Natural Durability of Timber Exposed Above Ground – a Survey

• Besides its inherent resistance against degrading organisms, the durability of timber is inﬂ uenced by design details and climatic conditions, making it difﬁ cult to treat wood durability as an absolute value. Durability classiﬁ cation is, therefore, based on comparing performance indicators between the timber in question and a reference timber. These relative values are grouped and related to durability classes, which can refer to a high range of service-lives. The insufﬁ cient comparability of such durability records has turned out to be a key challenge for service-life prediction. This paper reviewed literature data, based on service-life measures, not masked by a durability classiﬁ cation. It focused on natural durability of timber tested in the ﬁ eld above-ground. Additionally, results from ongoing above-ground durability studies in Europe and Australia are presented and have been used for further analysis. In total, 163 durability recordings from 31 different test sites worldwide based on ten different test methods have been considered for calculation of resistance factors. The datasets were heterogeneous in quality and quantity; the resulting resistance factors suffered from high variation. In conclusion, an open platform for scientiﬁ c exchange is needed to increase the amount of available service-life


UVOD
The natural durability of timber products is infl uenced by the interaction of wood properties, environmental conditions and structural design. Wood anatomy and the presence of natural protective chemicals (extractives) provide resistance against biodeterioration by microorganisms and insects. Communities of wood-destroying organisms vary between different locations, and their activity is infl uenced by climatic factors. Fungal decay and termite attack, for example, are generally more severe in warm and humid environs (Scheffer, 1971;Brischke, 2007;MacKenzie et al., 2007;Thelandersson et al., 2011). The extent to which timber components are affected by biodeterioration and weathering is also mediated by the design and maintenance of timber structures; for instance, the position of different structural elements and use of surface coatings alter their rates of wetting and drying, while untreated joinery and cracks in poorly maintained timber coatings may trap water and thus support decay (Norton and Francis, 2008).
Worldwide building codes and standards have traditionally provided natural durability information in a prescriptive context. Timber species are generally categorized into heartwood durability classes and the allowable uses of timbers belonging to those durability classes are prescribed (Stirling, 2009). Criteria for natural durability classifi cation differ between countries and include combinations of fi eld test data, laboratory test data, history of performance and expert experience (CEN, 1994;CEN, 2006;Standards Australia, 2008).
Many different fi eld and laboratory tests are used to measure natural durability. These include standardized and non-standardized methods, among which test environments, confi gurations and evaluation methods vary widely (Gobakken and Viitanen, 2004;Råberg et al.;2005;Stirling, 2009;Fredriksson, 2010). Tests that present a high biodeterioration hazard often involve soil contact or inoculation with microorganisms or insects. Above ground fi eld tests generally pose a lower biodeterioration hazard, but most test confi gurations are designed to accelerate decay by various moisture trapping elements. Durability evaluation procedures for fi eld tests commonly involve objective or subjective measures of strength loss, while mass loss is commonly measured for laboratory tests. Traditionally, fi eld test results are reported in a variety of ways, including mean or median measures of specimen service life or arbitrary scores that represent levels of biodeterioration. The performance of test species is commonly compared with the ones of non-durable reference species such as the sapwood of Scots pine (Pinus sylvestris L.) or southern yellow pine (Pinus spp.) for softwoods and common beech (Fagus sylvatica L.) wood for hardwoods. Beyond the relative performance of specimens in the circumstances of each test, however, the practical implications of durability test data are only beginning to be explored. Willeitner and Peek (1997) highlighted that comparing different durability tests is diffi cult, as in addition to the heterogeneity of test methodology, one may face results that are mostly codifi ed -sometimes in a cryptic way -or even incompletely published.
A major challenge remains to extract information from durability tests to help quantify the key factors that affect natural durability and integrate this information so that it is useful for predicting the service life of timber building products. Modern performance-based construction criteria require building products to be characterized in terms of the reliability that they will perform as expected over time. For timber, the current level of understanding of durability is far less developed than for other properties such as structural and fi re safety performance, and continued research is required to develop robust service life models (Foliente, 2000). Reliable service life data are also of crucial importance for Life Cycle Assessment (LCA) studies that are used to compare the environmental impacts of wood competing building materials.
Timber performance models have been developed that incorporate climate, durability classifi cation and design factors (Wang et al., 2008b;Viitanen et al., 2010;Thelandersson et al., 2011), however more data are sought for calibration and fi ne tuning. As an alternative to using durability class categories to represent wood properties in design guides (MacKenzie et al., 2007), the use of a resistance index and resistance classes has been proposed . Incorporation of 'durability factors' into a factor method has also been suggested (Dickinson, 2005;ISO 15 686-1, 2000).
Despite the importance of above ground structures in timber engineering, reports of natural durability studies involving above-ground exposures are relatively rare. Numerous laboratory decay tests have been reported, but their relationship with timber performance in service appears limited (Da Costa, 1979;Van Acker et al., 1999). Publications containing in ground 'graveyard' test data are more readily available, but their usefulness for service life modeling of above ground structures is unclear. The need for above ground durability to support performance modeling was more recently recognized, but due to their long duration, many above ground tests are incomplete and yet to be published. Above ground test results are likely to be most heterogeneous as they take a long time to complete and a wider range of standardized and non standardized methods may be used.
The aims of this review paper were to: (1) survey above-ground natural durability test data from published and ongoing fi eld studies; (2) examine the usefulness of data obtained for service life prediction; and (3) compute resistance factors and consider their implications for understanding the effects of differences between fi eld test sites and methods. Relevant literature was reviewed concerning the natural durability of timber species determined in fi eld tests above ground. Modifi ed and preservative treated timber was not considered as this would be unmanageable, due to increased amount of data and different testing approaches compared to non-treated timbers. Two a priori criteria for articles or data inclusion were set: (1) published in a peer reviewed journal, printed conference proceedings, international standard, project re-port or PhD thesis; (2) a focus on natural durability, fi eld testing or service life.

