Академический Документы
Профессиональный Документы
Культура Документы
PII: S0263-8223(19)31846-X
DOI: https://doi.org/10.1016/j.compstruct.2019.111543
Reference: COST 111543
Please cite this article as: Cheng, S., Feng, P., Meng, X., Li, Z., Du, J., Compression Behavior of Large-Scaled
Cylindrical GFRP Chimney Liner Segments, Composite Structures (2019), doi: https://doi.org/10.1016/
j.compstruct.2019.111543
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will
undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing
this version to give early visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Shi CHENGa,b, Peng FENGa,*, Xinmiao Mengc, Zhiyuan LIa, Jike DUd
a. Key Lab of Civil Engineering Safety and Durability of China Education Ministry,
Department of Civil Engineering, Tsinghua University, Beijing 100084, China;
b. China Coal Technology and Engineering Group, Beijing 100013, China;
c. Department of Civil Engineering, Beijing Forestry University, Beijing 100083,
China
d. Northwest Electric Power Design Institute Co., Ltd. Xi’an 710075, China.
* Corresponding author : Tel: +86 10 6277 2456; Email: fengpeng@tsinghua.edu.cn .
ABSTRACT:
As a pivotal anti-corrosion structure in the wet flue gas desulfurization system, the
as the chimney liner. However, the study on its mechanical properties is rare. Three
large-scale stiffened cylindrical GFRP chimney liner segments were tested under the
with two ring stiffeners, a specimen with two ring stiffeners cut into two segments and
joined by a hand-wound technique, and a specimen with an outside ring bracket. The
process, were acquired by test. The effects of ring stiffeners on the mechanical
behavior, service reliability of the joint between the two segments of the chimney
liner, and the reliability of the ring bracket were examined. Comparisons between
discussed. Finally, finite element analysis was performed to examine the failure
Joint; Bracket
1 Introduction
The requirements on the emission of clean exhaust gas from coal-fired power plants
(Fig. 1a) are steadily growing due to environmental protective regulations [1]. This
leads to the broader applications of Flue Gas Desulfurization (FGD) [2-4]. FGD is a
process of removing sulfur monoxide (SO) and sulfur dioxide (SO2) produced in
boilers as the by-products of flue gases and combustion at power plants (Fig. 1a).
Through this process, the pollutants of the exhaust gas are removed before being
discharged into the atmosphere. However, the vapor after FGD will condense on the
wall, absorb acidic gases, and then form acid drops, which will decrease the durability
of RC chimney severely. Hence, the FGD technique has created a large demand for a
type of chimney liner material for concrete chimneys with good performance in
protecting the primary outer structure from chemicals, as well as the thermal and
corrosive environment [5]. To meet these demands, GFRP liners have been proposed
fibers, where the polymer is usually epoxy, vinyl ester [9] or polyester thermosetting
plastic [10] and phenol-formaldehyde resin [11]. The fibers are usually glass [12-14],
carbon, aramid or basalt [15], among which the glass fiber is the most practical choice
for chimney liners due to the low cost and good performance. The GFRP chimney
liner often adopts the automatic filament-wound technique, and its wall construction
typically consists of three distinctly different layers. The outermost and innermost
layers are both corrosion barriers [16], which consists of one or more laminates of
chopped strand mat. These layers are nonstructural and are designed to protect the
sandwiched structural layer from damage caused by the corrosive flue gas. The
sandwiched structural layer consists several layers of GFRP laminates [17], which are
properties can be achieved by the winding angle of the filaments, the thickness and
number of laminates [18] , which meet the required strength to resist gravity, the
lateral loads due to flue gas, wind or earthquakes, and the thermal loads on the
chimney liner.
with lengths of 6-18 m [17] which are joined together by placing one on top of the
other to construct a continuous conduit inside the primary chimney. The chimney is a
tall and slender structure in compression (above support) or tension (below support)
as shown in Fig. 1a, where the tensile behavior of chimney liner was investigated by
Cheng et al. [19]. The wall of the liner is composed of anti-corrosion, structural, and
protective layers, as shown in Fig. 1b. The structural layer provides sufficient strength
to stand up. The joint between the two segments of the chimney liners is
After the construction of the chimney liners, the installation on a reinforced concrete
primary chimney structure is a substantial procedure that requires the existence of ring
brackets on the GFRP chimney liners for the purpose of lateral support.
