Application of instrumented micro-indentations to ‘in situ’ mechanical characterization of wooden structures:

Part II – Analysis of different species Barletta, M. Department of Mechanical Engineering, University of Rome “Tor Vergata”, Via del Politecnico, 1 - 00133 Rome, Italy

Summary Flat punch instrumented micro-indentation tests were performed and analysed by using an original approach to assess strength characteristics of different wooden species. The quality of the perpendicular compression strength-indentation load correlation, using this particular type of indenter, was investigated. It is found that some characteristic force values of the instrumented micro-indentation test are very well correlated to perpendicular compression ultimate strength. In this work, the characterization on pinewood was first led, by using highly selected ‘clear’ and ‘straight grained’ materials which, at the same time, weren’t affected by bad orientation of annual ring as well as by a moisture level higher than 12%. Then, in order to generalize the experimental procedure, a broadening of the method to generic ‘clear’ and ‘straight grained’ pinewood, fir wood and larch wood samples, characterized by a variable moisture level and annual ring orientation, was performed. The interpretation of the load over displacement trace, the development of an experimental set-up, the execution of tests on various materials and the estimation of reliability of correlations by a strict statistic approach are also discussed in detail. Keywords: Micro-Indentation, Wooden Species, Selectivity

1. Introduction Micro-indentation testing is becoming an attractive way for scientists and industry to obtain valuable information on the material intrinsic mechanical properties and particularly on metals. The plainness and non-destructive features of the test, the quickness of execution, the viability of ‘in situ’ measurements and the small size of the needed sample are just some of the attractive features of micro-indentation testing. In the nineties, several investigations were performed to extract the major number of information from the micro-indentation test [1-3]. One popular method is to instrument the test in order to measure the trend of ‘load versus displacement’. A chiefly successful geometry is the spherical indenter, from which tensile properties can be extracted using the ‘so called’ partial unloading technique [1]. Another well known indirect method to obtain the tensile properties is the punch test. A spherical or flat punch is used to pierce a small clamped disk. By comparing the two types of indenter, it is found that the flat punch has a decisive advantage over the spherical one. The load versus displacement curve displays a clear yield point that can be more easily and accurately correlated to tensile properties [2,3]. Concerning on what has just been mentioned, several attempts to correlate overall mechanical properties of several metal alloys with local properties, measured by instrumented micro-indentation testing, occurred in the past years [4]. A theoretical basis is also developed even if, as main drawback, the testing procedure is found to produce reliable experimental results only when a perfectly smooth surface and a homogeneous

material are indented. At the same time, some further attempts in generalizing such technique to non-metallic material like paper, polymer and web structures are currently being carried out, starting to provide first relevant results [5]. From these considerations, our research is started to evaluate the feasibility of extracting useful information from the instrumented micro-indentation test using a flat punch on a further different material, even on an orthotropic and naturally dispersed behaviour one, like the wood.

2. The aim of the work The definition of a proper technique to determine mechanical properties over a wide range of wood species and with a great potential for portable instruments, in situ and non destructive applications, has always been of major interest. As well known, the wood mechanical properties are obtained from conventional tests of samples termed “clear” and “straight grained”, because they do not contain characteristics such as knots, cross grain, checks, and splits. Variability, or variation in properties, is common to all materials. Nevertheless, because wood is a natural material and the tree is subject to many constantly changing influences (such as moisture, soil conditions, and growing space), wood properties vary considerably, even in clear material. By the light of previous considerations, the broadening of micro-indentation testing to wooden samples seems to be a very slippery slope. In this context, our research is consisted of the employment of a strict theoretical background to allow the interpretation of the load versus displacement trace in micro-indentation testing, by avoiding or limiting the bad influence on experimental results of sample defects or growth features. Experimental data are collected aimed at working the sample strength characteristics out from a few of indentation tests with good accuracy and repeatability. Pinewood highly selected samples, with no deviation in annual ring and moisture level below 12%, were used at this purpose. In addition, the development of an ‘ad hoc’ experimental set-up, the generalization of procedure to various wooden materials and the establishment of some peculiar correlation factors between the detected indentation load and the overall mechanical properties are also studied in depth. In particular, the capability of the flat punch instrumented micro-indentation test to get to the value of perpendicular compression ultimate strength is investigated. A strict statistic analysis of results is also led as support to experimental approach. As a result of the experimental findings, valuable data are first achieved in testing highly selected pinewood samples. Then, the procedure broadening to generic wooden samples is also successfully led at a large extent even if more care has to be spent to reach for more reliable results. At last, a rigorous validation of experimental results is also effectively done by using statistical considerations. Moreover, a rigid testing protocol is defined so as to minimize all negative and unpredictable influences on collected data and as support to the user in developing the testing procedure on more wooden samples and in different scenarios.

