender compo-
The influence of the wall thickness of the steel jacket on the behavior of slender
site columns can be seen in Figure 15a, where both standard and normalized equilibrium
composite columns can be seen in Figure 14b, which depicts the normalized equilibrium
paths
paths of elements from
of elements from series
series AS1C1
DS1C1and andBS1C1.
DS1C2Itare canpresented.
be noticedOn analyzing
that up to thethe pre-
vertical
sented data, it can be concluded that the use of stronger concrete
displacements equal to approximately 15 mm, the equilibrium paths of the columns made increases the ultimate
strength
from plates of columns,
3 mm thick but inliethebelow
post-peak part offor
the paths thecolumns
equilibrium made path,
of the residual
thicker steelbearing
sheets,
capacity was equal to approximately 450 kN and was a similar level
and therefore, exhibit smaller ductility. After reaching approximately 10–15 mm vertical for all depicted ele-
ments. This resulted in lower ductility of moderately slender CFST
displacement, the equilibrium paths of columns from series B start to increase, and therefore, columns filled with
stronger concrete
those elements that exhibited
exhibited a local
greater post peaktype of failure.
strength (and Itductility)
can be seen moredisplacements
for larger clearly when
analyzing the normalized equilibrium paths presented in
in comparison to columns from series A. This was caused by the local failure mode Figure 15b. In the post-peak
of all
zone, the relative force in a column
elements from series BS1C1 (see Figure 11c). filled with stronger concrete is, for the same displace-
ment,The lower compared
influence of thetoconcrete
the relative
strengthforce
on in
theaductility
column of filled with weaker
moderately slender concrete.
composite In
other words, the normalized equilibrium paths of the DS1C1 columns
columns can be seen in Figure 15a, where both standard and normalized equilibrium paths lie above the nor-
malized
of elements pathsfromof the DS1C2
series DS1C1 columns.
and DS1C2 are presented. On analyzing the presented data,
it can be concluded that the use of stronger concrete increases the ultimate strength of
columns, but in the post-peak part of the equilibrium path, the residual bearing capacity
was equal to approximately 450 kN and was a similar level for all depicted elements.
This resulted in lower ductility of moderately slender CFST columns filled with stronger
concrete that exhibited a local type of failure. It can be seen more clearly when analyzing
the normalized equilibrium paths presented in Figure 15b. In the post-peak zone, the
relative force in a column filled with stronger concrete is, for the same displacement, lower
compared to the relative force in a column filled with weaker concrete. In other words, the
normalized equilibrium paths of the DS1C1 columns lie above the normalized paths of the
DS1C2 columns.
A different conclusion can be drawn when analyzing the data presented in Figure 16a,
where normalized equilibrium paths of slender elements from series AS1C1 and AS1C2
are presented. It can be seen that the concrete strength had no influence on their ductility,
owing probably to the global buckling failure mode of those elements. In columns from
series A, most deformations (see Figure 11a) were focused on the plastic hinge formed in
the(a)center of the column as a result of the global sway failure (b) mode. The parts of the column
Figure 15. Comparison of (a) non-normalized and (b) normalized as
above and below the plastic hinge operate approximately a rigid body;
equilibrium pathstherefore,
of moderatelynot
much concrete
slender elements wasfromused
seriesinDS1C1
energyandabsorption duringconcrete
DS1C2 (different the loading process, as was the case
strength).
in all elements from series D, which exhibited local failure modes. The confirmation of the
A different conclusion can be drawn when analyzing the data presented in Figure
16a, where normalized equilibrium paths of slender elements from series AS1C1 and
AS1C2 are presented. It can be seen that the concrete strength had no influence on their
 sented data, it can be concluded that the use of stronger concrete increases the ultimate
strength of columns, but in the post-peak part of the equilibrium path, the residual bearing
capacity was equal to approximately 450 kN and was a similar level for all depicted ele-
ments. This resulted in lower ductility of moderately slender CFST columns filled with
stronger concrete that exhibited a local type of failure. It can be seen more clearly when
Materials 2022, 15, 2231
analyzing the normalized equilibrium paths presented in Figure 15b. In the post-peak 16 of 26

zone, the relative force in a column filled with stronger concrete is, for the same displace-
ment, lower compared to the relative force in a column filled with weaker concrete. In
other
abovewords, the normalized
considerations equilibrium
can be found in [34], paths
whereof
thethe DS1C1 columns
phenomenon lie above
of concrete the nor-
confinement
malized paths
was clearly of the DS1C2
observed only incolumns.
stub and moderately slender composite columns (L/D < 11).

