Вы находитесь на странице: 1из 10

COMPOSITE PRODUCTS RUPTURE UNDER LONG-

TERM LOADS: A TECHNOLOGY ASSESSMENT


In: Proceedings, 22d International particleboard/composite materials
symposium; 1988 March 22-24; Pullman, WA. Pullman, WA: Washington
State University; 1988: 247-256.

THEODORE L. LAUFENBERG
U.S. Forest Products Laboratory
Madison, WI

ABSTRACT structural particleboards, flakeboards, and


plywood. Some common baselines are offered
Information on wood-based structural which promote some consistency in the testing
composite panels performance under long-term and analysis of creep-rupture for composite
loads is currently needed to permit their use in products. The closing of this overview paper
engineered assemblies and systems. A broad is dedicated to assessment of the research needs
assessment of these time-dependent properties for this subject area.
is critical for creating databases and models of
the creep-rupture phenomenon which permit
application of reliability-based design
procedures. This paper provides a background INTRODUCTION
of information on the creep-rupture phenomena
by describing the microscopic mechanisms One of the better-known design aspects
presumed to control the process, the history of of solid wood products in that wood strength is
the treatment of creep-rupture in design of dependent on its time under load (Wood, 1947).
wood structures, the models commonly used Many scientists have studied this problem
to describe the creep-rupture time-to-failure phenomena and have introduced many diverse
behavior, and an overview of the literature for methods of modeling it. Another aspect of this

247
load-duration effect is the time-dependent of the macro-mechanical behavior and the
deformation caused by loads on structural responsible micro-structural mechanisms for
members which is commonly called creep. creep-rupture. From there, an overview will
Throughout this paper. the term “creep-rupture” then be presented of the history of the design
will he used to describe the combined factors used to account for creep-rupture in the
phenomena of increasing deformation and loss wood industry and past research work on creep-
of strength with time. rupture time-to-failure for composite products.
This will be followed by suggestions for
Composite wood products have similar standardizing the tests and data generated by
characteristics as solid wood products in regard these time-comsuming experimental programs.
to the creep-rupture phenomena. However. The paper closes with an assessment of research
being relatively new to structural uses, a long- needs for addressing the creep-rupture problem
term performance history based on common in light of the continued growth and structural
building practices has not been established for expectations for these composite wood
these products. Moreover, their rheological products.
properties present the creep phenomenon as
the more common concern than that of rupture,
as is true for solid wood products. WHAT IS CREEP-RUPTURE?

Composite wood products produced from The phenomena of creep-rupture has some
veneers, flakes, wafers, and particles are very humble beginnings and should be viewed
becoming accepted in structural systems that in this way prior to any analytical assessment.
may have traditionally employed solid wood From a simplistic standpoint. one sees a
or steel materials. A reason for this diversity material under a constant load with its
in product application is that these composites deformation increasing over time. This is
can be produced with a wide range of structural commonly called creep and was the subject of
properties. Manufacturers can control the a presentation last year in the Proceedings
production process and thus, the materials’ (1987). As this deformation increases, it will
short-term strength and stiffness and the at some point be seen as a “failure.” This may
variability of these properties. take the form of a massive crack in a beam
member, a buckled truss component, or simply
All wood-based products are susceptible a floor panel that has an excessive deflection.
to the creep-rupture effects noted previously. This failure or rupture under long-term loads
Thus, in-service stiffnesses must be known is the subject of this paper.
and controlled to limit the component’s creep
deformation and the possibility of rupture in Since it was the writer who introduced the
its intended service life. The relative lack of combined phenomena of duration of load and
information on creep-rupture in composite creep which he’s called creep-rupture, it will
materials hinders development of empirical be illustrated why it is correct to combine
design guidelines, but also precludes them and view this as a continuous process.
development of a reliability-based assessment. There are three basic stages in the creep-rupture
The writer’s view of a reliability-based design process: primary, secondary, and tertiary creep
approach has a format which sets deflection as distinguished by the decreasing, constant,
limitations that will address serviceability and increasing rates of deformation (Figure 1).
(deflection. in this care) concerns and time-to- These stages occur in many materials including
failure load simultaneously. This approach wood-based composites (Nielsen, 1972).
will provide the emphasis to assess deflection
as the controlling influence in design of Some obvious implications from this
traditional panel products (O’Halloran et al., simple presentation are that: (a) the primary
1988). creep is a short transition between the elastic
[or initial] deformation caused by the load and
In this paper, a background on the creep- the steady-state condition represented by the
rupture phenomena for composite materials secondary phase, (b) the secondary creep is of
will be provided. First, will be the description widely varying time length and at low loads

