Académique Documents
Professionnel Documents
Culture Documents
Abstract: The lack of a comprehensive, validated, and easily accessible data base for the durability of fiber-reinforced polymer 共FRP兲
composites as related to civil infrastructure applications has been identified as a critical barrier to widespread acceptance of these materials
by structural designers and civil engineers. This concern is emphasized since the structures of interest are primarily load bearing and are
expected to remain in service over extended periods of time without significant inspection or maintenance. This paper presents a synopsis
of a gap analysis study undertaken under the aegis of the Civil Engineering Research Foundation and the Federal Highway Administration
to identify and prioritize critical gaps in durability data. The study focuses on the use of FRP in internal reinforcement, external
strengthening, seismic retrofit, bridge decks, structural profiles, and panels. Environments of interest are moisture/solution, alkalinity,
creep/relaxation, fatigue, fire, thermal effects 共including freeze-thaw兲, and ultraviolet exposure.
DOI: 10.1061/共ASCE兲1090-0268共2003兲7:3共238兲
CE Database subject headings: Durability; Fiber reinforced polymers; Reinforcement; Retrofitting; Decks; Panels; Composite
materials.
Introduction durability is sparse, not well documented, and/or not easily acces-
sible to the civil engineer. In addition, there is a wealth of con-
Fiber-reinforced polymer 共FRP兲 composites are increasingly tradictory data published in a variety of venues that confuses the
being used in civil infrastructure in applications ranging from practicing engineer.
reinforcing rods and tendons, wraps for seismic retrofit of col- The lack of a comprehensive database on FRP materials makes
umns and externally bonded reinforcement for strengthening of it difficult for the practicing civil engineer and designer to use
walls, beams, and slabs, to all-composite bridge decks, and even FRP composites on a routine basis. Although a number of reviews
hybrid 共FRP composite in combination with conventional materi- have been published recently related to durability and test meth-
als兲 and all-composite structural systems. Anecdotal evidence ods 共Schutte 1994; Bank et al. 1995; Chin 1996; Liao et al. 1998兲
provides substantial reason to believe that, if appropriately de- the focus of each has been to summarize the state of knowledge in
signed and fabricated, FRP composite materials can provide general without emphasizing or attempting to prioritize critical
longer lifetimes and lower maintenance than equivalent structures areas in which needs are the greatest for collection, assimilation,
fabricated from conventional materials. However, actual data on and dissemination of data. This paper summarizes the results of
such an effort in detailing critical gaps in knowledge vis-à-vis the
1
Dept. of Structural Engineering, Univ. of California, San Diego, La durability of FRP composites to be used in civil infrastructure
Jolla, CA 92093-0085. E-mail: vkarbhari@ucsd.edu applications.
2
National Institute of Standards and Technology, Gaithersburg,
MD 20899.
3
National Institute of Standards and Technology, Gaithersburg, Scope and Background of Study
MD 20899.
4
Dept. of Civil Engineering, Univ. of Sherbrooke, Sherbrooke PQ,
Based on a series of workshops, held under the aegis of the Civil
Canada J1K 2R1.
5
Newport News Shipyard, Newport News, VA 23603. Engineering Research Foundation 共CERF兲 and the Market Devel-
6
Dept. of Mechanical Engineering, Texas A&M Univ., College opment Alliance 共MDA兲, with attendees from the user, owner,
Station, TX 77843-3123. construction, design, materials, and manufacturing sectors, the
7
Engineering Science and Mechanics Dept., Virginia Tech., use of FRP composites in the form of reinforcing bars 共rebar兲,
Blacksburg, VA 24061-0219. external reinforcement for strengthening on concrete structures,
8
Naval Surface Warfare Center, Carderock Division, West Bethesda, wraps, and jackets for the seismic retrofit of columns, piers and
MD 20817-5700. masonry walls, bridge decks and bridge systems, and wall panels
9
Civil Engineering Research Foundation, 2131 K Street NW, Suite and structural profiles, were identified as being of the maximum
700, Washington, DC 20037-1810. near-time market and application interest. Seven separate environ-
Note. Discussion open until January 1, 2004. Separate discussions
mental conditions, moisture/solution, alkali, thermal 共including
must be submitted for individual papers. To extend the closing date by
one month, a written request must be filed with the ASCE Managing temperature cycling and freeze-thaw兲, creep and relaxation, fa-
Editor. The manuscript for this paper was submitted for review and pos- tigue, ultraviolet, and fire, were then identified as being of pri-
sible publication on January 7, 2002; approved on May 3, 2002. This mary importance vis-à-vis durability of FRP composites in the
paper is part of the Journal of Composites for Construction, Vol. 7, No. application areas already highlighted. Synergistic effects 共i.e., ef-
3, August 1, 2003. ©ASCE, ISSN 1090-0268/2003/3-238 –247/$18.00. fects resulting from the combination of multiple environmental
• Effects of solution ions/salts in conjunction with stress and Katsuki and Uomoto 共1995兲, GangaRao and Vijay 共1997兲, Porter
temperature regimes need to be evaluated at the level of the and Barnes 共1998兲, Chin et al. 共1999, 2001兲, and Zhang and Kar-
fiber, interphase, matrix, and composite, bhari 共1999兲.