Many researchers have studied the dynamic behavior of cylindrical shells, the
have also focused on FRP chimney liners, including the prediction of ovalling
frequency for FRP chimney liners with circumferential stiffeners [23], the thermal
analysis of FRP chimney liners in uncontrolled fires [24] and the thermal blistering of
FRP chimney liners [17]. However, few studies have investigated the quasi-static
liners with the scaled-down ratio of 3:1 (practical to experimental size) subjected to
axial compression were carried out. This experimental investigation determined the
variations of lateral strains during the loading process. The effects of ring stiffeners on
the mechanical behavior, service reliability of the joint between the two segments of
the chimney liner, and the reliability of the ring bracket were examined. In addition,
stiffness were discussed. Finally, FEM analysis was performed using the software of
MSC. Marc 2010 to examine the failure modes and axial load-displacement behavior
of the specimens. Different diameter to thickness ratios and load eccentricity values
were used to examine their effects on the mechanical behavior of chimney liners.
Based on the experimental investigation and FEM analysis, suggestions on the design
of chimney liner structure were given.
2 Experimental Investigation
vinyl ester resin [25] and ECR glass fibers. In total, nine layers with different fiber
densities and resin percentages were employed, including the surfacing mat for
protection, hoop winding roving, chopped strand mat, and unidirectional cloth. The
stacking sequence and details of the specimens are listed in Table 1. The direction of
fiber is 0° in the hoop direction and 90° in the axial direction. It is noted that most
material properties were cut from the plates which had the same layers as the chimney
liner specimens, as shown in Fig. 2. These test specimens were 53 mm in the test
strain gauges centered on each specimen in the compressive direction. The averaged
results show the values of 12.3 GPa for the compressive elastic modulus, 146.2 MPa
for the compressive ultimate compressive stress, and 0.012 for the corresponding
ultimate strain. Additionally, to understand the material properties of FRP better, the
for tension [27], GB/T 1449-2005 for bending [28] and GB/T 1450.1-2005 for shear
2.2 Specimens
chimney liner specimens were fabricated with the scaled-down ratio of 3:1 (practical
to experimental size): YT-1, YT-2 and YT-3 (Figs. 3 a, b and c). The diameter and
wall-thickness of the GFRP chimney liners were both scaled down by a 3:1 ratio to
ensure the ratio of the diameter to wall thickness was a constant value. All specimens
had the same height of 3000 mm, diameter of 2800 mm, and thickness of 5.6 mm and
their applied conditions in a chimney liner, including the regular segment of YT-1,
Fig. 3 are the original filament-wounded specimens, which were not stiffened and
thus required further manufacturing. For the purpose of avoiding premature buckling,
two stiffeners were installed at the distance of one third and two thirds from top tip of
the specimen on specimen YT-1 and YT-2, which adopted the technique of foaming
The specimen YT-1 was designed to evaluate the load-bearing capacity and
different segments of a GFRP chimney liner due to the length limitation for a single
segment. The stiffened specimen YT-2 was manufactured by cutting the specimen
into two parts at the middle cross-section, then joining by the hand-wound GFRP to
form a ring band with a width of 100 mm and a thickness of 12 mm as the inside and
outside connection layers, respectively, as shown in Fig. 3d; this served as the final
experimental specimen, YT-2 (Fig. 4b). The specimen YT-3 (Fig. 4c) was designed to
validate the load-bearing capacity of the supporting measure, which was fabricated
with a ring bracket at the middle height of the specimen along the longitudinal
For the specimens YT-1 and YT-2, five longitudinal strain gauges with a length of 10
mm were attached towards the east, north, west and southward surfaces of the
specimen, and five transversal strain gauges with a length of 10 mm were installed
were installed on the east side at distances of 410, 1245, 1500, 1755 and 2590 mm
from the bottom of specimens YT-1 (Fig. 6a) and YT-2 (Fig. 6b). Eight lateral DTs
were installed at a distance of 1500 mm from the bottom of the specimens and
towards the east, north, west, south, northeast, northwest, southwest and
southeastward surfaces of YT-1 (Fig. 7a) and YT-2 (Fig. 7b). For specimen YT-3, the
strain gauges were attached at distances of 1500, 1755 and 2590 mm from the bottom
of the specimen and correspond to the same locations as for specimens YT-1 and
YT-2. The axial DTs were installed on the eastward surface of the specimen and at
distances of 2100 and 2550 mm (Fig. 6c). Six additional axial DTs were installed
around the middle of the specimen with a spacing of one-sixth of the circumferential
perimeter (Fig. 6d). The measurement ranges of all DTs, except DT3 to DT8 of
specimen YT-3 (Fig. 6d) were from the locations of DTs installed on the bottom of
the brackets. The measurement ranges of DT3 to DT8 (Fig. 6d) of specimen YT-3
were from the locations of DTs installed on the top of the specimen.
For convenient loading on the specimen, a total weight of 7,000 kg of the steel
cap and base were manufactured according to the size of the specimens. The steel cap
was made of three straight steel I-beams and one ring steel I-beam, as shown in Fig.