3. Experimental apparatus and procedure A description in depth of experimental apparatus and testing procedure can be found in the related work [6]. Further useful information on the employed testing methodology is reported below.

Moreover, some indications about the involved wooden species with the related properties and the UNI regulations which were followed during the conventional static tests were also provided.

3.1. Experimental apparatus

The execution of micro-indentation test was led by using a static test machine provided with an ultra-precision load cell which allows accuracy in load determination set at ± 1%. A sketch of the experimental apparatus is reported in a previous related work [7]. The machine was equipped with two automatically lined up tempered stainless steel work plains, articulated in spherical joints so as to assure the uniform distribution of load all over the specimen. On the upper working plain, it was possible to insert the equipment to hold the indenter to carry out the instrumented indentation test. The indenter consisted of a cylinder, made of ‘hard metal’ A09, 17.5 mm long and, typically, 0.5 mm as radius, equipped with an utterly flat head (± 0.005 mm). Load cells of 1, 10 and 50 kN as maximum load were used throughout the experimental tests to store data. In particular, a specifically designed and high sensitive load cell (200 N as maximum load) was employed to perform micro-indentations. Data recorded by static test machine were sent to a computer. A dedicated SW allowed storing of the experimental results so as to permit the repetition of test by just moving the sample kept fixed on a high precision (± 250 nm) CNC movement system, held on the bottom working plain.

3.2. Tested wooden species

Three wooden species were chosen: larch, white fir and pine. A thorough description about these species can be found in Wood’s Handbook [8]. In Table 1, just some main characteristics [8] for each species are summarized.

Wooden Sample Pine

(Pinus sylvestris) White Fir

(Abies alba Mill.) Larch

(Larix Decidua Mill.)

Specific Gravity

Green 880 kg/m3 920 kg/m

3 860 kg/m

3

Oven Dry 550 kg/m3 440 kg/m

3 650 kg/m

3

Structure

Texture Fine to medium Medium Fine

Fibrous Nature Dependent upon

climatic conditions Generally Straight

Generally Straight or Spiral

Mechanical Properties

Compression Strength

45N/mm2 35N/mm

2 51N/mm

2

Bending Strength

95N/mm2 70N/mm

2. 92N/mm

2.

Hardness poor poor poor

Impact Resistance

poor Poor-fair modest

Young’s Modulus

13.700N/mm2 14.000N/mm

2. 14.000N/mm

2.

Most common Structural Defects

Pitch Pockets, Knots, chromatic alterations

by fungi, other chromatic alterations

Irregular annual ring orientation, encased knots, inter-grown knots, canaster.

Canaster, irregular slope of grain, pitch

pockets; fungi attacks even on alive trees

Table 1: Wooden Samples Characteristics

3.3. Sample characteristics and experimental protocol

The preparation of samples and the execution of experimental tests followed the standardized procedures imposed by UNI regulations. Table 2 summarizes the indications of each specific regulation for static testing of the chosen wooden species.