• It is emphasized that testing over short periods of time fol- The results of the gap analysis based on the synthesis of data
lowed by extrapolation of results, especially for ambient tem- pertaining to the importance of data, and its availability in easily
perature cure systems, can lead to erroneous results, accessible, validated form, are presented in Table 4. Based on the
• Due to the effect of moisture on glass transition temperature, current state of knowledge, albeit incomplete, the following over-
composites must be cured such that the T g achieved is signifi- all aspects are highlighted:
cantly higher than the maximum service temperature 共a mini- • The use of mechanistic, and/or reliability based tools inte-
mum level of 30°F above maximum service temperature is grated with a risk-assessment methodology needs to be devel-
recommended for wet T g ), and oped for life-prediction,
• Taking into account effects of degradation and damage toler- • Since the polymeric resin plays a critical role in protecting the
ance requirements and the lack of sufficient data at present, the fiber and slowing the diffusion process, preference should be
stress level in the composite should be limited under sustained given to the use of appropriate epoxies and vinylesters. The
factored loads at less than 25% of guaranteed design strength use of polyester resins is not recommended,
for glass fiber-reinforced polymers 共GFRP兲, 30% for aramid • In order to decrease the possibility of rapid movement of
fiber-reinforced polymers 共AFRP兲, and 40% of guaranteed de- moisture and alkaline salts into the bulk composite, and to-
sign strength for carbon fiber-reinforced polymers 共CFRP兲. wards the fiber surface, it is critical that an appropriate thick-
ness of resin rich surface exist in FRP composites used in this
environment, with the resin layer remaining uncracked
Alkali Effects through the period of intended use,
• Acknowledging the role of undercure on increase in moisture
Although FRP composites can come in contact with alkaline susceptibility of resins, it is recommended that the resin and
media through interaction with a variety of sources, including FRP composite be fully cured prior to use in the field, and
alkaline chemicals, soil 共or solutions diffusing through soil兲, and • Taking into account effects of degradation and damage toler-
concrete, the main concern at the present time stems from the ance, the stress level in the FRP reinforcement should be lim-
potential effects of degradation due to concrete pore water solu- ited under sustained factored loads to less than 30% of guar-
tion, which is known to have a hydrogen ion concentration level anteed design strength for GFRP and AFRP, and 40% of
as high as 13.5. A large body of research exists on the degradation guaranteed design strength for CFRP.
of bare glass fibers in contact with 共or in兲 alkaline solutions, es-
pecially those derived from concrete, and there is no doubt that
bare glass fibers in this environment are severely degraded due to Thermal Effects
a combination of mechanisms ranging from pitting, hydroxyla-
tion, hydrolysis, and leaching. Although the presence of resins in Thermal effects considered in this section include response
FRP composites around individual filaments can be expected to changes due to temperatures above the cure temperature, freezing
protect the fibers from such attack, the alkaline solutions can ac- and freeze-thaw conditions, and temperature variations and
celerate the degradation of bond and of some resins themselves, cycles. It is acknowledged at the outset that not all thermal expo-
especially if not fully cured. A good summary of mechanisms and sure is deleterious since in a number of cases it can actually result
effects can be found in Sen et al. 共1993兲, Bank et al. 共1995兲, in much needed postcure of FRP components. It is noted that
resins and adhesives soften over a temperature range, which • FRP composites should not be used at temperatures above
causes an increase in viscoelastic response, a consequent reduc- their glass transition temperatures, and for purposes of design
tion in elastic mechanical performance levels, and, in a number of it is recommended that materials be chosen that have a T g at
cases, an increased susceptibility to moisture absorption. Prior least 30° above the maximum expected use temperature,
research, materials testing, and anecdotal data has shown that in • The synergistic effects of moisture/solution and thermal effects
general: need to be considered further with investigations into both
• Subzero temperature exposure can result in matrix hardening, materials and structural systems response changes, and
matrix microcracking, and fiber-matrix bond degradation, • Long-term effects of differences in coefficients of thermal ex-
• Freeze-thaw in the presence of salt can result in accelerated pansion and elastic properties of bonded materials need to be
degradation due to the formation and expansion of salt depos- considered. Effects of the presence of large heat sinks as with
its in addition to effects of moisture induced swelling and dry- concrete need to be assessed both for in-process and postpro-
ing, cess effects.