Materials 2022, 15, x FOR PEER REVIEW 17 of 29

ductility, owing probably to the global buckling failure mode of those elements. In col-
umns from series A, most deformations (see Figure 11a) were focused on the plastic hinge
formed in the center of the column as a result of the global sway failure mode. The parts
of the column above and below the plastic hinge operate approximately as a rigid body;
therefore, not much concrete was used in energy absorption during the loading process,
as(a)
was the case in all elements from series D, which (b) exhibited local failure modes. The
confirmation
Figure of the above
15. Comparison considerations and
of (a) non-normalized can (b)
be normalized
found in [34], where the
equilibrium phenomenon
paths of
of moderately
Figure 15. Comparison of (a) non-normalized and (b) normalized equilibrium paths of moderately
concrete
slender confinement
elements wasDS1C1
from series clearlyand
observed only in stub
DS1C2 (different andstrength).
concrete moderately slender compo-
slender elements from series DS1C1 and DS1C2 (different concrete strength).
site columns (L/D < 11).
A different conclusion can be drawn when analyzing the data presented in Figure
16a, where normalized equilibrium paths of slender elements from series AS1C1 and
AS1C2 are presented. It can be seen that the concrete strength had no influence on their

(a) (b)
Figure 16. Comparison of normalized equilibrium paths of: (a) slender elements from series AS1C1
Figure 16. Comparison of normalized equilibrium paths of: (a) slender elements from series AS1C1
and AS1C2 (different concrete strength), (b) moderately slender columns made from the same grade
and AS1C2 and
of concrete (different concrete
different strength),
grades of steel,(b) moderately
series DS1 and slender
DS2. columns made from the same grade
of concrete and different grades of steel, series DS1 and DS2.
The influence of the steel strength on the behavior of composite columns can be seen
The influence of the steel strength on the behavior of composite columns can be seen
in Figure 16b. It can be seen that the normalized equilibrium paths for columns made of
in Figure 16b. It can be seen that the normalized equilibrium paths for columns made of
stronger steel lie above the equilibrium paths of columns made of weaker steel. This can
stronger steel lie above the equilibrium paths of columns made of weaker steel. This can be
be explained by the greater concrete confinement in elements made of S355 steel. The av-
explained by the greater concrete confinement in elements made of S355 steel. The average
erage steel contribution ratios ξ in the DS2C1 and DS1C1 columns were equal to
steel contribution ratios ξ in the DS2C1 and DS1C1 columns were equal to (see Table 1),
ξ(see Table 1), ξ,DS2C1 = 1.51 and ξ,DS1C1 = 1.16, respectively.
,DS2C1 = 1.51 and ξ ,DS1C1 = 1.16, respectively.

5.3. Quantitative
5.3. Analysis of
Quantitative Analysis of Equilibrium
Equilibrium Paths
Paths
Figure 17
Figure 17shows
showsthe theductility
ductilityofof analyzed
analyzed columns
columns calculated
calculated usingusing proposed
proposed ener-
energetic
measure µω,0.7 .μItω,0.7
getic measure was . Itassumed
was assumed that equivalent
that equivalent axialNforce
axial force 0 is N 0 is
equal toequal
70% to
of 70%
the of the
ultimate
ultimatecapacity
bearing bearing capacity
of columns of columns
(ω = 0,7,(ω
see= Figure
0,7, see8).
Figure
In the8).column
In the column chart,
chart, the thecolors
same same
colors indicate
indicate composite composite and
and steel steel columns
columns made ofmade of thesteel
the same same steel and
profile profile
the and
samethe same
class of
class of structural steel. CFST elements filled with C25/30 concrete
structural steel. CFST elements filled with C25/30 concrete (designation C1) are shown (designation C1) are
shown as rectangular posts, and those filled with C35/40 concrete (designation C2) as
posts with rounded edges.
On analyzing the presented data, it can be seen that most of the conclusions drawn
on the basis of the qualitative analysis that included the wider spectrum of columns’ top
Materials 2022, 15, x FOR PEER REVIEW 18 of 29

from series D made of stronger steel S2 exhibited a 33% increase in ductility in relation to
Materials 2022, 15, 2231 columns made of weaker steel S1. The wall thickness had a significant influence on 17 the
of 26
ductility of slender CFST columns—the ductility of elements from series AS1C1 with D/t
= 10 was almost 2.5 times higher than the ductility of column BS1C1 with the D/t = 16,7.
When analyzingposts,
as rectangular the ductility
and thoseof filled
steel columns, it can
with C35/40 be concluded
concrete that the
(designation highest
C2) ductil-
as posts with
ity was exhibited
rounded edges. by elements made of thicker walls, and the steel strength had no influ-
ence on the results whatsoever.