Test Static Bending Tension

Perpendicular Compression Perpendicular

Compression Parallel

Normative UNI ISO 3133 UNI ISO 3346 UNI ISO 3787 UNI ISO 3132

Table 2: Conventional static testing: UNI normative. No regulation for micro-indentation test is currently available, being an absolutely novel characterization technique. So, as described in the related work [6], it was defined an ‘ad hoc’ experimental protocol and the samples employed for perpendicular compression were effectively used to carry out reliable micro-indentation experiments, as well. The first set of 100 samples was in pinewood. All the investigated samples were made from highly selected ‘clear’ and ‘straight grained’ wood with no deviation in annual ring and a moisture level set at below than 12%. A set of ten indentations for each sample was carried out and then, on the indented face, a perpendicular compression static test was led to evaluate their ultimate strength. Perpendicular compression test was properly chosen in order to correlate ultimate strength of wooden samples with stresses measured by micro-indentations. The reason of such choice was related to the solicitation typology affecting specimen throughout the last of the experimental test. In fact, the same solicitation occurs on specimen both during perpendicular compression test and instrumented micro-indentation test. The only difference stands in the solicitation distribution: in the conventional test, the load acts all over the face of the specimen; vice versa, in the instrumented micro-indentation test, load is concentrated locally, nearby the indented zone, in a very restricted area of the same face of the specimen. It is very important to remark that the relative direction between the load and the fibre is absolutely the same for both the tests, so the response of wooden samples is reliably correlated. A case of perpendicular tangential compression was examined for pinewood, being the samples chosen with the annual rings all directed tangentially to the compression direction. In particular, since the local and non-destructive typology of micro-indentation test, it was possible to carry out the perpendicular compression test on the same specimens, previously, subjected to indentation tests. Moreover, the choice of samples characterized by the same moisture level (below 12 %) was due to the attempt at minimizing external influences on overall mechanical properties of specimens. In the second part of the work, a generalization of the previous procedure to ‘clear’ and straight grained’ specimens, but with a variable orientation of the annual rings and moisture level, was carried out. The procedure foresaw the execution of four conventional static tests (Table 2) for each wooden species and the employment of, at least, 12 samples for each test. Moreover, as regards perpendicular compression, 8 out of 12 samples were first indented and then characterized with the static test machine so as to perform a straight comparison on the same wooden specimens. A total amount of ten micro-indentations for each specimen was carried out. Each wooden species was then subjected at least to 80 micro-indentations. Once led the experimental plan, a standardized protocol and an equal criterion was followed to automatically process all the experimental data and to discard those considered incorrect. A full description of the adopted procedure was reported in the related work [6]. In this context, it’s important to remark that data which were affected by:

Errors in the testing execution and, above all, in the starting position of indenter compared to wooden sample stricture

Specimens with visible defects, alterations or clear structural anomalies,

Tests with outcomes out of regulation expectations, were discarded in order to attain the best reliability of collected experimental results.

4. Experimental results and discussion

4.1. Analysis of pinewood samples

As mentioned, the first step in assessing wood properties, was the application of the testing procedure to highly selected pinewood samples. The idea was to carry out experimental tests on sample, which were affected by as least defects or alteration as possible, to seek a reliable correlation between pinewood ultimate strength in perpendicular tangential compression and indentation load. In such way, a minimization of all external disturbances to the determination of factor constraints between the two measures occurred. Figure 1 shows the experimental results. An averaged indentation stress of 13,68 MPa and an averaged stress in perpendicular tangential compression of 4,68 MPa were detected. A statistical analysis of achieved results was firstly performed. In particular, the ‘un-stacked’ ANOVA was performed in order to check the reliability of collected data by comparing the significance of results for all the ten samples. Figure 2 summarizes the results. Perpendicular compression data gave rise to Fisher’s value of 3,11 slightly higher than those tabulated due to a few of samples (less than 10 out of 100) with an unreliable behaviour, as expected. In fact, even if all cautions were taken in selecting pinewood samples, being wood a material naturally ‘dispersed’ some deviations from ideal behaviour were inevitable. A P-value of 0,12 characterized micro-indentation data, showing a higher reliability of procedure due, almost certainly, to the ten perforations performed to test each sample, which minimized disturbances, and to the local character of testing procedure. The definition of a correlation coefficient between collected data was so achievable. In fact, by simply working out the ratio between the averaged stress in perpendicular compression and stress so as measured by micro-indentation tests on the same sample, the correlation coefficients were estimated. An averaged value of 2,95 was found out (Figure 4). A strict statistical validation of correlation coefficient was led. First of all, the ‘un-stacked’ ANOVA checked out the reliability of results by comparing the significance of data achieved for all the ten samples (Figure 4). Then, a residual analysis was performed to give more support to experimental analysis. A Fisher’s factor of 1,39 and a related P-value of 0,204 confirmed the reliability of experimental findings. In addition, the test on residuals confirmed that they were randomly distributed and neither bias nor peculiar pattern were present.

