Building Research Establishment Information Paper
Reproduced for the information of members of The Vintage Club of Great Britain by courtesy of the Director, Building Research Establishment and by permission of the Controller of HMSO. Crown copyright.
Ageing of wood adhesives – loss in strength with time
D F G Rodwell, BA
Joints made with most common wood adhesives tend to lose strength as they age. This paper gives the results of longterm tests carried out at the Princes Risborough Laboratory which quantify these strength reductions over a period of 40 years for unstressed material and 27 years for stressed joints. Implications for glued structures are considered.
Adhesives are used extensively in the bonding of wooden structures and wood-based components and in many cases the glue line has a load-transmitting function. It is important therefore that the performance which can be expected from a bonded joint in service is known.
In the early 1940s R A G Knight and his co-workers at PRI, had foreseen the need for long-term ageing information and laid down the basis of an experiment to study the loss in strength of unstressed glued joints stored in interior conditions. At intervals over the years these samples have been removed from store and tested to destruction. The results of this ageing over 40 years are presented here.
Limited data are also available from a small-scale experiment with loaded joints, which has been running for 27 years.
The data stems from small single-lap joints made with urea formaldehyde (UF), phenol formaldehyde (P17), resorcinol formaldehyde (RF) and casein resins. Three versions of these joints were prepared:
- close-contact joints for adhesives designed to be used for assembly work where effective contact pressure can be applied and a thin glue line expected;
- Gap-filling joints for adhesives designed to be used where a close fit cannot be guaranteed and contact pressure may be poor;
- joints for plywood resins.
All three versions used thin 25.4 mm wide beech slips in a configuration, which gave a 25.4 mm overlap. Their ultimate strength was measured by pulling on the ends of the slips until fracture occurred in the lapped, glued area. The specifications covering the manufacture and testing of the joints were originally issued by the wartime Ministry of Aircraft Production and have since been replaced by British Standards BS 1203 and 1204 which are very similar to their predecessors. The close-contact and gap-filling joints used two beech slips which were 3.2 mm thick whilst the plywood joint contained three veneers each 1.6 mm. thick. The thickness of the gap-filling glue line was 1.3 mm.
Since manufacture, the joints have been stored loosely packed in covered boxes and kept in dry, interior conditions at normal laboratory temperatures, from time to time a number of samples (usually 12) have been removed for testing. All the adhesives considered here have been tested in the dry state and most have also been tested after 24 hours .of soaking in water at room temperature. Current BS specifications call only for a minimum strength when tested wet. The 24 hour soak has been shown in the past to be a fairly searching test of some aspects of a UF resin’s performance and it provides a guide to a resin’s durability in damp conditions. It will not of course simulate all aspects of a hazardous service environment.
Double lap joints (each comprising three 25.4 mint wide beech slips arranged to give a central 12.7 mm glued area) were made in 1956.
The joint design is detailed in BS 3544:1962 “Methods of test for polyvinyl acetate adhesives for wood”. Twelve joints were prepared for each of 3 resins: UF close-contact, UF gap-filling and resorcinol gap-filling. The samples were supported on edge near their ends and carried a sustained central load of 450 N. Figure 1 shows the arrangement used.
For the close-contact resin the 450 N represented 27 per cent of the breaking load of matched joints tested soon after manufacture, and for the gap-filling resin 30 per cent of their initial load-carrying capability.
Table 1 gives the forces required to break lap joints made with the 18 commercial adhesives used in the experiment. Values are given for their initial strength usually measured a week or two after manufacture and strength after 40 years of ageing in dry, interior conditions. The values given are the mean of (usually) 12 specimens. Variability between individual results for each resin was such that the coefficient of variation (the ratio of the standard deviation to the mean) was about 10 per cent.
Note: the strength referred to is the force required to break a 25.4 mm square lap joint in tension
Table 2 gives the minimum strength requirements for joints when tested wet and as specified in the current British Standards. It provides a basis for the comparison of the performance of joints in the tests reported here, both when first made and after 40 years.
Table 3 gives the strength retention of each adhesive, expressed as a percentage of its initial strength, at intervals of 10 years for 40 years. The data in this Table are presented in a simplified form in Figure 2. In this Figure the data are grouped by resin type and the lines shown present the mean values of all resins of the same type. In Figure 2a for example the line showing the fall in strength of UF plywood resins is the mean of the seven plywood resins shown in Table 3.
In Figure 2f the wet strengths of the RF and PF gap-filling resins are shown as dotted lines. The wet strength of these resins was only measured at the 40-year interval and the lines plotted assume that the wet and dry strengths were the same when the joints were first made.
Figure 3 shows the survival times of individual joints in the 3 groups of samples subjected to sustained loading. The strengths of the survivors were obtained in the dry state.
Initial failures of the UF close-contact joints occurred after 5 years and after 14 years only 2 of the 12 samples remained. The test was stopped after 16 years when the strength of the remaining 2 samples was about three-quarters of their initial value.