Type Series μω,0.7

AS1 19.1
AS2 18.2
steel BS1 10.0
DS1 7.4
DS2 7.8
AS1C1 30.5
AS1C2 28.1
AS2C1 24.6
BS1C1 12.3
CFST
BS1C2 12.5
DS1C1 10.7
DS1C2 8.8
DS2C1 14.3

Figure 17. Ductility of steel and CFST columns calculated using proposed μω,0.7 measure.
Figure 17. Ductility of steel and CFST columns calculated using proposed µω,0.7 measure.

5.4. Digital Image Correlation—Results

On analyzing the presented data, it can be seen that most of the conclusions drawn
Contour plots of qualitative
on the basis of the εxx, transverse
longitudinalanalysis εyy, and shear
that included εxy strains
the wider of selected
spectrum CFST
of columns’
specimens are presented in Figures 18–23. It was assumed that the strains describing
top displacements conducted in the previous chapter of this paper have been the
confirmed.
relative
In eachelongation
case, fillingarethepositive. Figures 18–20
steel cross-section withpresent
concretelongitudinal
caused a strains of moderately
significant increase in
slender columns
the ductility from
of the series
tested D. The individual
elements (for series strain contour
A: 35–60%, plots
series B:are marked
23–25%, andwith cap-
series D:
ital lettersincrease).
20–83% which correspond
In slender to the points
columns from marked
series Aonandthe Bequilibrium
that exhibited patha placed in the
global failure
lower
mode,left thecorner.
higherPoints
strengthA and B indicate
concrete fillingstrains
had no before reaching
significant the ultimate
influence on theload and
ductility
when the maximum force was reached, respectively. Points C, D and E mark selected
of the elements. However, in moderately slender columns from series D that exhibited
places in the modes,
local failure post-peakthepart of theofequilibrium
ductility specimens filledpath.with
On theC1 left
wasside,
22% the experimentally
higher in relation to
those filled with C2. Similarly, it can be observed that the ductility
obtained failure modes of specimen are presented. A red rectangle indicates the of moderately slender
area
columns
where from series were
displacements D made of stronger
measured. steel
Strain S2 exhibited
contour plots werea 33% increase
rotated in ductility
90 degrees coun-in
relation to columns made of weaker steel S1. The wall thickness had
terclockwise to the horizontal position. White spots indicate the areas where the strains a significant influence
on the the
exceed ductility
minimum of slender CFST columns—the
and maximum values assumed ductility
whileofplotting.
elements from series AS1C1
withOn D/t analyzing Figure 18, it can be seen that at point A alongofthe
= 10 was almost 2.5 times higher than the ductility column
heightBS1C1 with the
of the column,
D/t = 16.7. When analyzing the ductility of steel columns, it can
some horizontal stripes appeared, denoting places where strain increased several times be concluded that the
in
highest ductility was exhibited by elements made of thicker walls, and
relation to other cross-sections of the column. These are areas where local crushing of con- the steel strength
had no influence on the results whatsoever.
crete occurred, and in consequence the core locally lost its ability to transfer compressive
stresses. Therefore,
5.4. Digital in order to maintain a constant force in the weakened cross-section,
Image Correlation—Results
the strains and consequently stresses in the steel jacket increased. At point B, a few more
Contour plots of longitudinal εxx , transverse εyy , and shear εxy strains of selected CFST
stripes appeared, and their distribution along the height of the column became approxi-
specimens are presented in Figures 18–23. It was assumed that the strains describing the
mately uniform. In the post-peak part of the equilibrium path (points C and D), new
relative elongation are positive. Figures 18–20 present longitudinal strains of moderately
stripes did not form. In the third from the right strip, where highest strains were observed
slender columns from series D. The individual strain contour plots are marked with capital
from the beginning, further localization of strain occurred. Local outward buckling of steel
letters which correspond to the points marked on the equilibrium path placed in the lower
was observed, denoting the area where shear and sliding failure occurred (points D and
left corner. Points A and B indicate strains before reaching the ultimate load and when the
E).
maximum force was reached, respectively. Points C, D and E mark selected places in the
post-peak part of the equilibrium path. On the left side, the experimentally obtained failure
modes of specimen are presented. A red rectangle indicates the area where displacements
were measured. Strain contour plots were rotated 90 degrees counterclockwise to the
horizontal position. White spots indicate the areas where the strains exceed the minimum
and maximum values assumed while plotting.
Materials 2022, 15, 2231 18 of 26
Materials 2022, 15, x FOR PEER REVIEW 19 of 29

Figure 18. Evolution of strains εxx in column DS1C2_1, steel S235, stronger concrete, ξ = 0.90.Points
(A,B) indicate strains before reaching the ultimate load and when the maximum force was reached,
respectively. Points (C–E) mark selected places in the post-peak part of the equilibrium path.
 Materials 2022, 15, x FOR PEER REVIEW 20 of 29

Figure 18. Evolution of strains εxx in column DS1C2_1, steel S235, stronger concrete, ξ = 0.90.Points
Materials 2022, 15, 2231 19 of 26
(A,B) indicate strains before reaching the ultimate load and when the maximum force was
reached, respectively. Points (C–E) mark selected places in the post-peak part of the equilibrium
path.