MATERIALS AND METHODS
The reference lists in the articles found and publication lists from durability researchers worldwide were checked for additional articles. The studies which met the a priori criteria used only four different test methods: the horizontal lap-joint test (CEN, 2003;Palanti et al., 2011); the horizontal double layer test (Augusta, 2007; Brischke et al., 2009); the cross brace test (Eslyn et al., 1985;Highley, 1995); and the accelerated L-joint test (Van Acker and Stevens, 2003). The principal confi gurations of these methods are illustrated in Figure 1. In most cases, there were minor variations in the basic set up for each test method between different studies, for instance in terms of shading, distance to ground, test rig size and material. Untreated control specimens included in tests of treated timber were included if necessary and appropriate.

Above ground fi eld tests 2.2. Testovi trajnosti drva iznad zemlje
In addition to published information, data from ongoing tests, which had not been published to date but were accessible to the authors, were included (Tab. 1). L-joint test units were constructed according to Fig. 1d using timber 35 x 35 mm² in cross section. Half of the specimens for each species were painted. Each joint was pulled apart after painting to completely break the paint fi lm along the frame of the joint and therefore create a uniformly high decay hazard by allowing moisture to enter and remain in the joint and under the broken paint. The 35 x 35 mm² faces at the distal ends of the joint components were sealed with bituminous tape.
At each site L-joints were placed on exposure racks that were constructed using CCA treated plywood and durable framing timbers that are resistant to insect attack. Plastic strips and brackets were fi xed to the racks to support L-joints and prevent them from coming into direct contact with each other or the plywood. At all locations the racks were faced north, and they were constructed the way that L-joints placed on them were oriented 10° backward from vertical to channel moisture toward the joint.
Assessment of the specimens was undertaken after 3,5,7,9,11,16,19 and 21 years of exposure. Only the 35 x 35 x 11 mm 3 face of the tenon part of each joint -the component most susceptible to decay -was assessed. The depth and distribution of decay was detected using the pick test, which involves fi rm probing using a small knife. Decay scores were assigned between 0 (sound, resistant to probing and no apparent loss of structural integrity) and 4 (failure, severe decay through the 11 x 35 mm 2 tenon part of an L-joint) according to Carey et al. (1981).  (CEN, 2003) Horizontalni test lap-spoja Specimens (38 x 85 x 300 mm³) are exposed horizontally on test rigs with supports 1 m above ground. The two lap-joint segments are fi xed through stainless steel clamps or plastic cable strips. The end-grain of each lap-joint is sealed with polyurethane or silicone.

Horizontal double layer test
Horizontalni test dvostrukog sloja Specimens (500 x 50 x 25 mm 3 ) are exposed horizontally in double layers according to Augusta (2007) with the upper layer displaced laterally by 25 mm to the lower layer. Supports are 25 cm above ground and made from aluminum L-profi les or Norway spruce beams with or without a bituminous foil.