• Exposure to temperature above that of processing can result in
an initial postcure followed by degradation due to thermal ef-
fects, and CreepÕRelaxation
• The coefficients of thermal expansion of adhesives can be or-
It has been well established that aramid and glass fibers have a
ders of magnitude different from those of bulk resins and/or
higher level of susceptibility to creep rupture at lower stress lev-
composites, and hence thermal gradients/exposure can cause
els than carbon fibers, with carbon fibers exhibiting little to no
premature debonding along the FRP composite-adhesive-
chemical-induced strength degradation, but having a larger spread
concrete interfaces.
in median failure times under stress rupture conditions as shown
Results of infrastructure related testing and reviews of appli-
in Table 6 共Chiao and Moore 1971; Chiao et al. 1972; Moore
cations can be found in Lord and Dutta 共1988兲, Dutta 共1988兲,
et al. 1974兲. Qualifications for this data are that aramid and glass
Gomez and Casto 共1996兲, Karbhari and Engineer 共1996兲, Green
fibers 共Bentur et al. 1985兲 are very susceptible to alkali-induced,
et al. 共2000兲, and Miyano et al. 共1999兲. The results of the gap
chemical-induced
analysis for both elevated temperature conditions and freeze/
freeze-thaw conditions are shown in Table 5. Aspects that need to
be emphasized include: Table 6. Fiber Stress Rupture Level for 10% Population Failures
• The greatest concern with temperature effects on composite After 75 Years Continuous Stress Exposure Under Ambient
structures in civil engineering applications is that freeze/thaw Conditions
conditions can potentially result in debonding of laminates, 10% Failure probability
• Failure is also possible if the laminating resin or adhesive stress rupture level Spread in median time
softens excessively. The upper use temperature, the Material Fiber type after 75 years 共%兲 to fail in decades
Operational Limit, of a given laminating resin defined as the
Carbon 75 6
temperature at which the flexural strength decreases to half the
Aramid 60 3
room temperature value, needs to be clearly followed for field
Glass 50 2.5
implementation,
strength degradation, which, over long periods, will generally pre- • Significant testing and characterization needs to be conducted
dominate over any fiber stress rupture attributes. Carbon fibers for material systems cured under ambient conditions taking
exhibit no chemical-induced strength degradation but exhibit a into account the synergistic effects of moisture, stress levels,
much larger spread in median failure times than aramid or glass and temperature,
fibers. • Effects of shearing deformations which are often erroneously
For any practical civil infrastructure composite, the creep- neglected need to be accounted for in the prediction of time-
stress relaxation properties are dominated by the resin matrix de- dependent behavior, and
pendent properties, rather than fiber or interfacial properties. Un- • The use of standardized materials forms as in metals is highly
dercured resins are susceptible to significant creep and possible recommended.