Materials 2022, 15, x FOR PEER REVIEW D 21 of 29

Figure 19. Evolution of strains εxx in column DS2C1_2, steel S355, weaker concrete, ξ = 1.51. Points
Figure
(A,B)19. Evolution
indicate strainsofbefore
strains εxx in column
reaching DS2C1_2,
the ultimate steel
load and whenS355, weaker concrete,
the maximum force wasξ = 1.51. Points
reached, respectively. Points (C–E) mark selected places in the post-peak part of the equilibrium
(A,B) indicate strains before reaching the ultimate load and when the maximum force was reached,
path.
respectively. Points (C–E) mark selected places in the post-peak part of the equilibrium path.
Materials 2022, 15, 2231 20 of 26
Materials 2022, 15, x FOR PEER REVIEW 22 of 29

Figure 20. Evolution of strains εxx in column DS1C1_1, steel S235, weaker concrete, ξ = 1.16. Points
Figure 20. Evolution
(A,B) indicate strains of strains
before εxx in column
reaching DS1C1_1,
the ultimate steel
load and S235,the
when weaker
maximumconcrete, = 1.16. Points
force ξwas
reached, respectively. Points (C–E) mark selected places in the post-peak part of the equilibrium
(A,B) indicate strains before reaching the ultimate load and when the maximum force was reached,
path.
respectively. Points (C–E) mark selected places in the post-peak part of the equilibrium path.
Materials 2022, 15, 2231 21 of 26
Materials 2022, 15, x FOR PEER REVIEW 23 of 29

(a)

(b)

Figure 21. (a) Longitudinal εxx and (b) transverse εyy strains in column DS1C1_2 captured in the
same
Figure
Materials 2022, 15, x FOR PEER REVIEW21.point
(a) of the equilibriumεxx
Longitudinal path.
and (b) transverse εyy strains in column DS1C1_2
24 captured
of 29 in the
same point of the equilibrium path.

(a)

(b)

Figure 22. (a) Longitudinal εxx and (b) transverse εyy strains in column DS2C1_1 captured in the
Figure (a) Longitudinal
22. point
same of the equilibrium and (b) transverse εyy strains in column DS2C1_1 captured in the
εxxpath.
same point of the equilibrium path.
Materials 2022, 15,
Materials 2022, 15, x2231
FOR PEER REVIEW 2522of
of 29
26

(a) (b)
Figure
Figure 23.
23. Longitudinal strainsεxxεxxinin
Longitudinalstrains slender columns:
slender (a) (a)
columns: BS1C2_1 (ξ =(ξ1.52)
BS1C2_1 and and
= 1.52) (b) AS1C1_3 (ξ =
(b) AS1C1_3
4.12). Description in the text.
(ξ = 4.12). Description in the text.

Significantly
On analyzingdifferent
Figure 18, strain
it canlocalization
be seen thatpatterns
at pointcan be observed
A along the height in of
Figure 19 for
the column,
column DS2C1_2 stripes
some horizontal made ofappeared,
stronger steel and weaker
denoting concrete.
places where Until
strain the ultimate
increased loadtimes
several was
reached (point B), the horizontal localizations of strains were not observed.
in relation to other cross-sections of the column. These are areas where local crushing of On the other
hand, in the
concrete post-critical
occurred, and inpart of the equilibrium
consequence path (point
the core locally lost itsC), diagonal
ability stripes
to transfer appeared.
compressive
This denotes that in specimens made of weaker steel and filled with
stresses. Therefore, in order to maintain a constant force in the weakened cross-section, stronger concrete
the
(DS1C2—horizontal
strains and consequently yieldstresses
areas),ina the
different concrete
steel jacket failureAt
increased. type occurred
point B, a fewcompared
more stripesto
those madeand
appeared, of stronger steel and filled
their distribution alongwith weakerofconcrete
the height the column(DS2C1—diagonal yield ar-
became approximately
eas).
uniform. In the post-peak part of the equilibrium path (points C and D), new stripes did
A combination
not form. In the thirdof from
the two
thepreviously
right strip, described behaviors
where highest strains can be observed
were observedin Figure
from the
20, where strain contour plots of DS1C1_1 column made of weaker steel
beginning, further localization of strain occurred. Local outward buckling of steel was and filled with
weaker
observed,concrete
denotingarethe
presented.
area where It can
shearbeand
seensliding
that although in the pre-peak
failure occurred (points Dpartandof E).the
Materials 2022, 15, 2231 23 of 26