Cross brace test -Križni test
Test units are constructed of 19 x 76.2 x 152.4 mm³ boards, that are nailed together at their centers to form a cross (e.g. Highley, 1995) and installed on test fences.

L-joint test (CEN, 1993)
Test L-spoja Specimens with dimension of 38 x 38 mm² in the cross section with a machined mortise and tenon joint are used. The members measure 203 mm in length. The whole L-joint assembly is either coated and afterwards disassembled or stays uncoated. Different modifi cations of the standard procedure are also considered.

Accelerated L-joint test
Ubrzani test L-spoja A modifi ed version of the L-joint test (CEN, 1993) is applied, e.g. according to Van Acker and Stevens (2003). L-joint tenon members are made from the test species. In contrast, the mortise member is half made of beech, and half made of Scots pine sapwood acting as feeder specimen. c. d. e.

Bundle test Type A Test svežnja tipa A
Each specimen consists of three segments, which are stakes of 25 x 50 x 500 mm³ and are ex-posed as a bundle.

Bundle test Type B
Test svežnja tipa B Each specimen consists of two segments, which are stakes of 25 x 50 x 500 mm³ and are exposed as a bundle.

Bundle test Type C
Test svežnja tipa C Each specimen consists of three segments, one bottom stake of 25 x 50 x 500 mm³ and two top stakes of 25 x 50 x 250 mm³, which are exposed as a bundle.

Bundle test Type D Test svežnja tipa D
Each specimen consists of two segments, which are stakes of 25 x 50 x 500 mm³ and are exposed as a bundle. The upper specimen has four circular drill holes with a diameter of 20 mm to allow water trapping.

Ground-proximity multiple layer test Višeslojni test u blizini zemlje
Each test unit consists of ten specimens of 22 x 95 x 250 mm³, stacked two by two in fi ve crossed layers, bottom layer on the ground, e.g. according to Edlund (2004). To avoid weed growth the ground is covered with a geotextile. Either the two upper boards or the two bottom boards are assessed (indicated as 'upper' and 'bottom'). h. i. j.

Horizontal double layer tests in Europe
Horizontalni dvoslojni testovi u Europi Double layer tests have been performed at 23 different European test sites to establish dose-response functions for above ground wood decay with wood moisture content (MC) and temperature. A detailed description of the study and a corresponding dose-response performance model is given by Brischke and Rapp (2008a, 2010. Specimens made from Scots pine sapwood (Pinus sylvestris L.) and Douglas fi r heartwood (Pseudotsuga menziesii (Mirb.) Franco) were monitored in terms of MC, wood temperature and the progress of fungal decay up to a period of eight years. The specimens (500 x 50 x 25 mm 3 ), according to EN 252 (CEN 1989), were exposed horizontally in double layer test rigs (see Fig. 1b) producing a decay risk corresponding to European Use Class 3 (CEN 2006). The upper layer was displaced lat-erally by 25 mm with respect to the lower layer. The lower layer consisted of seven pine sapwood specimens and six Douglas fi r specimens; the upper layer consisted of six pine sapwood specimens and fi ve Douglas fi r specimens. The whole test set-up formed a closed deck (73 x 65 x 21 cm 3 ). The specimens were evaluated yearly through a pick-test using a small knife and rating the extent and distribution of decay according to EN 252 (CEN 1989) as: 0 (sound), 1 (slight attack), 2 (moderate attack), 3 (severe attack) or 4 (failure).

Horizontal double layer tests in Norway
Horizontalni dvoslojni testovi u Norveškoj Horizontal double layer tests (Fig. 1b) were conducted with 29 different wood species (Tab. 6) as described by Evans et al. (2011) and Flaete et al. (2008Flaete et al. ( , 2011. Specimens were exposed at three different locations in Norway: Oslo (exposed in 2002), Bergen and Ås (exposed in 2004). In Oslo the test site is on the roof of the Norwegian Institute of Wood Technology, an 8 fl oor building, while the two test sites in Bergen and Ås are on ground level. Test set up and assessment of the specimens were identical with the above described procedure apart from the test rack size, which was larger due to a higher number of tested wood species. Samples were evaluated every year.