8a, and the top 300 mm region was fixed with an inner ring steel clamp (Fig. 8b) and
an outer ring steel clamp (Fig. 8c) which were connected by 24 bolts with a radius of
22 mm. The steel base was used to fix the specimen along with an inner ring steel
clamp (Fig. 8b) and an outer ring steel clamp (Fig. 8c) that were connected by 24
bolts with a radius of 22 mm. The steel base was fixed to the floor with mortar. The
bottoms of specimens YT-1 and YT-2 were placed on the floor. In contrast, the
bottom of specimen YT-3 was suspended 50 mm above the floor. YT-3 was supported
half way up the specimen by a bracket in order to test the service reliability of the ring
bracket.
The axial compressive experiments for the three GFRP chimney liner
specimens were simultaneously carried out using two 2,500 kN Universal Loading
Machines. The two machines were loaded at any two locations along the same straight
steel I-beam as shown in Fig. 8d, with a manually controlled displacement rate of 1
mm/min.
3 Experimental Results
The conventional signs of axial compressive displacement as well as strain and lateral
inward deformation are symbolized by the plus sign (+), whereas the axial tensile
displacement, and strain and lateral outward deformation are denoted by the minus
sign (-). The top 25% region, the middle region between the two stiffeners, and the
lower 25% region are conventionalized as part I, part II and part III, respectively.
In general, three chimney liner specimens showed similar mechanical behavior when
subjected to axial compression, which failed due to overall buckling that developed
from local buckling. However, the major locations of buckling were different, which
were induced due to the existence of a joint band for YT-2 and the existence of a
Initially, the loading of axial compression was carried out on specimen YT-1.
The specimen did not show any clear lateral deformations on the surface until the load
reached 70% of the maximum load-bearing capacity. At this loading level, inward
local buckling was observed on the northeastward surface in the center of the
specimen (part II). During the loading process, lateral deformation of local buckling
developed, and outward local buckling was observed in addition to the inward local
buckling. This local buckling was mainly produced in the center of the specimen (part
II), particularly along the height direction and on the northwest, east and southward
surfaces, until the local buckling developed to a continuous sine-wave shape on the
circumferential surface in the center of the specimen (part II) (Fig. 7a). In addition,
significant local buckling was observed on the surface of the top 25% region (part I)
and lower 25% region (part III) (Fig. 7a). The loading process of the YT-1 specimen
was manually terminated when the load-bearing capacity was reduced to 75% of the
maximum load-bearing capacity during the period after maximum load. Subsequently,
during the unloading process, the axial and lateral deformations due to axial loading
gradually decreased and the specimen almost regained its original state, even the
blenching traces due to wrinkles could be visually recognized (Fig. 7b). This
demonstrated that the specimens were restored to their original shape without residual
deformation after unloading and the buckling of FRP is mainly elastic buckling.. The
failure mode of specimen YT-2 was similar to YT-1; however, the location of
buckling was concentrated on the top 25% region of the specimen (part I) (Figs. 7c
and d). This deformation resulted from the strengthening of the joint in the middle
region of the specimen. Additionally, specimen YT-3, with a bracket at its center,
displayed a similar behavior to specimens YT-1 and YT-2, developing buckling in the
same region (part I) (Figs. 7e and f) as YT-2. The overall failure mode of YT-3 was
buckling, rather than the strength failure of the ring bracket; this indicates that the
strength-bearing capacity of the bracket can meet the reliable service requirements.
Axial load-axial displacement relationships are shown in Figs. 6a, b, c and d for
specimens YT-1, YT-2 and YT-3. Each of the axial displacements was obtained from
Fig. 6a shows five axial load-axial displacement relationships for YT-1, where
the positive and negative displacements signify the compressive and tensile
deformations, respectively. As shown in Fig. 6a, DT1, DT2, DT3 and DT4 were
installed along the same longitudinal line and at distances of 410, 1245, 1500, and
1755 mm from the top of the specimen respectively, through which similar axial
load-axial displacement responses were measured. These four curves started from an
elastic period and reached a peak load of 662.3 kN and followed by a softening
load-bearing capacity, loading was manually terminated. The specimen entered the
unloading period and terminated when the load-bearing capacity reached 0 kN.