microcrack initiation during the early stages of service environ-
ment exposure. Absorbed moisture and higher service environ-
ment exposure temperatures both enhance creep susceptibility, Fatigue
which is ultimately diminished by enhanced further cure. There
has been a study on ambient cured amine epoxies that shows the Fatigue, which is generally defined as the physical phenomenon
unreacted epoxy monomers can recrystallize in the undercured that causes a material or component to fail after the application of
resin, causing a permanent brittle resin, unless such crystals are an applied condition or conditions 共cycles兲 even though the level
melted at 42.5°C 共Morgan and O’Neal 1978兲. There has been one of that condition is not high enough to cause failure on the first
study of the long-term creep behavior of vinyl and polyesters as a cycle of application, is an important consideration for the dura-
function of cure conditions using flexural creep tests at ambient bility and safety of civil infrastructure. The loading may be me-
temperature 共Bradley et al. 1998兲. Vinylesters that have been chanical 共due to vehicle traffic, for example兲, thermal 共from varia-
cured at room temperature had a greater creep exponent 共n ap- tions in temperature兲, or chemical 共from seasonal road treatments,
proximately 0.20兲 for power law creep (t n ) than vinylesters that oxidation, NOX effects, water, etc.兲. Significant research is
were postcured to cross link completion at 93°C 共n approximately needed to develop a comprehensive understanding of the pro-
0.12兲. The total creep compliance as well as the time exponent n cesses and mechanisms associated with fatigue failure in civil
decreased systematically with increasing cure condition and time, infrastructure components fabricated of FRP composites. Good
with a creep compliance for room temperature cure for one day reviews of the phenomena are given by Mandell 共1982兲, Liao
that is 250% more than that for a neat vinylester cured for four et al. 共1998兲, and Konur and Matthews 共1989兲, with the effect of
hours at 93°C. environment; constituent materials and processes of fabrication
In most cases, at the level of a structure or component, creep being elucidated in Mandell et al. 共1985兲, Branco et al. 共1995兲,
and stress relaxation can be guarded against or reduced signifi- and McBagonluri et al. 共2000兲. The results of the gap analysis are
cantly by taking advantage of the fact that creep and stress relax- shown in Table 8 for various environmental conditions.
ation response is likely to be resin dominated for most practical Data are currently available for a limited set of ‘‘fatigue’’ con-
civil infrastructure applications. Thus appropriate selection and ditions, most notably constant amplitude fatigue at frequencies
processing of resins, and the designed placement of fibers can ranging from 1 to 10 Hz. Based on the work of Mandell 共1982兲
solve a large part of the challenge. Readers are referred to the and McBagonluri et al. 共2000兲, the slope of the S-N curve for
excellent reviews by Liao et al. 共1998兲 and Scott et al. 共1995兲 for most planar glass/polymer composites can be expected to range
further explanations. The results of the gap analysis are shown in from 10–12% ultimate tensile strength 共UTS兲/decade of life.
Table 7. Based on the current state of knowledge at the materials There are existing life prediction methodologies such as devel-
and structural levels the following aspects are highlighted: oped by Reifsnider 共1991兲 that can be used to combine processes,
their sequence, and severity to estimate life under known condi- of the FRP from direct ultraviolet exposure. However, it must be
tions. However, again there is a lack of understanding to accu- noted that the use of a polymeric protective coating does not
rately accelerate the individual and combined service conditions prevent UV-induced damage from occurring, but serves as a
and environments to appropriately assess in a short period of time ‘‘self-sacrificing’’ layer to prevent the FRP surface being directly
if a material system or structure will survive an extended service exposed to UV radiation. The protective coating itself will even-
period. tually be degraded by UV radiation and will need to be main-
tained. The most deleterious effects of UV exposure are probably
not due to the actual UV damage, which is limited to the surface,
Ultraviolet „UV… Radiation
Ultraviolet radiation that reaches the Earth’s surface comprises Table 9. Overall Ranking of Gaps for Ultraviolet Effects
about 6% of the total solar radiant flux and has wavelengths be- Structural Aesthetic
tween 290 and 400 nm. Since most polymers have bond dissocia- concerns concerns
tion energies on the order of the 290 to 400 nm wavelengths in Application area Aa Bb Cc A B C
the ultraviolet region, they are greatly affected by exposure to this
portion of the solar spectrum. The effects of ultraviolet 共UV兲 Internal reinforcement
exposure, or photodegradation, are usually confined to the top few Rebar Xf X X X X X
microns of the surface. However, in some cases, degradation at External reinforcement
the surface of a polymeric component has been shown to affect Beamsd X X X X X X
mechanical properties disproportionately, as flaws that result from Slabsd X X X X X X
surface photodegradation can serve as stress concentrators and Columns 10 X X 2 X X
initiate fracture at stress levels much lower than those for unex- Seismic retrofit
posed specimens. The effect of ultraviolet radiation is also com- Columns, piers 10 X X 2 X X
pounded by the action of temperature, moisture, wind-borne abra- Wallse 10 X X 2 X X
sives, freeze-thaw, and other environmental components. A Deck systems
survey of the literature pertaining to outdoor exposure and accel- Conventional beams/girders 8 X X 2 X X
erated laboratory tests of polymers indicates that a great deal of Integral/composite beams/girders 8 X X 2 X X
variability and uncertainty is associated with the current testing Structural elements
methodology. Furthermore, it is not always clear if changes in Wall panels, profilese 10 X X 2 X X
laminate mechanical properties can be attributed solely to UV
Note: A ranking of 1 indicates a low level of criticality, whereas a ranking
effects or to a combination of UV and moisture. The overall rank-
of 10 represents a high level of criticality.