Lap-joint and ground-proximity multi layer tests in Sweden
Lap-spoj i prizemni višeslojni testovi u Švedskoj The lap-joint tests were started in 1996 and the specimens were assessed after 5, 8, 10, 12, 13, and 15 years of exposure. The ground-proximity trials were started in 2001 and assessed after 1, 2, 3, 5, and 10 years. Each ground-proximity multi layer test unit consisted of ten specimens, 22 x 95 x 250 mm³, that were stacked two by two in fi ve crossed layers, with the bottom layer on the ground. The assessment of the specimens in the stacks was carried out separately for the bottom and the Table 3 Mean decay rating according to EN 252 (CEN 1989) of specimens exposed in horizontal painted and unpainted L-joint tests after 21 years of exposure at ten different test sites in Australia. upper part (Tab. 1) using the pick-test. To avoid weed growth around the stacks, the ground had been covered with a geotextile, permeable for micro-organisms.

Bundle tests in Germany
Testovi svežnja u Njemačkoj Bundle tests of four different types (A-D) after  were conducted in Northern Germany. The specimens were made from Norway spruce as illustrated in Fig. 1f-i and exposed in 2007. Afterwards they were evaluated annually by using the picktest and rating the extent and distribution of decay according to EN 252 (CEN 1989).

Durability measures 2.3. Mjere trajnosti
Numerous evaluation and assessment procedures were analyzed with respect to their signifi cance and informative value for the prediction of service life. The following ranking of preference was applied to the different durability assessment measures: ; if n is even Where SL i is the service life of a single specimen (the year when a specimen was recorded to have failed) [y], v i is the decay rate of single specimen [y -1 ], R is the decay rating (score), t is the exposure time [y], and n is the number of replicate specimens.
Decay rate, as represented by the rate of change in decay rating over time, was considered as less desirable quantity with which to determine resistance factors. Whilst decay does not necessarily proceed at a linear rate, it was necessary to consider it as such for the purposes of this study. Different decay rating schemes had been applied, e.g. the fi ve step scales according to EN 252 (CEN, 1989) and EN 330 (CEN, 1993). Alternatively, the decay rate was expressed as 'mass loss rate v ML ', when only mass loss, but not decay ratings were available (e.g. Van Acker and Stevens, 2003).

Resistance factors 2.4. Čimbenici otpornosti
To make the different durability measures comparable, they were related to the respective reference species and resistance factors f were calculated according to 6 Where f SL

Aktualna istraživanja trajnosti
In total, results from six published and fi ve different ongoing durability studies were considered for this survey. To illustrate the latest state of the ongoing studies, which took place at different locations around the world and made use of seven different tests methods, the mean decay ratings are presented for all timber species tested (Fig. 2 and Tab. 2 to 6).   Site characteristics were found to affect the performance of particular wood species differently. The mean decay ratings for Douglas fi r heartwood after 6.5 years of exposure in horizontal double layer tests at 11 different locations in Europe is shown in Fig. 2 in order descending severity of decay. The respective 'non-durable' reference Scots pine sapwood did not show the same trend for decay severity amongst the 11 test sites. The differing ratio between mean decay rating for Douglas fi r and the reference species was presumably caused by a combination of their respective wood properties and climatic differences between sites. The particular properties of each species, such as moisture permeability and potential for leaching of protective extractives, may cause differences in the effects of climatic conditions, such as rainfall and temperature. Similar observations were made for the horizontal double layer samples exposed at three Norwegian test sites (Tab. 6). For instance, the mean decay ratings of grey alder (Alnus glutinosa L.) and Scots pine sapwood were almost the same after 6 years of exposure in Oslo and Ås, whilst the mean decay rating was signifi cantly higher for grey alder in Bergen compared to the Scots pine sapwood reference (Tab. 6). For other species, such as aspen (P. tremula), the ratios between tested timber and reference were nearly the same at all three test locations.
In addition to differences in decay progress between species at climatically different locations, the impact of test methods and test design became apparent. As shown in Tab. 5, the ratio of the mean decay ratings for seven wood species differed signifi cantly between the upper and bottom parts of ground-proximity multi layer tests in Borås, Sweden. The higher moisture load and limited potential for re-drying in the bottom parts of the stack diminished the differences between different timbers, which coincides with the reports by Augusta (2007) and Rapp et al. (2010), who compared the decay development of different European wood species under different exposure conditions above ground. For instance, the good moisture performance of the heartwood of European larch (L. decidua), Douglas fi r (P. menziesii) or Scots pine (P. sylvestris) is abolished when permanent wetting is provoked. For further comparative analyses of the different above ground trials considered for this survey, resistance factors were considered.