Figure 1: Micro-Indentation and Perpendicular compression

stress data on pinewood samples

Figure 2-3: Dotplots of data collected in perpendicular compression

and in micro-indentation on pinewood

Figure 4-5: Dotplots of correlation coefficient and

normal probability plot of residuals for pinewood samples So, external factors and the sequence of experiments had no relevant influence on data, confirming, once more, the reliability of performed investigation. Normal probability plot of residuals (Figure 5) shows points, which fit very well a straight line. Therefore, residuals got a normal distribution and no further investigations were required as W-test for normality confirmed. Once examined the experimental results, the averaged correlation coefficient of 2,95 was used to check out the capability of procedure in estimating the ultimate strength of pinewood samples. Well then, the achieved results gave rise to very consistent results with expectations, as Figure 6 shows. All the calculated results fell in the measured range by using the perpendicular compression static test.

Figure 6: Estimation of ultimate strength of pinewood by using the correlation coefficient

By comparing micro-indentations with other non destructive or almost non destructive methodologies [8], significant advantages arise. In fact, all the ultrasonic wave techniques seem to be strongly influenced by the bad propagation of waves in wood caused in turn either by the fibrous and discontinuous nature of material or by related defects and alterations, with overall mechanical properties measures severely affected [9-10]. At the same time, massive penetration techniques, being affected by wood instability and friction troubles and influenced by the hardness and compactness properties of wood fibres, strictly linked to the specific wood species, fibre direction and water content, produce low reliable and repeatable measures as well as quite tricky to generalize to singular testing scenarios [11]. On the contrary, as before highlighted, the micro-indentation test appears to provide useful cues in determining overall mechanical properties of wooden samples by just using an ‘in situ’ applicable, local, non destructive as well as fast and cheap test.

4.2. The generalization procedure: conventional static tests

The first step in the attempt to generalize the experimental procedure was the accurate measurement of the overall mechanical properties of the, namely, generic wooden samples. In agreement with indications provided by sector regulations, bending, perpendicular tensile and perpendicular and parallel compression static tests on all the three investigated wooden species were carried out to deduce the major number of cues from the behaviour of each sample set. Figure 7-8 and 9-10 show in turn the results achieved for static bending and perpendicular tensile test and for perpendicular and parallel compression test.

Figure 7-8: Bending test and perpendicular tensile test trend over wooden density.

Figure 9-10: Parallel and perpendicular compression test over wooden density.

In every test, the ultimate stress stored for larchwood was found to be the highest if compared with the values reached for firwood and pinewood. This result was more evident reporting the ultimate stress over samples density. As expected, mechanical properties resulted to be quite dispersed because of the generic moisture level and orientation of annual ring of chosen samples so as described in previous section. However, a certain scattering of experimental results was expected in perfect agreement with indications coming from scientific and technical literature [6]. In particular, as regards the collected values, a good match with available literature results and theoretical expectations was found out [6].

4.3. The generalization procedure: instrumented micro-indentation test

Experimental trends of micro-indentation tests didn’t show wide differences for the three

investigated wooden species, as Figure 11 states. As seen in conventional static tests and in according with cues coming from scientific and technical literature [8], the indentation test made the larch record higher stress values rather than those reached for pine and fir. In agreement with experimental results attained in the related paper [6], it was possible to stand out a quite irregular trend of measured stress from 0.2 – 0.3 mm as indentation depth. A flatly linear trend was attained in the first part of the test and this conferred a great reliability to collected data. Moreover, as stated in the related paper [6], the stress value achieved for 0.1 mm provided the most useful indications in the process of identifying the mechanical properties of wooden samples.

Figure 11: Typical trend of micro-indentation on all the three wooden species.