248
Figure 1. --Idealized deformation (∂) versus time curves
for three stress (τ) levels in a material. Only the τ − τ3
curve shows all three stages of the creep process which
ends with an excessive deformation. Deformation is
represented with two components, elastic (∂o) and creep
(∂o )

could be infinite [or nearly so], and (c) the some wood fibers and adhesive bonds that are
tertiary phase of creep is short-lived and redistributed within the microstructure. This
represents the onset of massive fracture of the process of micro-failures and micro-creep
material. followed by stress redistribution is presumed
To justify this view of creep-rupture as to be occurring at the general rates indicated
a continuous phenomena, an insight is needed by the three creep phases indicated earlier.
into the microscopic structure. For most wood- Clearly, there are differing mechanisms for
based composites, the stresses carried within these microscopic processes, but the scenario,
the material are dependent on such local factors as presented, stays basically the same.
as fiber orientation, density, adhesives type Also, there are environmental conditions
and content, etc. Due to the construction of which confound the creep-rupture process. In
these composites, some constituents are simple terms, the changing of temperatures
carrying larger stresses than others. These and moisture contents cause the deformation
highly stressed fibers are the first to unload rate to increase as well. That rate is dependent
through such mechanisms as disbonding, fiber on the material itself, the load level, material
failure, and fiber rotation. These micro-failures volume, and the rate of environmental change.
cause infinitesimal increases in overall material Each of these factors results in a gradient of
deformation and cause the remaining fibers to structural properties which change the
be more highly stressed. In addition to these deformation and failure characteristics of the
micro-failures, it is expected that wood fibers material under load. Considering the
and adhesive bonds under stress will exhibit complexity of the environmental stresses placed
micro-creep. These continuously increasing on composite products in-service, this may
deformations represent a stress relaxation of well be the most difficult effect to quantify.

249
HISTORY OF DESIGN FACTORS During the 1940s, a comprehensive study
FOR CREEP-RUPTURE was undertaken by Wood (1947) of the effect
on sustained bending loads on strength of clear
A s e a r l y a s about 1840, Haupt wood specimens. The results of that wood
recognized that long-term loads affect the (Figure 2) indicated that if the S-minute
bending strength of wood (Thurston, 1881). duration corresponded to 100% of the standard
With additional experimental evidence from test strength, then the 9/16 factor (56%) was
Tiemann (1909), a “duration of load” factor achieved at 27 years. Further analysis of the
evolved to assure that design loads stayed below data (Wood, 1951) provided a curve that fit
the elastic limit. This factor, 9/16, was Wood’s rapid (ramp) loading as well as long-
recognized as the ratio of the elastic limit to time (constant) load failures in this hyperbolic
the modulus of rupture for wood (ASTM, 1915) form:
for “longtime loading.” No clear definition
was given for long-time loading until Combs
(1939) extrapolated existing data to give 10
years as a value for “normal loading” as it is where S = percentage of 5 minute test strength
known today. D = duration to failure in seconds

Figure 2.--Time-to-failure data for clear Douglas fir under constant loads as presented
by Wood (1947)