Resistance factors 3.2. Čimbenici otpornosti
The computation of resistance factors allowed the wide range of previous and ongoing tests to be compared, irrespective of test confi gurations and assessment methods. We found, however, that the number of durability recordings that were freely accessible from publications and relevant for service life prediction was generally sparse. Apart from the fact that above ground durability studies are rare, many of the reported studies contained insuffi ciently detailed results. The condensed format of presenting test results that is often used for publication inhibited the calculation of resistance factors with suffi ciently high statisti-cal reliability. The signifi cance of this problem can be illustrated by considering the Australian L-joint test, which includes 29 different wood species represented by painted and unpainted specimens installed at various locations, and the test has been assessed eight times to date. If the results were reported together, there would be 1808 mean scores alone. It is obviously beyond the scope of one publication to deal with this volume of data, so selected results have been published over time. If only mean scores at a particular time are reported in a single publication, they are not very useful to timber engineers researching service life prediction, as they attempt to fi nd and compile a complete set of data for analysis ( Tab. 2 and 3). Furthermore, representative measures of durability may need to be transformed for analysis, for example from ratings (scores) to service life values, so raw data are required. While it is possible to seek data directly from researchers managing durability tests, they may be diffi cult to fi nd. Individual publications may not reveal the full extent of an entire durability test when only specifi c elements of data are reported.
Tab. 1 gives an overview of the data regarded for this survey. In total, 163 durability measures from 31 different test sites have been considered for the calculation of resistance factors: 37 for hardwoods and 126 for softwoods. Only three reference species were used to compare the different durability tests: Scots pine sapwood (P. sylvestris), Radiata pine sapwood (Pinus radiata D.Don) and southern yellow pine sapwood (Pinus spp.). The resistance factors for six selected wood species, for which most durability records were found, are presented in Tab. 7 and 8. Several of these timbers are commonly used untreated for above ground structures that are exposed to the weather, including oak (Quercus spp.), spotted gum (Corymbia spp.) and western red cedar (T. plicata).
Most of the durability recordings were based on preliminary test results, and consequently, decay ratings after 4 to 21 years were used for calculating resistance factors. For most species the range of resistance factors was quite high, for example between 0.90 and 4.54 for Douglas fi r (P. menziesii), and in extreme -between 15.88 and 43.03 -for spotted gum (Corymbia spp.). In the case of Douglas fi r this can be translated to durability classes (DC, according to EN 350-1, CEN 1994) between DC 5 (non durable) and DC 2 (durable). This variation and how it can be related to at least three, in some cases even to four or fi ve durability classes, is shown for six selected wood species in Fig. 3. The importance of this variation becomes even more obvious when calculating the expected service life: Based on a mean service life of 6.5 years of the Scots pine sapwood reference (Tab. 8), the service life to be expected for Douglas fi r ranges from 7.4 to 29.5 years. Even more drastic is the range for spotted gum (Corymbia spp.), which is from 18.7 years and 473.3 years. These fi ndings highlight the potential value of service life modeling to greatly increase the accuracy and relevance of information available regarding the expected durability of timber used at different locations.     Although most of the results are still preliminary, they indicate that the resistance factor, and hence the relative durability of different species, is not necessarily the same at climatically different places. This is confi rmed by the results for European oak (Quercus robur / Quercus petraea): While the resistance factors for eight German test sites differed only between 1.35 and 1.83, a variation between 1.67 and 3.00 was found for three Norwegian sites. As there were only a few species for which multiple recordings were available, no clear relationship between the test site and resulting relative durability was discernible. Signifi cantly more durability recordings from different sites are needed. As previously discussed, chemical and anatomical properties of different species may infl uence the extent to which they are affected by climate variables, and this topic requires further investigation.
Another example is illustrated in Fig. 4, where the resistance factors of eight wood species determined in L-joint tests have been compared between ten test sites in Australia. Many additional wood species were installed at the Beerburrum site, while only nine wood species were installed at all ten sites. It would be ideal if resistance factors for the nine species tested at all sites could be used to gauge the performance of the additional species at Beerburrum, if they were used at the other locations. No simple relationship between relative resistance factors and test location was observed that represented all species. The higher the resistance factor -and thus the expected service life -the higher was the site-specifi c variation. In extreme, the factors differed between 4 and 32. For those wood species, showing resistance factors below 5, which is equivalent to durability class 2 = 'durable' according to EN 350-1 (CEN, 1994), the variation between most of the sites diminished, while differences between sites for the species with higher resistance factors showed the opposite. The test sites represent a wide range of climatic conditions, and preliminary analysis revealed that there is a strong relationship between climate variables and relative durability of each wood species exposed at different locations (Francis and Norton, 2006). The infl uence of the analyzed climate variables differed amongst the eight species. Further research is required to explore the possibility of using resistance factors to predict durability between different locations based on indicating wood species that are selected to represent groups of wood species with similar properties. For example, the resistance factors for spotted gum may more accurately predict the service life of dense hardwoods that contain extractives that are highly toxic to decay fungi, while resistance factors for brush box may be used to predict the service life of dense hardwoods that contain moderately toxic extractives.
To further examine the potential relationship between the severity of a test site and respective durability of timber species, resistance factors were correlated with the service life (mean or median) of the reference wood species for all sites at which these data were available. As shown exemplarily for three softwoods and three hardwoods in Fig. 5, no clear relationship was obtained. It leads to the conclusion that other factors than the site-specifi c decay intensity determine the relative resistance, such as climatic peculiarities, different decay types, or detoxifying agents.
In addition to potential site-specifi c effects, the test method and especially the durability measure seem to infl uence the resistance factors. While no clear differences between the use of mean or median service lives on the one hand and decay rates after certain exposure times on the other hand were observed, the use of mass loss differences led to signifi cant outliers for English oak (Quercus robur), Scots pine and Norway spruce (Picea abies) and the relative effects of durability measures, therefore, need to be verifi ed.
The infl uence of the test methods on the resulting resistance factors of a certain wood species is superposed by the effect of climatic conditions. Basically it is the microclimate within a wood specimen that determines the conditions for fungal growth and decay. Consequently, the combination of mesoclimate (environmental conditions at the test site) and the design of the respective test set up affect the microclimate. This is demonstrated by considering resistance factors calculated for Douglas fi r, which varied as follows: in double layer tests between 1.08 and 4.54, in uncoated L-joint tests between 0.90 and 1.74, in coated L-joint tests between 1.12 and 1.95, and in cross brace tests between 2.00 and 2.13 (Tab. 8). Obviously the variation within one test method was higher compared to the variation between the different test methods, which coincides with the fi ndings of De Groot (1992), who exposed Southern yellow pine sapwood in Mississippi, USA, and in a rainforest in Panama using 18 different test designs. While he found a signifi cant impact of the test design in the temperate location, differences diminished in the tropical rainforest. Within this study, test data from different test methods at the same test location were available only for a few wood species, so the potential effect of the test confi gurations was not quantifi able.