4.4. Comparison between collected data and definition of a correlation

coefficient

The last part of the paper was in turn dedicated to generalize the correlation procedure and to define new correlation coefficients between results coming from conventional static test and instrumented micro-indentation test. Some cautions had to be taken on new samples because of the differences in their moisture level and in annual ring orientation which produced highly dispersed data. Concerning on what has just been mentioned, a strict protocol was followed in order to determine a correlation coefficient between conventional test and micro-indentation results for such generic samples. Many parameters were considered, as below reported:

Perpendicular compression ultimate stress. It was worked out in according with the normative and, opportunely, manipulated by using an empirical formula [8] so as to

account the orientation of annual rings. No corrections accounted the moisture level in wooden samples, because the same samples were employed for both the conventional static and micro-indentation tests.

The ultimate stress of instrumented micro-indentation test. Experimental results were recorded for an indentation depth set at 0.1 mm and tests were performed with an indentation speed constant and set at 0.4 mm/min (in according with indications provided by the related work [6] to attain the most reliable results).

The influence of the orientation of annual rings was quite relevant. Figure 12-14 report the ultimate stress values worked out by considering or not such contribution. Corrected data represent the stresses which would have been recorded for specimens in which all the annual rings were oriented tangentially to the compression direction (case of perpendicular tangential compression). In Figure 15, the averaged values of stresses and standard deviations were collected for all the three wooden species. In particular, for each species, it was represented (continuous line) the overall averaged value of all the performed indentations, as well.

Figure 12-13: Perpendicular compression ultimate stresses stored by machine and worked

out by accounting the orientation of pinewood and firwood annual ring

Figure 14: Perpendicular compression ultimate stresses stored by machine and worked

out by accounting the larchwood orientation of annual ring

Figure 15: Comparison between micro-indentation results on all the three wooden species. The local character of the micro-indentation test was balanced by performing, as previously discussed, twenty indentations all over the face of each specimen. At this regard, standard deviations, reported in Figure 15, give a more reliable idea of differences which arose between each local test and the overall properties of the specimen. By the light of previous experimental findings, the correlation coefficients, defined as ratio between the recorded stress for the instrumented micro-indentation test and perpendicular compression, was evaluated and Table 3 summarizes the results. The reliability of the average correlation coefficient for each species was estimated by a strict statistical approach. The experimental results stated the chance to employ such averaged correlation coefficients in estimating, accurately, the ultimate stress value of each sample by using a whatever indentation stress related to a generic indentation test on its face. The averaged error of this procedure is worth about 2% for firwood and pinewood and about 5% for larchwood. The reliability and repeatability of experimental results got this coefficients being able to predict overall mechanical properties of wooden samples. Figure 16-18 report for each wooden species, the differences between the ultimate perpendicular compression stresses recorded by static test machine and those worked out by using the averagred correlation coefficients and the indentation stresses. A good fit between analytical and experimental data arises, confirming, once more, the capability of the procedure in determinining overall mechanical properties of wooden species by using a local, non-destructive and ’in situ’ applicable test.

Species Firwood Pinewood Larchwood

Correlation Coefficient

% Error Correlation Coefficient

% Error Correlation Coefficient

% Error

Sample 1 2,70 1,26 2,94 -3,16 2,36 10,75

Sample 2 2,80 -2,52 2,78 2,44 2,70 -1,92

Sample 3 2,70 1,03 2,75 3,37 2,57 2,99

Sample 4 2,77 -1,33 2,86 -0,35 2,45 7,41

Sample 5 2,60 4,89 2,92 -2,42 2,88 -8,83

Sample 6 2,74 -0,49 2,84 0,36 2,81 -5,96

Sample 7 2,82 -3,18 2,80 1,89 2,64 0,30

Sample 8 2,72 0,34 2,91 -2,13 2,77 -4,45

Species averaged correlation coefficient

2,73 1,88 2,85 2,015 2,65 5,326

Table 3: Correlation coefficient for each wooden specie.

Figure 16-17: Comparison between ultimate perpendicular compression stresses

stored by machine for firwood and pinewood and those worked out by coefficients.