250
The adoption of load duration factors CREEP-RUPTURE TIME-TO-
based on “normal” loading first appeared in FAILURE MODELING
the National Design Specification (NDS) in
1948. After several years of experience and Creep rupture is a continuous deformation
readjustment, the 1952 NDS (National Forest phenomenon which may end in catastrophic
Products Association, 1986) fully embraced failure of the material under stress. Past
Wood’s hyperbolic curve and retained its use experience has provided two basic behaviors
to the present time (Figure 3). This hyperbolic to model: deformation and time-to-failure.
equation, often called the Madison or Wood Models employed to characterize either the
curve, is used for adjusting all design stresses, deformation or the time-to-failure are intended
all grades and species of solid wood, and all to provide a mathematical description of the
composite structural components and materials behavior for a given environmental condition
that have assigned design stresses. and a range of loading conditions so that the
long-term behavior of a material may be
The history behind the design adjustment predicted in a variety of structural applications.
used for calculating deflection caused by long- The writer reviewed the models used to
term dead loads is not as well documented as describe creep deformation in the previously
the strength adjustment. Traditionally, the mentioned Proceedings (Laufenberg, 1987)
deflection caused by long-term loads on a and thus, will only review the time-to-failure
flexural member have been doubled when models here.
checking the design’s deflection criteria. This
doubling of long-term load deflection is The time-to-failure models that have
recommended for use of unseasoned lumber, been applied traditionally to composite wood
but deflection need only be increased by 50% products are of the exponential form
for use of seasoned material (National Forest
Products Association, 1986). S = A + B log10t

Figure 3 .--Duration of load stress adjustment curve as adopted in the 1952 National
Design Specification (NDS) (National Forest Products Association, 1986)

251
where
S = percentage of “static” strength
A, B = constants
t = time-to-failure

Of course, normally this form is applied


to data collected under constant steady-state where b and τ are parameters from the
stress and environmental conditions. Of creep function, J(t):
growing interest is the use of more sophisticated
J(t) = [1.0 + (t/τ) ] /E
b
models which are termed damage accumulation
models (Schaffer, 1973; Foschi and Barrett,
1982; Gerhards and Link, 1983). These models with
predict rate of damage (dα/dt) and assume
failure when damage reaches a given value. E = Young’s modulus
The models are summarized as shown: σL = material strength without flaws
σcr = infinite rate of loading strength
1/b
q = (b + 1 )(b + 2)/2)
K(t) = dimensionless crack length ratio

(Schaffer, 1973) Parameters for these advanced models are


difficult to generate and in some cases may
have little or no means of assessing confidence
in the estimates (Link et al., 1988). Thus, their
usage in reliability-based design formats may
(Gerhards and Link, 1983) be limited. Yet to be developed, however, is
the model which can link the deformation and
failure process.

(Foschi and Barrett, 1982)


TIME-TO-FAILURE: A REVIEW OF
where THE LITERATURE FOR WOOD
COMPOSITES
a(t) is the damage at time t (when α = 1,
failure is assumed) One of the earliest studies to encompass
a, b are constants the emerging product known as structural
c, λ are lognormally distributed random particleboard was that by Bryan (1960). Scope
variables of the testing was limited to assessing the time-
τs, is short-term strength to-failure for two particleboard densities and
τ(t) is stress applied at time t two adhesive types. Stress levels of 65, 70, 75,
σ is a threshold stress below which no 80, and 85% of the average bending strength
damage accumulates were chosen for the long-term loading. Board
density and adhesive type were not found to
It has been illustrated by Schaffer (1973) influence the stress/time-to-failure relationship.
and later by Caulfield (1985) how response Similarly, Kufner (1970) tested bending
under constant load is linked to ramp load specimens of eight types of particleboard at
behavior and simplifies to the traditional several different constant stress levels.
exponential time-to-failure model. Moisture contents were reported between 10
and 12%. Tests were completed on specimens
Another promising approach is that parallel and perpendicular to the panel length.
proposed by Nielsen (1985) which embodies a McNatt (1975, 1978) conducted an
fracture mechanics backbone where the rate of extensive test program on the creep-rupture
crack growth is: behavior of several commercially-produced
particleboards. The program included rate-of- model described earlier. have been
loading tests in tension, bending, interlaminar summarized. The summary illustrates the
shear, and edgewise shear. Tension tests and range of results found in the literature.
interlaminar shear tests were conducted for However, the work of these researchers in
constant load time-to-failure. Rate of loading identifying time-to-failure relationships can’t
results showed that bending, tension, and shear be compared directly. With variable materials,
strengths decreased 6-11% for each ten-fold geometrics, loadings, environments, and
increase in time to maximum load. McNatt analysis methods, the best that can be gleaned
(1985) also tested a variety of laboratory from the sundry results is an insight for the
flakeboards in a limited study of rate-of-loading sensitivity of time-to-failure to the many
effects in bending and tension and constant variables.
load behavior in tension.