ZAKLJUČCI
We do not claim that this literature and data survey on above ground durability tests is complete. This is mainly due to the fact that many studies around the world are known to exist, but respective data are not freely available. The lack of freely available data is strongly indicated through the fact that 80 % of durability records used for this study was unpublished. Furthermore, in many cases information was too condensed and incomplete, which is inescapable for journal articles, but prevented the data transformation necessary to calculate specimen service life measures.
The range of test results observed for each wood species further highlighted that the current timber durability classifi cation systems, which assign a species to a durability class irrespective of site and design, are not precise enough for many scientifi c and engineering purposes. Data need to meet a number of requirements in terms of specifi city, background information and formatting.
We conclude that further research into the relative effects of climate on decay progress amongst different species is required, and future comparative studies should focus not only on differences between test sites, but also on different test confi gurations at the same location to determine the effects of structural design on timber durability. To facilitate this goal, a suitable platform is needed to increase the quantity and availability of useful data. Service life related durability recordings should be shared amongst the scientifi c community to allow the exchange and advancement of knowledge in this fi eld. The value of these durability data is expected to rise through collaborative comparative studies and meta analyses.
Similar or even more complex challenges are faced for predicting the service life of modifi ed and preservative treated wooden material because additional information of treatment agents and processes are needed. Wood used outdoors is commonly treated with different wood preserving agents and formulations, and fi eld studies on the durability of preservative treated and modifi ed timber include additional parameters, including preservative type, penetration and retention. For these reasons, a proposal for a 'Durability Data Base' has been made to the 'International Research Group on Wood Protection, IRG-WP' (Brischke et al. 2012). Requirements and feasible formats for durability recordings have been suggested for all types of wood products: naturally durable timber, thermally and chemically modifi ed timber, water repellent and preservative treated timber as well as for composite products. The database shall allow availability of test results from fi eld and laboratory studies dealing with wooddegrading fungi, insects, and marine borers.