According to the observations, local buckling started to occur when the load-bearing
capacity reached around 70% of the maximum load-bearing capacity. Buckling then
developed and eventually formed a shape similar to a sinewave in the middle of the
specimen (Fig. 9a). Moreover, both the maximum axial displacement and the slope of
the elastic period obtained from each axial load-axial displacement curve decreased
with an increase in distance from the top of the specimen. In addition, the four curves
However, DT5, which was installed below DT4 at a distance of 2590 mm from the
top of the specimen, measured a state of tensile deformation. This curve had an
ascending bi-linear shape, which indicated slight tensile displacement occurred (from
DT5) before the axial load reached 70% of the maximum load-bearing capacity, and a
significant increase in the development of axial displacement. This met the visual
observations of the specimen under axial compression loading. From 70% of its
the region of part II (Fig. 9a) and the bottom of the specimen.
which were measured using five displacement transducers. The transducers were
installed along the same longitudinal line, and DT1 to DT5 were distributed at 410,
1245, 1500, 1755 and 2590 mm from the top of the specimen, respectively. All curves
except the curve measured by DT1 show a similar shape developed from an elastic
period to a plateau period; the peak load was reached at 710.3 kN. Subsequently, an
unloading period below the loading curve occurred. The curve measured by DT1
indicated an elastic period under compression occurred before reaching 65% of its
maximum load-bearing capacity. The axial strain had a reverse of 200 during the
loading process. This reverse resulted from the displacements measured by DT2 to
DT5 and had a significant increase from the point corresponding to 65% of its
maximum load-bearing capacity. During the loading process, when the load-bearing
capacity reached 78% of its maximum of capacity, the local buckling was first
observed at the top 25% region (part I) (Fig. 9d) and then relative local buckling was
observed from the lower 25% region (part III). With further loading, the local
sinewave, corresponding to the plateau period. The end of the plateau period was the
peak load-bearing capacity. Finally, the unloading process was carried out until the
load reached 0 and the residual deformation could be obtained to a certain degree. It is
worth noting that the residual displacements of DT2 to DT5 sustained low levels
and d. Fig. 6c shows the axial load-axial displacement curves measured by the two
DTs installed along the same longitudinal line. It illustrates both the curves following
a same path from 0 to 400 kN (60% of the maximum load-bearing capacity), entering
a slight deviation. The curve measured by DT1 had a larger increase than that of DT2.
This indicates that the axial deformation measured by DT1 had a 4 mm larger
maximum axial displacement than that of DT2. Additionally, Fig. 6d illustrates the
axial load-axial displacement curves measured by the six axial DTs installed around
the specimen at the top 25% region (part I) with a spacing of one-sixth of the
circumferential perimeter. The measurement ranges of these six DTs were from the
top of the specimen to the top of the bracket. Therefore, the axial displacements
towards the specimen (including east, northeast, north, northwest, west, southwest,
south and southeastward) could be obtained, which showed that all curves followed a
similar shape. Specifically, the curves followed the elastic axial load-displacement
relationships until 400 kN (60% of the maximum load-bearing capacity) and then the
curves branched out. With further loading, the load reached a maximum load-bearing
capacity of 662.1 kN and then entered the softening period until 522 kN (78% of the
the axial load reached 0 kN, in which the axial compressive deformation of the
specimen was almost reversed to the original state. In general, the maximum axial
displacements measured by DT3, DT4, DT5, DT6, DT7 and DT8 decreased, and the
The axial load-lateral displacement relationships of YT-1 and YT-2 are plotted in
Figs. 7a and b, respectively. Each figure shows eight curves measured by eight DTs as
well as the locations of all DTs installed. Moreover, each figure is divided into two
parts, including the right part illustrating the inward buckling and the left part
indicating the outward buckling. The curves in each figure have a similar shape.
When loading started, the curves followed an elastic path until they reached the
displacements was received. A softening period followed and occurred until 75% of
the maximum load-bearing capacity was obtained; this was followed by the unloading
process. The initial period of unloading had a lower lateral deformation before the
load decreased to 200 to 300 kN, then the lateral deformation started to reverse to the
original state before loading. During the unloading process, the decrease in lateral
which resulted from most of the lateral deformations concentrated on part II nearly
regaining their original condition. However, this stiffness was lower than the
shows similar curves to Fig. 7a, which developed the elastic periods until the axial
capacity. The unloading process followed and occurred until the load reached 0 kN.
All curves showed residual deformation to a certain degree, especially DT11 and
DT12, which measured significant lateral displacements in the inward buckling area.
Figs. 10a, b, c and d show the distance from the bottom of specimen YT-1 against
axial strain towards the specimen including east, north, west and southward surfaces,
respectively. According to the arrangements of the axial strain gauges attached on the
surface of YT-1, each curve was determined by five points corresponding to the axial
strains measured by five axial strain gauges at certain stress levels, which were
attached along the same longitudinal line and at distances of 410, 1245, 1500, 1755
and 2590 mm from the bottom of the specimen. All curves show a common
conclusion that the strain gauges attached at the distances of 1245 and 1755 mm
obtained larger maximum strains (2500 to 7500 respectively) compared to other
heights. The strains measured at 2590 mm show a significantly lower level. Compared
to the visual observations, hardly any deformations were observed. The strains
measured at 1500 mm indicate a much lower value than the strains measured by the
adjacent strain gauges. This is due to the two adjacent gauges corresponding to the
transition region between the two inward buckling regions. Therefore, much lower
axial strain values were obtained. Particularly, the strain values measured at 410 mm
were negative values, as shown in Fig. 10a, which resulted due to the tension in this
Similarly, Figs. 10e, f, g and h show the distance from the bottom of YT-2
against axial strain towards the specimen including east, north, west and southward
surfaces, respectively. The arrangement of the axial strain gauges was the same as
YT-1. From the curves shown in Figs. 10e, f, g and h, it can be concluded that the
maximum strain values measured from the strain gauges attached at 2590 mm (1800
to 7000 ) were larger than the strains obtained from the other heights. The
secondary large strain values were obtained from the strain gauges attached at 410
mm from the bottom. The minimum strain values were obtained from the middle of
YT-2, corresponding to 1245, 1500 and 1755 mm from the bottom of the specimen.