ing of gaps for UV effects is given in Table 9. The effects of UV a
A: Effects at the fiber-reinforced polymers composite level.
exposure on composites can be reviewed in Roylance and Roy- b
lance 共1976兲, Trabocco and Stander 共1976兲, George et al. 共1997兲, B: Effects at the interface/bond/adhesive level.
c
Lucki et al. 共1981兲, Chin et al. 共1997兲, Monney et al. 共1998兲, and C: Effects at the substrate level.
d
Liau and Tseng 共1998兲. Ultraviolet assumed not to be a factor on the nonexposed part.
e
A common practice in outdoor applications of FRP composites In exterior applications only.
f
is to use a gel coat or other protective coating to shield the surface X: Not applicable.
but to the potential for increased moisture ingress in the damaged Milke and Vizzini 共1993兲, Ohlemiller and Cleary 共1999兲, Scud-
regions. It is recommended that future research in UV radiation amore 共1994兲, Dao and Asaro 共1999兲, and Ohlemiller and Shields
effects on composites be conducted in the following areas: 共1999兲.
• Development of polymeric matrix resins that have inherent
stability to UV radiation, which would allow for the elimina- Summary and Recommendations
tion of the protective coating, Based on the gap analysis conducted for each of the selected
• During UV exposure, specimens should be subjected to me- environmental conditions, it appears that there is a substantial
chanical loads similar to the type that they will encounter in commonality of needs, which provides for the selection of a set of
service, and data/research requirements that is critical to the generic imple-
• New UV testing methodologies based on reliability theory and mentation of FRP composites in civil infrastructure. These needs,
the idea of effective dosage, already being implemented in the in no particular order of priority 共since it is difficult to transition
polymer coatings field, should be extended to FRP materials. or compare the level of need within one category of environment
to that in another兲, are as follows:
• Collection, assessment, and appropriate documentation of
Fire available data in a form useable by the civil engineer/designer,
• Testing over extended 共18⫹ months兲 time periods. Tests con-
A significant concern in any application of organic matrix-based ducted over short time periods 共less than 18 months兲 can yield
composites is the possibility that an accidental 共or deliberate兲 fire misleading results due to effects of postcure and slow inter-
may ignite the composite material. This may result in the spread phase and fiber level degradation, and can provide an errone-
of flame on the composite surface, and may also release heat and ous level of comfort in some cases,
generate potentially toxic smoke. The fibers displace polymer • Testing under combined conditions 共stress, moisture, solution,
resin, making less fuel available to the fire. When the outermost temperature, and/or other regimes兲 at both the materials and
layers of a composite lose their resin due to heat-induced gasifi- structural levels is critical,
cation, they act as an insulating layer, slowing heat penetration • Assessment and characterization of the effects of incomplete
and evolution of gases from the depth of the composite. It was cure and undercure, especially for ambient temperature cure
recognized that fire-related issues associated with composite ma- systems, are essential,
terials are more severe in confined spaces 共such as tunnels and • Development of standardized solutions and conditions for
buildings兲 as opposed to open spaces 共such as roads and bridges兲. laboratory studies that closely simulate actual field conditions,
As such, the gap analysis was conducted on two bases: 共1兲 fires in and
open spaces and 共2兲 fires in confined spaces. The effect of fire is • Development of appropriate resin systems, gel coats, and coat-
initially exhibited by the heating up of the composite surface. ings that would serve as protective layers for the bulk compos-
Over the depth of composite material heated up to temperatures ite against external influences including environmental condi-
past the glass transition temperature, the composite exhibits a tions, intended, and accidental damage.
corresponding loss of modulus. Below the temperature of chemi- Based on the results of the gap analysis conducted through the
cal degradation, this loss in modulus is reversible. Further in- present study, and on the overall results of the investigation
crease in temperature, such as above 450°F for glass/vinylester, 共through review of literature, discussions with experts in the area
results in the degradation of the chemical structure of the resin. of durability, results of discussions of the user and supplier pan-
This thermal damage results in irreversible loss in load bearing els, and subsequent discussions with members of the FRP com-
characteristics. The overall ranking of gaps for fire effects is posites and civil engineering industries兲 a three-pronged approach
given in Table 10. A representative set of data, as an overview, is recommended for future activities in continuation of this study
can be gleaned from Sorathia et al. 共1997兲, Sorathia et al. 共1993兲, as described in the following.