The only published work on time-to- A COMMON BASELINE FOR CREEP-


failure which encompassess oriented strand RUPTURE EXPERIMENTS
board (OSB) and softwood plywood is that by
Hoyle and Adams (1975). Long-term bending Quite obviously, these long-term tests
tests were completed at 12% moisture content represent a significant technical investment.
on solid wood. OSB, and plywood at 82%, However, the results of these programs could
71%, and 66% of static strength for selected be better utilized if they were conducted
periods of loading. Residual strength of according to the same baseline set of
surviving specimens were determined to procedures, a protocol, for determining time
provide additional confidence in the stress time- dependent strength properties. The writer
to-failure data. proposes that the creep test standard presented
at the last Symposium (Laufenberg, 1987)
Lundgren (1957) presented data on the provides a backbone upon which time-to-failure
time-to-failure and strain-at-failure for tests may be completed. However, some
tempered hardboards, particleboards, and additional suggestions need to be made in
common fiberboards stressed in tension. regards to sampling sizes. load levels, and data
Regardless of applied stress and duration of presentation.
load applied to tempered hardboards, the strain-
at-failure was found to be fairly constant at Compared to creep test sampling, the time-
approximately 0.9% elongation. to-failure sample sizes need to be quite large
In a broad program of study on strength to produce usable results. Twenty-five
loss of hardboards over time, McNatt (1970) specimens at each of three load levels are
evaluated six commercial 1/4 in. thick recommended as minimum samples. Materials
hardboards with rate-of-loading and constant with high variability may need to be assessed
load tests. All major modes of loading were with significantly larger samples to provide
assessed using the rate-of-loading test while additional confidence in the test results.
only tension tests were completed for the
constant load time-to-failure. Strength losses Load levels to be applied to the specimen
of 8% for each ten-fold increase in time-to- groups need to be assessed in such a way that
failure were noted for different rates of loading median failure times cover the widest range of
and constant load levels. time permissible. The highest load level should
be chosen to produce median time-to-failure
Similarly, constant load bending tests on of less than two days while the lowest load
eight commercial brands of 1/4 in. tempered level expected to produce failures should
hardboards were reported by Chan (1971). A provide the longest time that the researcher
total of forty tests were completed for each can allow. The baseline for strength of the
brand and five constant stress levels (30, 40, material should be established using a load-
50, 60, and 70% of static) were used. controlled ramp test with a target time-to-failure
of one minute. This ramp test is crucial in
The results of these constant load time-to- establishing a common baseline for all products
failure tests (Table 1), using the exponential and modeling them in a common data pool.

253
Table 1 .--Summary of creep-rupture data in the form of exponential models of this form,
S = A + B log10D

Property Material A B

Tensile strength—McNatt (1985) Structural flakeboard 86.4 -7.9


Tensile strength—McNatt (1975) Structural particleboard 84.8 -8.3
Tensile strength—McNatt (1970) Hardboard 80.3 -7.8
Tensile strength—Lundgren (1957) Hardboard 83.2 -9.5
Interlaminar shear—McNatt (1978) Structural particleboard 83.0 -8.1
Bending strength—Wood (1947) Clear Douglas fir 90.4 -6.3
Bending strength—Hoyle/Adams (1975) OSB and plywood 85.1 -7.6
Bending strength—Ghan (1971) Hardboard 78.2 -6.8
Bending strength—Kufner (1970) Structural particleboard 80.3 -7.9
Bending strength—Bryan (1960) Structural particleboard 83.4 -6.5