This resulted due to the existence of the joint. Therefore, the effects of strengthening
decreased the axial strain, especially at the distance of 1500 mm, where a significantly
lower level of axial strain was obtained. The strain values measured at 410 mm were
negative values (as shown in Fig. 10f), which resulted due to the tension in this part of
specimen.
In Figs. 10i, j, k and l, the distance from the bottom tip-axial strain relationships
of YT-3 towards the specimen including east, north, west and southward surfaces,
respectively, are shown. The arrangement of axial strain gauges at 1500, 1755 and
2590 mm were the same as YT-1 and YT-2. The results show that the values obtained
from the strain gauge attached at 1755 mm were the largest. Moreover, this maximum
axial strain value at 1755 mm was within the range of 8000 to 11000 on the north,
west and southward surfaces. However, the strain values at 1755 mm on the eastward
surface were negative, which indicates the tension in this part of the specimen. The
strain measured at 2590 mm was the second largest compared to the strain observed
elsewhere. The strain values measured at 1500 mm were almost close to 0 , as they
Radar graphs were adopted (Figs. 11) to express the variation of lateral strain at a
certain section. Each figure presents six curves, which represent the corresponding
strains to six strength levels, including 25, 50, 75 and 100% of the maximum axial
stress during the strength increasing period, and 80% of the maximum axial stress
during the strength decreasing period, as well as the ultimate strength. The ultimate
strength corresponds to the maximum lateral strain. The radar graph has four radiative
axes, which represent the lateral strain values towards the specimen, including the
northeast, northwest, southwest and southeast-facing surfaces. Each axis ranges from
negative strain values to positive values, which correspond to the outward and inward
buckling from visual observations.
Figs. 11a, b, c, d and e illustrate the variation of lateral strain of YT-1 at five
sections including 410, 1245, 1500, 1755 and 2590 mm from the top of the specimen,
respectively. In general, the maximum lateral strain was measured at 1245, 1500 and
1755 mm, which all correspond to the middle region (part II) of the specimen. The
maximum positive lateral strain of these three sections was within the range of 5900
to 6100 which was measured from the southwestward surface of the specimen.
The maximum negative lateral strain of these three sections was within the range of
-140 to -4600 and were obtained from the northeastward surface of the specimen.
Thus, the significant inward buckling mainly occurred on the southeast, southwest and
northwestward surfaces in the center (part II) of YT-1. The lower part (part III) of the
specimen are in tensile state, for that the strain values measured from the southeast,
southwest and northwestward surfaces were negative values (Fig. 11e). In addition,
significant lower strain values were obtained at 410 mm, which was in accord with the
visual observations. Moreover, Fig. 11a shows the inward buckling on the southeast
and northwestward surfaces, but the buckling is relatively lower than that of other
locations.