Significantly different strain localization patterns can be observed in Figure 19 for

column DS2C1_2 made of stronger steel and weaker concrete. Until the ultimate load was
reached (point B), the horizontal localizations of strains were not observed. On the other
hand, in the post-critical part of the equilibrium path (point C), diagonal stripes appeared.
This denotes that in specimens made of weaker steel and filled with stronger concrete
(DS1C2—horizontal yield areas), a different concrete failure type occurred compared to
those made of stronger steel and filled with weaker concrete (DS2C1—diagonal yield areas).
A combination of the two previously described behaviors can be observed in Figure 20,
where strain contour plots of DS1C1_1 column made of weaker steel and filled with weaker
concrete are presented. It can be seen that although in the pre-peak part of the equilibrium
path no strain localization was observed (similarly to the DS2C1_1 column), when the
ultimate load was reached (point B), horizontal yield zones appeared, similarly to the
DS1C2_1 column. However, as the column shortening increased at the transition from
point B to point C, new diagonal yield zones developed.
It was previously observed that the ductility of moderately slender columns from
series DS2C1 made of stronger steel S2 exhibited a 33% increase in ductility in relation to
columns DS1C1 made of weaker steel S1. This was due to the greater concrete confinement
in such elements. This may indicate that there is a correlation between the intensity of the
occurrence of diagonal yield zones and the level of concrete confinement in elements that
exhibit local shear and sliding or elephant foot-type failure. In CFST columns, concrete
confinement is always accompanied by transverse tensile strains of the steel jacket. This
can be observed in Figures 21 and 22, which present contour plots of longitudinal and
transverse strains in the post-peak part of the equilibrium paths of the DS1C1_2 and
DS2C1_1 columns, respectively. It can be seen that the transverse strains in diagonal yield
areas are higher in comparison to strains in surrounding zones. Similar diagonal failure
mechanisms were observed in [56], where localization of hoop strains in cylindrical FRP-
wrapped stub concrete columns was investigated. Depending on the length of the confined
concrete cylinder, a number of fully localized failure mechanisms developed, each one of
them then involving solid wedge movements along a shear frictional surface, causing stress
concentrations in the FRP wrapping and thus precipitating its premature failure.
Figure 23a,b presents the longitudinal strains in slender columns AS1C1_3 (ξ = 4.12)
and BS1C2_1 (ξ = 1.52), respectively. BS1C2_1 was the only specimen from series B that
exhibited global buckling failure mode. The top and bottom strain contour plots were
captured close to the ultimate force and in the post peak part of the equilibrium paths,
respectively. In both specimens, a combination of global and local outward buckling
occurred. In the middle of both columns, plastic hinges denoted by elliptical shapes
representing the bulged steel jacket (showed in the lower contour plots) were formed. It
seems interesting that similarly to series D columns, which all exhibited local failure modes,
in the BS1C2_1 element, after reaching the ultimate force, the diagonal yield pattern also
appeared, but was located below the plastic hinge that formed in the middle. The reason
for this may be larger imperfections in this specimen that caused the global buckling to
occur before the local concrete shear could have been fully developed. An outline of similar
diagonal yield patterns was also observed in other specimens from series B, but not in
any of the specimens from series A. This may indicate that the geometric and material
properties of elements from series BS1C2 delimit the global and local failure modes of
axially compressed CFST columns.

6. Conclusions
In this study, a total of 21 composite CFST and 5 steel rectangular columns of mod-
erate slenderness were tested to investigate their ductility under axial compression. The
importance of the vertical ductility of columns was discussed, and a new ductility mea-
sure considering the equivalent preloading prior to the extreme event was proposed and
utilized to examine the ductility of tested specimens. The test results were presented and
Materials 2022, 15, 2231 24 of 26