Initial failures of the UF gap-filling resin occurred after 13 years and after 15 years only 4 of the initial 12 samples remained. This test was also stopped after 16 years when the strength of the remaining 4 samples was about two-thirds of their initial value.
Failures of the RF resin began to occur after 13 years and 8 of the 12 remained intact after 27 years. Four of these have been tested to destruction and it was found that they had retained approximately half their strength.
DISCUSSION OF RESULTS
Although the resins used in the unstressed samples were formulated more than 40 years ago their performance in joints made at that time was well up to modern specifications. Their strength has steadily declined however over the years and now only a few of the aged joints reach the values listed in Table 2. Although these British Standard minimum requirements are not meant to be used as yardsticks for aged material they nevertheless provide an indication of the extent to which the initial bond performance has deteriorated.
On average the UF plywood resins were reduced to half strength in 30 to 40 years but individual resins had only one third of their original strength. Of special note is the loss in strength of the UF resins used in the close-contact joints; when tested in the wet they were very weak, retaining only 25 to 30 per cent of their initial strength. Note that the plywood samples were all initially stronger in the wet state than in the dry. The likely explanation of this stems from the fact that the plywood test piece is made from thin cross-banded veneers which are very flexible when wet. This flexibility reduces the stress concentrations, which occur in more rigid samples and allows the wet plywood to resist higher loads.
Unfortunately the original wet strength data for the PF and RF gap-filling resins are not available. The dry test data show that these resins have retained about three-quarters of their original strength and if it is assumed that the initial wet strength was similar to the initial dry strength, then this would indicate that the RF and PF resins have retained 50-60 per cent of their initial values. These are the resins that are most unlikely to have been used in the construction of load-bearing beams and for this reason their ageing characteristics are of special interest.
Experience has shown that RF and PF resins are better than most at coping with adverse service conditions and the data presented here tend to confirm their relative durability.
In calculating the design stresses of wood-based structures no account is made of the glue line; only the stresses relevant to the timber itself are used. Any factor that reduces the bond strength of a glue line below that of the wood is important, therefore, since it changes one of the design assumptions.
In well-made joints in normal service conditions the assumption that a glue line is as strong as the wood substrate is likely to be valid during the early years of a joint’s life for most reputable commercially available adhesives. This was the case with the resins tested here. When the joints were tested to destruction the fracture surfaces at first showed significant amounts of wood failure in nearly all cases. As time passed however only the joints made with PF or RF resins have sustained their performance in this respect. The aged joints incorporating resins of other types tended to fracture within the glue or at the glue/wood interface. This effect is more pronounced on samples tested in the wet state.
The mechanisms which cause a reduction in the strength of glued joints are complex and not fully understood, although it is known that UF resins are susceptible to degradation by acid hydrolysis. At room temperatures and normal glue line acidities, this is a slow process but, over the long periods considered here it may well be a significant factor In determining the performance of the UF resins.
Resin ageing is usually accompanied by shrinkage and embrittlement which tend to make the joint more vulnerable to stresses and strains induced by changes in temperature or moisture content or which stem from external sources. The precise methods of bond failure will depend largely on the type of wood/adhesive bonding formed during resin cure and a more thorough knowledge of this process could help to explain the strength-reduction phenomenon.
Ultimately it is to be hoped that a fuller understanding of the fundamental processes would lead to improved adhesives and better predictive durability tests. PRL has already initiated work via an extra-mural contract with Leicester Polytechnic on some of the fundamental issues and work at PRL itself will concentrate on the development of improved methods of predicting the service life of glued joints.
In practice, most timber load-bearing members would have a minimum factor of safety of 2 to 2.5, so that the design load would represent about 40 to 50 per cent of the minimum failing load. For the majority of their service life however many structures carry less than the design load. A few roof members might for example be subjected to a continuous long-term load of 20 to 25 per cent of falilng load; most would carry far less. In the tests reported here the stressed samples were subjected to sustained loads of roughly 30 per cent of failing load, i.e. rather higher than most service conditions.
It is often difficult to relate experimental results to performance in service. In the experiment described here for example the type of stress induced in the test piece differs significantly from that which is likely to occur in practice, and in terms of loading the experiment is probably more severe. On the other hand, for some service situations the climate of exposure may be more demanding. In a loadbearing structure the bonded joints may not of course be the weakest points in the design. Even at design load the shear stresses in timber members at glued joints are frequently less than the permissible stress.
There is no evidence at present of widespread de-bonding problems and most failures can be attributed to specific causes either in manufacture or environment of use. However if the progressive deterioration of bond strength evident from the unstressed joints is considered together with the stressed joint behaviour then the margin of safety may be less than intended.
The resins tested here were all made in the 1940s. Most of them are no longer available in their original form but there are nevertheless very similar modern counterparts of the same chemical types. The strengths of today’s UF’s and PF/RF mixtures are also similar to those tested here and they can be expected to age in a similar manner.