Figs. 11f, g, h, i and j show the variation of lateral strain of YT-2. The
arrangement of the lateral strain gauges was the same as specimen YT-1. Generally,
the maximum lateral strain was concentrated at the heights of 410 mm and 2590 mm,
which correspond to the top 25% region (part I) and lower 25% region (part III). The
maximum positive lateral strain was measured from the southeastward surface; the
values were within the range of 2500 to 4000 The strain was measured from the
southwest and northeastward surfaces at 410 mm and 2590 mm, respectively, and
show positive values; however, the values are relatively lower than that of the
surface at 410 mm and 2590 mm mainly show negative values, which occurred due to
the outward buckling at these locations. Apart from this, the strains measured at 1245,
1500 and 1755 mm show significantly lower values, which are consistent with the
visual observations that little lateral deformation was observed. Figs. 11 also shows a
trend of inward buckling on the southwest and northeastward surfaces; the maximum
The variation of lateral strain obtained from YT-3 is shown in Figs. 11k, m and
n; the arrangement was the same as described previously. From Fig. 11k, it can be
seen that the maximum lateral strain was 3800 on the northwestward surface and
the negative strain values on southwest and northeastward surfaces decreased with an
increase in load. The maximum lateral strain on the southeast and northeastward
surfaces were within the range of 4000 to 5000 at 1245 mm, which is in agreement
with visual observations of significant inward buckling. The strains observed at 1500
mm height show significantly lower values, all of which were below 150 and result
4.1 Stiffness
The initial compressive stiffness of GFRP chimney liners is a significant factor
affecting the mechanical behavior and the resistant ability of elastic deformation. The
chimney liner specimens reflect the effects of different fabrications on the initial
compressive stiffness. The measured stiffness from the experiments can be obtained
as the tangent value at the point when the axial load reached 10% of the maximum
load. The theoretically calculated stiffness can be achieved by the product of the
sectional area of the specimen and the compressive modulus obtained from the
results, where the theoretical results show the lowest stiffness value and a 10%
increase in YT-3, a 63% increase in YT-1, and a 166% increase in YT-2 compared
with the stiffness of the theoretical values. Results show that the stiffness of YT-3 is
the closest to the theoretical value, which resulted from YT-3 without stiffeners. In
addition, the 10% increase is due to the existence of the ring bracket. The stiffness
values of YT-2 and YT-3 illustrate a significant increase, which resulted from the
4.2 Comparisons
The comparison between YT-1 and YT-2 reflects the effects of the joint on the
shows that the locations of buckling were congregated mainly in the middle region
(part II) of YT-1. However, the existence of a joint caused the locations of buckling to
congregate in the top 25% region (part I). Therefore, the locations of buckling were
transferred, and the stiffness of the specimen increased due to the existence of the
joint. The maximum strength was also improved; the maximum strengths of YT-1 and
YT-2 were 13.5 MPa and 14.4 MPa, respectively, where a 7% strength increase was
achieved. Fig. 6a illustrates the strength of YT-1 after maximum strength enters a
softening period. The strength of YT-2 after the maximum strength slightly increased
is illustrated in Fig. 6b. Therefore, the existence of a ring joint is an effective method
illustrates the effects of a ring bracket on the mechanical behaviour of GFRP chimney
liners. The maximum strengths of YT-1 and YT-3 were both 13.5 MPa, which
illustrates that the existence of ring bracket barely affects the axial strength. The
effects of a ring joint on axial strength and the failure mode should be considered in
future design [30]. However, the effects of a bracket on axial strength can be ignored.
was used to model the mechanical behaviors of specimens using MSC software, Marc
2010. Three models were established, including models of YT-1, YT-2 and YT-3, to
band at a height of 100 mm, and a specimen with a ring bracket; these models
investigation, respectively.
In the modelling process, the size of the model was the same as the
from the top of the specimen. There were 1500 elements for the FRP chimney liner
tube and 480 elements for the outer stiffeners. The elements of these two different
materials shared the same nodes in the interface based on the full bond assumption.
The 4-node quadrilateral thick shell element (#75 shell) was used to simulate the FRP
tube and the 8-node hexahedral solid element (#7 solid) was used to simulate the
stiffeners. The material property of the model was determined according to the
experimental material property results and the values provided by the manufacturer
The boundary conditions for specimens YT-1 (Fig. 13a) and YT-2 (Fig. 13b)
included: (1) all nodes on the bottom section were fixed on all degrees of freedom; (2)
all nodes on the top section were fixed on all degrees of freedom, except for the
specimen YT-3 (Fig. 13c) included: (1) all nodes in the lower half-height region of the
specimen were fixed on all degrees of freedom; (2) the nodes on the top section were
fixed on all degrees of freedom, except for the displacement in the z-axis (longitudinal
displacement on one additional top node above the specimen, which is linked and tied
to all the nodes at the top section of the specimen, as shown in Figs. 13a, b and c. The
progressive failure analysis option was initiated with a residual stiffness coefficient of
0.01 to achieve a gradual degradation of the elastic modulus during the failure
In Figs. 14a, b, c and d, the failure modes of simulated specimens, which agree
well with the experimental specimens (Fig. 9a, b and c), are illustrated. All specimens
subjected to axial compression failed due to overall buckling, which developed from
local buckling. However, the major locations of buckling were different, which were
induced due to the existence of a joint for YT-2 and the bracket for YT-3. Figs. 14a
and b show the first and second mode of eigenvalue analysis for specimen YT-1,
which present the alternative inward and outward buckling between the two stiffeners
under different models. Fig. 14c indicates the first mode of eigenvalue analysis of
YT-2, which shows the alternative inward and outward buckling concentrated on the
upper and lower segments of the specimen due to the strengthening effects by the
joint band. Fig. 14d illustrates the first mode of eigenvalue analysis of YT-3, which
shows the inward buckling within the upper region of the specimen due to the
relationships between the simulated and experimental results, which agree well in
terms of the initial stiffness, peak load-bearing capacity and softening stiffness.
