![[NF-2000 Database - AIR Program]](../images/Air.gif) |
AIR2-CT93-1057
Duration of Load Effect on Different Sized Timber Beams |
 |
Proposal No: | AIR2-CT93-1057 |
| Date Prepared: | November 1999 | |
| Source: | Final report October 1997 |
Summary
The objective of this project was to establish a new scientific basis for the determination of the long term performance of reconstituted timber, including glulam and laminated veneer lumber (LVL), under (mechanical and climatic) load with a duration of a few months. This information was needed before finalisation of EUROCODE 5. At the start of the project, there was a great uncertainty as to the correct value of the strength of large members under long term loading: D.O.L. (Duration Of Load) strength reduction (kmod) is the same for solid wood and glulam, for climatic class 1 and class 2, for all the problems appearing in timber designing (beams in bending, in tension; curved and tampered beams; notched beams; joints). It was necessary to formulate a project including experimental as well as theoretical and modelling aspects, covering a wide range of usual situations (dimensions of elements: size and shape; climatic variations).
Three general problems were studied (bending of straight beams, of notched beams, of curved beams) in 6 laboratories through a consistent experimental program, including short term tests (5 minutes) at two moisture contents, step by step slowly increasing loading or constant loading leading to failure at middle term (some months), under different climatic loading (constant, daily or monthly variations). The results of the project confirm the need: kmod strength factors are different for the three problems, curved beams and tension perpendicular to the grain being most sensitive to DOL effect in varying climate; kmod is very high (near) for notched beams on which mean moisture content and moisture gradient have no true influence; kmod is greater for LVL than for glulam straight beams.
Modelling tasks were necessary to explain the observed DOL effects: refined computational methods including new concepts on damage in wood are now capable to give an exploitation to the strong size effect (width effect) for short term strength of curved beam as well as to model the damage accumulation leading to long term failure in beams without stress concentration. Models developed from Fracture Mechanics can describe correctly the problem, of beams with stress concentration and can be applied to other situations than notched beams.
Activities
The project was divided into 3 operational Parts in each of which a constant experimental database was produced, using the facilities and the competencies of the various laboratories involved, and theoretical computer assisted models were prepared. These covered:
c. four point bending of curved beams, with an associated subtask concerning tension perpendicular to the grain.
Materials Two reference materials were chosen: glulam (parts a, b, c) and L.V.L. (parts a and b). For each task different specimen sizes were selected and different environments were chosen. These included controlled constant climate; uncontrolled outdoor sheltered climate, called nature climate, with sealed (moisture content considered to be constant) or unsealed specimens; controlled cyclic climate (between 55 and 90 % RH, with a period of one month). For each subtask (defined by the size and shape of the beam or specimen), short term reference tests (20 C, 65 % RH, failure in 5 minutes) were performed in a number great enough to have a good idea of the cumulative distribution of strength; this series of tests is repeated at a higher Moisture Content. For long term tests, the minimum size of the samples is a 10 specimen one; the test begins at a predefined stress level, which is based on the short term results the load is maintained at this level for 2 weeks or one month, after which, and for survived specimens, the load level is increased in a step wise manner with a new period of equal duration; the experiments are stopped when more than 50% of the specimens have failed. For most of the sub-tasks, this step loading procedure is completed by a constant loading method.
Analysis For interpretation of the results, the matched sample method was adopted: for the definition of the stress level SL in an experiment, the actual stress (or force) was divided by the stress (or force) in a short term experiment for the specimen having the same number in the order of increasing difficulty to fail; for the time of failure tf it was measured from the beginning of the last step. For comparison of the sub-tasks it was decided to draw a straight line between the points representing the median value of (SL, log tf) for the short term and the long term experiments, and to extrapolate it up to tf = 6 months. A kmod coefficient was then calculated for a duration of load of 6 months.
Experimental rigs During the first year of this project the following rigs were built:
The materials were fabricated in different places. For glulam, spruce was cut during the winter of 1994 in the south-east part of Sweden, in the vicinity of the Kinda sawmill, where it was sawn and graded (visually and by a Computer-Matic machine); the glulam was realised in Late-Rakenteet (Finland) for part c, Paul Mathis S.A. (France) for part a and Töreboda Linträ AB (Sweden) for part b. The LVL was manufactured by Finnforest Oy (Finland).
Modelling Most of the laboratories were involved in a modelling task; different approaches were used:
Results
At the end of the experimental program, the summary of short term strengths showed some unexpected results, the most important concerning curved beams, with characteristic values for tensile strength perpendicular to grain (MPa) as follows:
| dimensions (mm) | 90 x 600 x 5400 | 90 x 600 x 7400 | 140 x 600 x 7400 |
| 50% | 0.848 | 0.706 | 0.568 |
| 5% | 0.716 | 0.596 | 0.412 |
A synthetic view of resulting kmod (6 months) from the long term experiments show the benefits of such a project: the traditional value (the so called Madison curve) of kmod (6 months) is 0.7.
Discussion
Bending of straight beams Experimental results were presented in terms of kDOL and kmod. A small size effect on medium term strength between two heights (100 and 150 mm) of LVL beams was noticed, despite the fact no size effect has been found on short term strength, the Duration Of Load effect is almost constant for sealed (coated) beams and the moisture content variations seem to lighten this DOL effect; the different seasons (constant, increasing or decreasing mean Moisture Content) give slightly different kDOL .