analyzed in detail to provide information about the factors influencing the ductility of
axially compressed CFST columns.
The conducted analyses showed that the key feature influencing the ductility of axially
compressed columns is their ability to dissipate the energy of imposed loads. The larger
the volume of a material in the element that may permanently deform (in the case of steel)
or crush (in the case of concrete) and consequently dissipate the energy, the greater this
ability. Due to the confinement of a concrete core by a steel jacket, and most importantly,
the limitation to local buckling of the steel cross-section by the concrete core, all tested
composite columns showed greater ductility than their steel counterparts.
The analysis concluded that the vertical ductility levels of axially compressed CFST
columns with rectangular cross-sections highly depend on their failure modes. Therefore,
the detailed conclusions were divided into two categories, taking into account the failure
modes of the analyzed columns. In general, the ductility of specimens with local failure
modes was higher in comparison to columns that exhibited global buckling failure.
The conclusions for elements that underwent global buckling:
• The concrete core failure consisted of crushing the concrete in the plastic hinge on the
side of the cross-section with high compressive stresses. In elements with large D/t
ratios, the additional formation of diagonal yield zones in the other cross-sections of
the columns occurred.
• The higher strength concrete filling had no significant influence on the ductility of
slender test elements.
• The increase in the wall thickness had a positive effect on the ductility of slender
specimens.
• For elements that underwent the local failure mode:
• In shear and sliding failure mode, the concrete core failure consists of the formation
of diagonal or horizontal yield zones along the height of the column, causing stress
concentrations in the steel jacket, and in the weakest yield zone, shear solid wedge
movements along a shear frictional surface develop. In elephant foot failure mode, the
concrete core failure consists of crushing the concrete in the weakest cross-section of
the column.
• Symmetrical elephant foot failure mode yielded higher vertical ductility in comparison
to shear and sliding failure modes.
• The use of higher-grade steel increased the vertical ductility of the tested columns.
• The use of stronger concrete caused a decrease in the ductility of the test specimens.

Author Contributions: Conceptualization, B.G.; data curation, B.G.; formal analysis, B.G.; investi-
gation, B.G.; methodology, B.G.; project administration, B.G.; software, B.G.; supervision, E.D.S.;
Validation, E.D.S.; visualization, B.G.; writing—original draft, B.G.; writing—review and editing,
E.D.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Kiakojouri, F.; Sheidaii, M.R.; De Biagi, V.; Chiaia, B. Progressive collapse of structures: A discussion on annotated nomenclature.
Structures 2021, 29, 1417–1423. [CrossRef]
2. Cai, J.; Pan, J.; Wu, Y. Mechanical behavior of steel-reinforced concrete-filled steel tubular (SRCFST) columns under uniaxial
compressive loading. Thin Walled Struct. 2015, 97, 1–10. [CrossRef]
3. Le Hoang, A.; Fehling, E. Numerical study of circular steel tube confined concrete (STCC) stub columns. J. Constr. Steel Res. 2017,
136, 238–255. [CrossRef]
Materials 2022, 15, 2231 25 of 26