*S in percent and D in hours

Aside from the environmental conditions produced plywood, oriented strand board, and
and measurements cited in the proposed creep waferboard of varying constructions are being
test standard, the only additional data needed tested to identify the variables most influential
is the time the specimen is under load prior to in their flexural creep and creep-rupture
failure. This data should be presented in tabular performance. Environmental conditions will
form and, if any modeling is done. the be varied for creep tests, but held constant for
exponential form should be presented along creep-rupture tests. The test program is
with any other models fitted to the data. intended to provide a broad data base on the
rheological properties and duration-of-load
RESEARCH FOR THE PRESENT behavior of these structural panel products.
AND FUTURE NEEDS This database will then allow researchers to
formulate a test program to generate the data
One very clear need in the creep-rupture necessary for development of design stresses.
field for structural composites is for a consistent
baseline of data. Standardized test methods, Regrettably, a number of other needs will
materials, and analysis techniques, as noted remain for the research community. Greater
earlier, are needed to formalize approaches to insight into the microscopic and chemical
the problem, yet provide adequate flexibility processes responsible for creep are needed to
to test creep-rupture hypotheses. A cooperative allow for the prediction and control of the
research program has been established which phenomenon. This is especially true for the
addresses these basic needs. interaction of stress, moisture content, and
temperature. Understanding these micro-level
Under the guidance of a task group from phenomena will probably allow for progress
the National Forest Products Association on a global model of load and environment
membership, a joint program on composite history effect, as mentioned previously.
panel creep-rupture has been established by Researchers need to know why creep, and
Forintek Canada Corporation and the U.S. thence, rupture, is accelerated in cyclic
Forest Products Laboratory. The program is environments and what micro-structural
being supported by both the Canadian Forest elements are contributing. Only with this level
Service and the United States Forest Service, of knowledge will it be possible to effectively
as well as the Waferboard Association and the design materials to resist creep effects.
American Plywood Association. Although the
program is two years old, results of the joint Another rather straightforward need is for
program are not expected to be published until the establishment of performance requirements
1989 (Laufenberg, 1986). Commercially for structural systems and components to

254
provide designers with guidelines for deflection LITERATURE CITED
limits under long-term loading conditions. As
discussed earlier in this paper, there is a need American Society for Testing and Materials.
for a creep-rupture model which can 1915. Report of Committee D-7 on
incorporate loading and environmental histories standard specifications for timber.
of structural members. Obviously, both the Proceedings of ASTM 18th Annu. Meet.,
performance requirements and the creep- Part II. Philadelphia, PA. p. 382.
rupture model will be useful in reliability
analysis of wood structures.
Bryan, E. L. 1960. Bending strength of
particleboard under long-term load. Forest
Prod. J. 10(4):200-204.

CONCLUDING REMARKS
Caulfield, D.F. 1985. A chemical kinetics
A number of basic concepts surrounding approach to the duration-of-load problem
the creep deformation and time-to-failure in wood. Wood and Fiber Science 17(4):
phenomena are discussed both separately and 504-521.
jointly with the term creep-rupture. The
continuous nature of the joint phenomenon is
discussed as a microscopic process which Chan, W. 1971. The structural properties of
ultimately produces microscopic failure. A hardboard. FPR 195/01, Forest Products
history of the development of design factors Research Laboratory. Princes Risborough,
for creep-rupture in solid wood products is Aylesbury, Buckinghamshire. United
offered here as a background for the Kingdom.
development of design guidelines in composite
products. Creep-rupture time-to-failure models Combs, T. C. 1939. A seismic design of wood
are summarized as a prelude to presentation of structures: Duration of stress
the literature on the subject of structural considerations. Bull. Seis. Soc. of Am.
composite panel time-to-failure. 29(4):539-547.
The world literature is shown to be a
collection of research studies meant to solve Foschi, R. O. and Barrett, J. D. 1982. Duration
specific needs, but it is not possible to collect of load effects in western hemlock lumber.
them all in a single meaningful database. J. Struct. Div. ASCE 108(ST7): 1494-1510.
Creep-rupture information is collected in such
a diverse manner that the writer could not
provide a definitive summation. However, Gerhards, C. C. and Link, C. L. 1983. Use of
time-to-failure information from several a Cumulative Damage Model to Predict
sources has been tabulated for this class of Load Duration Characteristics of Lumber.
products through use of the simple exponential Presented at IUFRO Div. 5 Conference.
model. This section is closed by offering Madison, WI.
suggestions for providing a common baseline
of experimental data with standards for testing
and data presentation. Hoyle, R. J. and Adams, R. D. 1975. Load
duration factors for strand wood, plywood,
Lastly, an overview of a large North and clear wood. Proceedings, Ninth
American test program is given which fulfills International Particleboard Symposium,
a basic need for a broad database on structural T. M. Maloney, Ed. Pullman, WA. pp.
composites creep-rupture behavior. Several 83-109.
other broad research needs are offered to this
audience as an open invitation for additional Kufner, M. 1970. Creep in wood particleboard
creative research on creep-rupture for this under long-term bending load. Holz a1s
expanding family of products. Roh- und Werkstoff 28(11):429-446.