Damage accumulation models are very useful tools for presenting and using D.O.L. experimental results : they can provide predictions for very long term loading (10 years). However the long term extrapolation has to be carefully studied: it has been demonstrated that the long term strength factor is only strongly dependent on the method used for presenting the results and to fit a damage accumulation model. Once this method is set, the influence of the variability of the parameters (number of tested beams, stress level, time to failure) has to be studied to assess their value for a medium term test according to a required accuracy of the prediction.
The calculated stress-strain state of the beam is monitored against a failure criterion. Since the failure criterion is based on strain energy density, its distribution across the cross-section is calculated as function of stress and strain. The criterion is different for tension and compression. A beam is considered to fail as soon as either tension or compression side fails. Spatial distribution of strain energy density and size effect is taken into account by applying Weibull's weakest link theory.
Modelling of long term capacity based on strain energy density, has some important advantages compared to traditional damage accumulation modelling. Strain energy is at least to some level of accuracy a measurable quantity, which gives it a more reliable character than a pure computational damage parameter. It also has a clear physical meaning and it can be calculated with the help of a suitable constitutive law, which in turn can be based on results of experiments. The fact that creep experiments can be used to support predictions of long term strength gives a clearly more reliable basis. Hence, the computational model used in this project seems to describe the DOL-effect.
Bending of notched beams It is believed that the results. obtained in this project may be of great value in development of timber design building codes. For analysis of the short term strength of a notched beam with a homogeneous moisture content it seems that linear elastic fracture applied through classical equation and the "initial crack" consideration gives reasonably good results. The effect of moisture content seems to be negligible for a common range of RH, i.e. for constant RH between 65% and 85%. Knots and grain deviations at the notch do not seem to decrease strength, but on the contrary improve the load capacity. Annual variation in climatic conditions resulting in non-homogenous content was found to have strong effect on load capacity. Depending on choice of beam failure criterion, i.e. if clearly visible cracks are accepted or not at the notch, the consideration to the effect of moisture variations becomes different. If a crack is not accepted, models suggest that a good moisture sealing at the tip of the notch is needed in common climatic conditions even if the beam is allowed to carry only a small external load. Experiments confirmed that there is no difference in terms of DOL effect in Summer or Winter, when there is no moisture gradient at the crack tip, between sealed or unsealed beams; the propagation time is really shorter in Spring and Autumn for sealed beams: moisture content gradients which appear at the crack tip at these seasons make the crack propagate faster. This project proved that, on loaded and cracked structures, the influence of natural daily humidity variations is important and quasi-instantaneous because it does not affect the entire volume of the beam but only the damaged zone at the crack tip.
Bending of curved beams Tensile stresses perpendicular to grain in curved beams, are found to be higher in the middle section, in the plane where pith is located, and much lower in the rest of the cross-section. The same is observed also during moisture cycling in long term loading. As a result, the failure will start when stresses in the middle exceed the strength. These are new scientific observations. In practical situations maximum stresses are obtained when a humid period is long. Tensile stresses at surface occur during dry periods, and they may become critical when drying is fast, width of beam is large or lamella includes the pith. When the experimental values in cyclic and constant climate were compared, it was concluded that when the beams are exposed to the test cycle, at least 50% of the duration of load effect is caused by the changing stress distribution, and less than 50% by the weakening of the material. The effect of stress redistribution seems to depend on beam width:
The results of the research work with EN standard glulam volume specimens (V = 0.01 m3) prove a significant and extreme difference of the DOL effect in service class 1 and 2 for tension perpendicular to the grain of glulam and further that the absolute values are well below those assumed today in Eurocode 5. The DOL tests with the larger glulam volumes (V = 0.03 m3) indicate that the volume effect, conforming to a Weibull shape factor of about m = 5 in short term tests is somewhat less severe in long term loading. The long term tests with solid wood specimens according to EN 1193 specifications for ramp load tests revealed a considerable smaller (67%) DOL effect compared to glulam. Extensive numerical investigations revealed that this fact is to a large extent due to the deviating boundary conditions resulting from respective specimen inherent configurations. Thus, DOL tests with solid wood specimens according to EN 1193 cannot be used to substitute far glulam tests.
Based on the DOL results with two significantly different glulam volumes at three different climates and related ramp-load tests, a set of kmod values is proposed which in contrast to EC 5 accounts for the strength reduction due to climate influence at service class 2 conditions. As the investigations were conducted with glulam of high strength classes (appr. GL 32 and GL 36) the proposals given subsequently refer primarily to these grades; the applicability to lower quality materials seems sensible but should be checked.
Whereas the kmod values for service class 1, given in EC 5 and derived here, coincide roughly, the values for service class 2 in EC 5 should be reduced significantly by about 40%. It shall be mentioned that the proposed service class 2 values are slightly higher compared to empirical values for the 0.01 m3 volumes, thus accounting for the somewhat reduced size effect in DOL tests.
Conclusions
Although, almost by definition, it is impossible to plan the outcome of a research project, it is believed that the information and knowledge that have been obtained in the present project at the least is of as great a value as initially estimated. In basic research as well as before future development of timber engineering codes with respect to duration of load and climate effects on strength of structural members, the results obtained add much new qualitative and quantitative knowledge and can therefore be assumed to be of significant future value.
© Copyright 2006 Policy Statements
Updated
by CPL Press:
03/07/2007
- biomatnet@biomatnet.org
 |
 |
 |
News |
 |
Events |
|
|