4. Jiang, Y.; Silva, A.; Castro, J.M.; Monteiro, R. Experimental Assessment of the Behaviour of Rubberized Concrete Filled Steel Tube
Members. In Proceedings of the V International Conference on Behavior of Steel Structures in Seismic Areas, Shanghai, China, 15
July 2015.
5. Lin, S.Q.; Zhao, Y.G. Ultimate Stress of the Steel Tube in Circular CFT Stub Columns Subjected to Axial Compression. Int. J. Eng.
Technol. 2018, 10, 315–319. [CrossRef]
6. Mendoza, R., Jr.; Yamamoto, Y. Simulation of Failure Localization and Buckling Behavior of Steel-tube Confined Concrete
Columns using Coupled RBSM-Shell FEM Analysis. In Proceedings of the Computational Engineering Conference JSCES,
Maastricht, The Netherlands, 12 June 2018.
7. Liew, J.R.; Xiong, D.X. Ultra-high strength concrete filled composite columns for multi-storey building construction. Adv. Struct.
Eng. 2012, 15, 1487–1503. [CrossRef]
8. Portolés, J.M.; Serra, E.; Romero, M.L. Influence of ultra-high strength infill in slender concrete-filled steel tubular columns. J.
Constr. Steel Res. 2013, 86, 107–114. [CrossRef]
9. Li, D.; Uy, B.; Aslani, F.; Hou, C. Behaviour and design of spiral-welded stainless steel tubes subjected to axial compression. J.
Constr. Steel Res. 2019, 154, 67–83. [CrossRef]
10. Park, R.; Paulay, T. Reinforced Concrete Structures; John Wiley & Sons: Hoboken, NJ, USA, 1975.
11. Grzeszykowski, B.; Szmigiera, E. Ductility assessment of two-chord composite steel-concrete battened columns. Struct. Infrastruct.
Eng. 2017, 13, 1414–1424. [CrossRef]
12. Kitada, T. Ultimate strength and ductility of state-of-the-art concrete-filled steel bridge piers in Japan. Eng. Struct. 1998, 20,
347–354. [CrossRef]
13. Schneider, S.P. Axially loaded concrete-filled steel tubes. J. Struct. Eng. 1998, 124, 1125–1138. [CrossRef]
14. Feng, P.; Cheng, S.; Bai, Y.; Ye, L. Mechanical behavior of concrete-filled square steel tube with FRP-confined concrete core
subjected to axial compression. Compos. Struct. 2015, 123, 312–324. [CrossRef]
15. Abdulraheem, M.S. Experimental investigation of fire effects on ductility and stiffness of reinforced reactive powder concrete
columns under axial compression. J. Build. Eng. 2018, 20, 750–761. [CrossRef]
16. Hadi, M.N.S.; Khan, Q.S.; Sheikh, M.N. Axial and flexural behavior of unreinforced and FRP bar reinforced circular concrete
filled FRP tube columns. Constr. Build. Mater. 2016, 122, 43–53. [CrossRef]
17. Al-Nimry, H.S.; Al-Rabadi, R.A. Axial–Flexural Interaction in FRP-Wrapped RC Columns. Int. J. Concr. Struct. Mater. 2019, 13, 53.
[CrossRef]
18. GangaRao, H.V.; Taly, N.; Vijay, P.V. Reinforced Concrete Design with FRP Composites; CRC Press: Boca Raton, FL, USA, 2006.
19. Kusumawardaningsih, Y.; Hadi, M.N.S. Comparative behaviour of hollow columns confined with FRP composites. Compos.
Struct. 2010, 93, 198–205. [CrossRef]
20. Wang, J.; Shen, Q.; Jiang, H.; Pan, X. Analysis and Design of Elliptical Concrete-Filled Thin-Walled Steel Stub Columns Under
Axial Compression. Int. J. Steel Struct. 2018, 18, 365–380. [CrossRef]
21. Szmigiera, E. Dwugał˛eziowe Słupy Stalowo-Betonowe; Budownictwo; Prace Naukowe Politechniki Warszawskiej: Warsaw, Poland, 2012.
22. He, A.; Wang, F.; Zhao, O. Experimental and numerical studies of concrete-filled high-chromium stainless steel tube (CFHSST)
stub columns. Thin Walled Struct. 2019, 144, 106273. [CrossRef]
23. Gao, B.; Wang, J.; Shen, Q.; Wang, C.; Yu, Z. Eccentric compression tests and design of welded steel tube strengthened RC stub
columns with high-strength grout. J. Build. Eng. 2021, 35, 102072. [CrossRef]
24. Ghanbari-Ghazijahani, T.; Nabati, A.; Azandariani, M.G.; Fanaie, N. Crushing of steel tubes with different infills under partial
axial loading. Thin Walled Struct. 2020, 149, 106614. [CrossRef]
25. Zhao, X.L.; Hancock, G.J. Tests to determine plate slenderness limits for cold-formed rectangular hollow sections of grade C450.
Steel Constr. 1991, 25, 2–16.
26. Yang, H.; Lam, D.; Gardner, L. Testing and analysis of concrete-filled elliptical hollow sections. Eng. Struct. 2008, 30, 3771–3781.
[CrossRef]
27. Wang, J.; Wang, F.; Shen, Q. Numerical modelling and design recommendation of axially-loaded thin-walled RCFST slender
column. Thin Walled Struct. 2019, 135, 210–226. [CrossRef]
28. İpek, S.; Erdoğan, A.; Güneyisi, E.M. Compressive behavior of concrete-filled double skin steel tubular short columns with the
elliptical hollow section. J. Build. Eng. 2021, 38, 102200. [CrossRef]
29. Park, R. Evaluation of ductility of structures and structural assemblages from laboratory testing. Bull. N. Z. Natl. Soc. Earthq. Eng.
1989, 22, 155–166. [CrossRef]
30. Gardner, N.J.; Jacobson, E.R. Structural behavior of concrete filled steel tubes. J. Proc. 1967, 64, 404–413.
31. Mander, J.B.; Priestley, M.J.N.; Park, E. Theoretical stress-strain model for confined concrete. J. Struct. Eng. 1988, 114, 1804–1826.
[CrossRef]
32. Ho, J.; Lam, J.; Kwan, A. Effectiveness of adding confinement for ductility improvement of high-strength concrete columns. Eng.
Struct. 2010, 32, 714–725. [CrossRef]
33. Knowles, R.B.; Park, R. Strength of concrete filled steel columns. J. Struct. Div. 1969, 95, 2565–2587. [CrossRef]
34. Knowles, R.B.; Park, R. Axial load design for concrete filled steel tubes. J. Struct. Div. 1970, 96, 2125–2153. [CrossRef]
35. Zhu, J.Y.; Chan, T.M. Experimental investigation on octagonal concrete filled steel stub columns under uniaxial compression. J.
Constr. Steel Res. 2018, 147, 457–467. [CrossRef]
Materials 2022, 15, 2231 26 of 26