For further investigation on the effects of the ratio of diameter to thickness (D/t)
of FRP tubes on the mechanical behavior, a larger range of ratios (D/t) were chosen
and analysed, including 100, 200, 300, 400, 600, 700, 800, 900 and 1000, which
correspond to the FRP thicknesses of 28.0, 14.0, 9.3, 7.0, 4.7, 4.0, 3.5, 3.1 and 2.8 mm,
respectively. Results show that the failure modes were different for the specimens
with thickness lower than 3.5 mm (Fig. 16a) and higher than 14.0 mm (Fig. 16a),
which illustrates the inward and outward buckling of the entire surface in the
buckling as shown in the experimental tests (Figs. 9a and 14a). The axial
stress-displacement curves are plotted in Fig. 17, which show all specimens have the
same initial stiffness and increasing level of peak stress and corresponding strain with
results were plotted in Fig. 18, which shows an approximately linear relationship.
large range of load eccentricity values were chosen to examine the effects of
to radius of specimen sections at 0.1, 0.2, 0.4, 0.6 and 0.8, which correspond to load
eccentricities of 140, 280, 560, 840 and 1120 mm, respectively. The load eccentricity
was simulated by covering a rigid plate on top of the simulated specimen, as shown in
Figs. 19 a and b, and the load point was carried out at the desired node on the rigid
plate. From Fig. 20, results show the significant effects of load eccentricity on the
advised that high quality manufacturing of FRP tubes as well as the exact alignment
capacity.
5 Conclusions
This paper studied the mechanical behavior of GFRP chimney liners subjected to
axial compression with three scaled-down specimens. The failure modes, axial
variation of axial strain along the specimen height, and the variation of lateral strain
during the loading process, were determined. Additionally, the effects of ring
discussed. Finally, FEM analysis was performed using MSC software, Marc 2010, to
examine the failure modes and strength of the specimens. The method of design was
(1) The three chimney liner specimens showed similar mechanical behavior
when subjected to axial compression, which failed due to overall buckling that
developed from local buckling. The existence of the joint on specimen YT-2 between
two chimney liner parts was able to support enough axial strength and generate better
stiffness compared to specimen YT-1. The good service reliability of the ring bracket
to the visual observations. Results indicate that the axial deformation congregated at
regions with lower stiffness, such as the larger axial strain in the middle of specimen
YT-1. However, hardly any axial strain was measured at the region where the joint
(3) Compared to the theoretical results of stiffness, the existence of the ring
bracket increased the stiffness by 10%, and the existence of two stiffeners increased
the stiffness by 63%. The existence of both stiffeners and the joint increased the
stiffness by 166%.
ring joint in specimen YT-2. In addition, the locations of buckling were transferred to
the top 25% region from the one-half region of the specimen due to the existence of
the joint. Therefore, the effects of a ring joint on the axial strength and failure mode
should be considered in design. The effects of a bracket on the axial strength can be
ignored due to the slight difference in strength and failure mode found between YT-1
and YT-3.
(5) FEM analysis was used to examine the reliability of the experimental
investigation; the results of the FEM analysis agree well with the results of the
Different ranges of diameter to thickness ratios and load eccentricity values were
regular segment is competent to bear the axial compressive stress of 13.5 MPa.
(2) The round stiffeners could effectively avoid premature buckling of the
chimney liner. In practical applications, the height of each chimney liner segment is
around 9 m, in which two stiffeners are advised to be installed on each segment at one
third and two thirds distance from the top of the chimney liner.
(3) The method to connect different GFRP chimney liner segments proposed
in this study could strengthen the chimney liner. In practical applications, the
thickness of the GFRP sheet used to connect two segments is generally more than 2
(4) The ring bracket proposed in this study could effectively provide the
support needed to the chimney liner. The support stress is larger than the buckling
stress of a GFRP chimney liner (13.5 MPa). In practical applications, the ring bracket
can be scaled up or down in design according to the practical size of the chimney
liner.
ACKNOWLEDGMENTS
This research was supported by the National Key Research and Development
Foundation of China, China (Nos. 51708330, 51522807 and 51661165016) and the
[1] Chambers RE. ASCE design standard for pultruded fiber-reinforced-plastic (FRP)
[2] Chen L, Dick WA, Nelson S. Flue gas desulfurization by-products additions to
2001;114(2): 161-168.
2004;43(21): 6714-6722.
[4] Bigham JM, Kost DA, Stehouwer RC, Beeghly JH. Mineralogical and engineering
[5] Feng P, Wang J, Wang Y, Loughery D, Niu DT. Effects of corrosive environments
[6] Kelley D. The use of FRP in FGD applications. Reinf Plast. 2005;51(1): 14-19.