255
Laufenberg. T. L. 1986. Creep and creep- National Forest Products Association. 1986.
rupture in reconstituted panel products: National Design Specification for Stress-
A cooperative North American study. In Grade Lumber and Its Fastenings.
Proceedings of International Workshop National Forest Products Association.
on Duration of Load in Lumber and Wood Washington DC.
Products. Special Publication No. SP-27.
Forintek Canada Corporation. Vancouver, Nielsen, A. 1972. Rheology of Building
B.C., Canada. pp. 61-66. Materials. Document D6. National
Swedish Institute for Building Research.
Laufenberg. T. L. 1987. Creep testing of Stockholm 27, Sweden.
structural composite panels: A literature
review and proposed standard. In Nielson, L. F. 1986. Wood as a cracked
Proceedings, Twenty-First International viscoelastic material. In Proceedings of
Particleboard/Composite Materials International Workshop on Duration of
Symposium, T. M. Maloney, Ed. Load in Lumber and Wood Products.
Washington State University. Pullman. Richmond, B.C., Canada. Special Publ.
WA. pp. 297-313. No. SP-27. Forintek Canada Corp.,
Vancouver, B.C. pp. 67-89.
Link, C. L., Gerhards, C. C., and Murphy, J. F.
1988. Estimation and Confidence O’Halloran, M. R., Johnson, J. A., Elias, E. G.,
Intervals for Parameters of a Cumulative and Cunningham. T. P., Jr. 1988,
Damage Model. USDA Forest Service. Consideration of reliability-based design
Research Paper FPL-RP-484. Forest for structural composite products. Forest
Products Laboratory. Madison, WI. Prod. J. 38(4): 35-43.

Lundgren, S. Ake, 1957. Hardboard as Schaffer, E.L. 1973. Effect of pyrolytic


construction material—a viscoelastic temperatures on the longitudinal strength
substance. Holz als Roh- und Werkstoff of dry Douglas-fir. J. Testing and
15( 1): 19-23. Evaluation 1(4): 319-329.

McNatt, J. D. 1970. Design stresses for


hardboard—effect of rate, duration, and Thurston, G. H. 1881. On the effect on
repeated loading. Forest Prod. J. prolonged stress upon the strength and
20(1):53-60. elasticity of pine timber. Amer. Assoc.
for Adv. Sci. Summarized Proc. pp. 9-16.
McNatt, J. D. 1975. Effect of Rate of Loading
and Duration of Load on Properties of Tiemann, H. D. 1909. Some results of dead
Particleboard. USDA Forest Service. load bending tests of timber by means of a
Research Paper FPL 270, Forest Products recording deflectometer. Proceedings of
Laboratory. Madison. WI. ASTM 12th Annu. Meet. Philadelphia,
PA. pp. 534-548.
McNatt. J. D. 1978. Stress-rupture of
particleboard in interlaminar shear. Forest Wood, L. W. 1947. Behavior of wood under
Prod. J. 28(9):34-36. continued loading. Eng. News-Record
139(24):108-111.
McNatt. J. D. 1985. Rate- and Duration-of-
Load Behavior of Lab-Made Structural Wood, L. W. 1951. Relation of Strength of
Flakeboards. USDA Forest Service, Res. Wood to Duration of Load. U.S. For.
Note FPL-0252. Forest Products Prod. Lab. Rep. No. R-1916. Forest
Laboratory. Madison. WI. Products Laboratory. Madison, WI.

256

Вам также может понравиться