36. Han, L.H.; Li, W.; Bjorhovde, R. Developments and advanced applications of concrete-filled steel tubular (CFST) structures:
Members. J. Constr. Steel Res. 2014, 100, 211–228. [CrossRef]
37. Patel, V.I.; Hassanein, M.F.; Thai, H.T.; Al Abadi, H.; Elchalakani, M.; Bai, Y. Ultra-high strength circular short CFST columns:
Axisymmetric analysis, behaviour and design. Eng. Struct. 2019, 179, 268–283. [CrossRef]
38. Han, L.H. Tests on stub columns of concrete-filled RHS sections. J. Constr. Steel Res. 2002, 58, 353–372. [CrossRef]
39. Zhang, L.; Yang, S.L.; Fu, B.; Tong, G.S.; Tong, J.Z.; Jing, T. Behavior and design of concrete-filled narrow rectangular steel tubular
(CFNRST) stub columns under axial compression. J. Build. Eng. 2021, 37, 102166. [CrossRef]
40. EN-1994-1-1:2004; Eurocode 4: Design of Composite Steel and Concrete Structures-Part 1-1: General Rules and Rules for Buildings;
European Union: Brussels, Belgium, 2004.
41. Yang, C.; Yu, Z.X.; Sun, Y.P.; Zhao, L.; Zhao, H. Axial residual capacity of circular concrete-filled steel tube stub columns
considering local buckling. Adv. Steel Constr. 2018, 14, 496–513.
42. Lundberg, J.E. The reliability of composite columns and beam columns. Univ. Minn. 1993, 1867.
43. Dai, X.H.; Lam, D.; Jamaluddin, N.; Ye, J. Numerical analysis of slender elliptical concrete filled columns under axial compression.
Thin Walled Struct. 2014, 77, 26–35. [CrossRef]
44. An, Y.F.; Han, L.H.; Zhao, X.L. Behaviour and design calculations on very slender thin-walled CFST columns. Thin Walled Struct.
2012, 53, 161–175. [CrossRef]
45. Phan, D.H.H.; Patel, V.I.; Liang, Q.Q.; Al Abadi, H.; Thai, H.T. Numerical investigations of circular double-skin steel tubular
slender beam-columns filled with ultra-high-strength concrete. Eng. Struct. 2022, 254, 113814. [CrossRef]
46. Patel, V.I.; Liang, Q.Q.; Hadi, M.N. Nonlinear analysis of circular high strength concrete-filled stainless steel tubular slender
beam-columns. Eng. Struct. 2017, 130, 1–13. [CrossRef]
47. Yu, T.; Teng, J.G.; Wong, Y.L.; Dong, S.L. Finite element modeling of confined concrete-I: Drucker–Prager type plasticity model.
Eng. Struct. 2010, 32, 665–679. [CrossRef]
48. Yu, T.; Teng, J.G.; Wong, Y.L.; Dong, S.L. Finite element modeling of confined concrete-II: Plastic-damage model. Eng. Struct. 2010,
32, 680–691. [CrossRef]
49. Tao, Z.; Wang, Z.B.; Yu, Q. Finite element modelling of concrete-filled steel stub columns under axial compression. J. Constr. Steel
Res. 2013, 89, 121–131. [CrossRef]
50. Azad, S.K.; Uy, B. Effect of concrete infill on local buckling capacity of circular tubes. J. Constr. Steel Res. 2020, 165, 105899.
[CrossRef]
51. Zarringol, M.; Thai, H.T.; Ngo, T.; Patel, V. Behaviour and design calculations of rectangular CFST beam-columns with slender
sections. Eng. Struct. 2020, 222, 111142. [CrossRef]
52. Zarringol, M.; Thai, H.T.; Thai, S.; Patel, V. Application of ANN to the design of CFST columns. Structures 2020, 28, 2203–2220.
[CrossRef]
53. Zarringol, M.; Thai, H.-T.; Naser, M. Application of machine learning models for designing CFCFST columns. J. Constr. Steel Res.
2021, 185, 106856. [CrossRef]
54. Zarringol, M.; Thai, H.T. Prediction of the load-shortening curve of CFST columns using ANN-based models. J. Build. Eng. 2022,
51, 104279. [CrossRef]
55. Aramis, Version 6.1; User Manual-Software; GOM mbH: Braunschweig, Germany, 2009.
56. Tabbara, M.; Karam, G. Parametric Investigation of the Effects of Localization and Slenderness on the Stress–Strain Response and
Confinement Efficiency in FRP-Wrapped Concrete Cylinders. Appl. Sci. 2020, 10, 3432. [CrossRef]