2005;49(10): 20-25.
[8] Kelley, D. H. 1994. “Pollution control of wet flue gas desulfurization (FGD)
[9] Taillemite S, Pauer R. Bright future for vinyl ester resins in corrosion applications.
U.S. Patent 4,222,929, Washington, DC: U.S. Patent and Trademark Office. 1980.
[11] Sreekala MS, George J, Kumaran MG, Thomas S. The mechanical performance
steel tube with FRP-confined concrete core subjected to axial compression. Compos.
concrete-filled square steel tube with FRP-confined concrete core. J. Compos. Constr.
2018;22(4): 04018015.
[15] Bank LC. Progressive failure and ductility of FRP composites for construction:
[16] Banks J. The performance record of liners for power chimneys using
[17] Zarghamee MS, Brainerd ML, Tigue DB. On Thermal Blistering of FRP
composites for civil engineering structures. Constr Build Mater. 2003;17(6): 405-437.
chimney liners subjected to axial tension. Compos. Part B. Eng. 2019;177: 107411.
1966;3(3): 242-251.
cylindrical shells using a wave propagation method. Shock. Vib. 2001;8(2): 71-84.
2011;doi: 10.1177/0892705711411342.
[25] Häg N, Harrison P. Experience with the use of Derakane vinyl ester-based GRP
properties. 2005.
[28] GB/T 1449-2005. Fiber-reinforced plastics composites-Determination of flexural
properties. 2005.
[30] ASTM D 5364. Standard guide for design, fabrication, and erection of fiberglass
Figure Captions
Fig. 1 Cylindrical GFRP chimney liner: (a) GFRP chimney liner and (b) configuration of
Fig. 3 Cylindrical GFRP chimney liner specimens: (a) YT-1, (b) YT-2, (c) YT-3, and (d)
Fig. 4 Set-up and instrumentation: (a) YT-1, (b) YT-2, and (c) YT-3.
subjected to axial compression: (a) DT1–DT5 measured for the YT-1 specimen, (b)
DT1–DT5 measured for the YT-2 specimen, (c) DT1 and DT2 measured for the YT-3
subjected to axial compression: (a) DT6–DT13 measured for the YT-1 specimen and
Fig. 8 Steel accessory equipment: (a) steel I-beams for loading, (b) inner steel clamps,
(c) outer steel clamps, and (d) loading locations.
Fig. 9 Failure modes of specimens: (a) failure mode of YT-1 under loading, (b) failure
mode of YT-1 after unloading, (c) failure mode of YT-2 under loading, (d) failure
mode of upper quarter region of YT-2, (e) failure mode of YT-3 under loading, and (f)
Fig. 10 Variations of the axial strains of the specimens from different towards: (a)
east view of YT-1, (b) north view of YT-1, (c) west view of YT-1, (d) south view of YT-1,
(e) east view of YT-2, (f) north view of YT-2, (g) west view of YT-2, (h) south view of
YT-2, (i) east view of YT-3, (j) north view of YT-3, (k) west view of YT-3, and (l) south
view of YT-3.
Fig. 11 Variations of lateral strains of certain sections at different distances from the
top tips of the specimens: (a) at 410 mm of YT-1, (b) at 1245 mm of YT-1, (c) at 1500
mm of YT-1, (d) at 1755 mm of YT-1, (e) at 2590 mm of YT-1, (f) at 410 mm of YT-2,
(g) at 1245 mm of YT-2, (h) at 1500 mm of YT-2, (i) at 1755 mm of YT-2, (j) at 2590
mm of YT-2, (k) at 410 mm of YT-3, (m) at 1245 mm of YT-3, and (n) at 1500 mm of
YT-3.
stiffnesses.
Fig. 13 Finite element modeling of stiffened GFRP chimney liner: (a) YT-1, (b) YT-2,
Fig. 14 FEM analysis of stiffened GFRP chimney liner: (a) first mode of eigenvalue
analysis for YT-1, (b) second mode of eigenvalue analysis for YT-1, (c) first mode of
eigenvalue analysis for YT-2, and (d) first mode of eigenvalue analysis for YT-3.
Fig. 15 Simulated axial load–displacement relationships: (a) YT-1, (b) YT-2, and (c)
YT-3.
Fig. 16 Failure modes of simulated stiffened GFRP chimney liners with different D/t
ratios: (a) specimen with lower level of FRP thickness and (b) specimen with high
D/t ratios.
Fig. 19 FEM analysis of the effects of the load eccentricity on the mechanical
behavior: (a) model for load eccentricity and (b) failure mode under load
eccentricity.
Conflict of Interest
Authors: Shi Cheng, Peng Feng, Xinmiao Meng, Zhiyuan Li, Jike
Du