Figure 18: Comparison between ultimate perpendicular compression stresses stored by

static test machine for larchwood and those worked out by usign the correlation coefficient. In particular, both firwood and pinewood show a very consistent trend of experimental data over measured, as Figure 16-17 show. Slight higher scattering appears to characterize the results reached for the larchwood samples. This is probably due to the nature of annual growth ring of the larch, very narrow (seldom they overcome 1 mm), determining a severe impact on micro-indentation results and affecting collected data reliability, as Figure 18, ultimately, shows. So, more cautions have to be taken to characterize such kinds of wood. No straight comparison can be performed between results collected in Figure 16-18, with similiar results, because no analogous data are reported in literature. This novel experimental procedure can be so considered as definitely advanced in the field of non-destructive methodology for the characterization of overall mechanical properties of wooden structures.

5. Conclusion

This work deals with the definition of a totally novel experimental procedure to estimate the overall mechanical properties of wooden samples by using an ’in situ’ applicable, local, fast, cheap and non-destructive testing methodology: the flat punch instrumented micro-indentations. Two main results were achieved: the first one is related to the calibration of the procedure on pinewood ’clear’ and ’straight grained’ samples showing, also, the same annual ring orientation (tangential to the compression direction) and moisture level (below 12%) so as

to minimize external distrubances in determining overall mechanical properties. The second one consists of a generalization of the procedure to ’clear’ and ’straight grained’ firwood, pinewood and larchwood samples with generic and variable orientation of annual ring and moisture level to broaden the field of applicability of the testing methodology. Very consistent results with more valuable expectations arose from the investigation of highly selected pinewood samples. A correlation coefficient of 2.95 was worked-out and its reliability was succesfully checked out both analytically (comparing overall mechanical properties experimentally achieved with those worked out by using the correlation coefficients) and statistically. In addition, the generalization of procedure successfully occurred: three very close correlation coefficients were achieved for the three investigated wooden species. Best fit between experimental data and those analytically calculated by using the correlation coefficients were achieved by firwood and pinewood. Slight worse results were reached for the larchwood, probably due to the too narrow annual ring which definitely affected the micro-indentations results. At last, a rigid testing protocol could be extracted by our analysis so as to minimize all negative and unpredictable influences on collected data, in the attempt to generalize the procedure to other wooden samples and as support to the user in applying the testing procedure in different scenarios.

6. References

[1] Haggag FM, In-Service non-destructive measurements of stress–strain curves and fracture toughness of oil and gas pipelines: examples of fitness-for-purpose applications. Fifth International Conference on Pipeline Rehabilitation & Maintenance; Bahrain 2002. [2] Lucas GE, Odette GR, Sheckherd JW. Shear punch and microhardness tests for strength and ductility measurements. The use of small-scale specimen for testing irradiated material, ASTM STP, 888; 1986, p. 112–40. [3] Foulds JR, Wu M, Srivastav S, Jewett CW. Fracture and tensile properties of ASTM cross-comparison exercise A 533B steel by small punch testing. Small specimen test techniques, ASTM STP, 1329; 1998. p. 557–74. [4] M. Scibetta,1, E. Lucon, R. Chaouadi, E. van Walle, Instrumented hardness testing using a flat punch, International Journal of Pressure Vessels and Piping 80 (2003) 345–349 [5] J.J. Pawlak, D.S. Keller, Measurement of the local compressive characteristics of polymeric film and web structures using micro-indentation, Polymer Testing 22 (2003) 515–528 [6] Barletta, M., Application of instrumented microindentation to ‘in situ’ mechanical characterization of wooden structures, Submitted to Building Materials and Consturctions, 2005. [7] Barletta, M. Sura, E. Application of microindentation tests to ‘in situ’ characterization of mechanical properties of different woods, Technical Report, European FAR projects for the development of Southern Italy, 2004 [8] Wood Handbook, Wood as an engineering material, United States Department of Agriculture, Forest Product Laboratory. [9] Sandoz, JL. Wood testing using acusto-ultrasonic. Proceedings of the First European Symposium on Non Destructive Evaluation of Wood. Sopron, Hungary: University of Sopron, 21-23 September 1994. [10] Ross RJ, Pellerin RF. Nondestructive testing for assessing wood members in structures. A review. Tech. Rep. FPL-GTR-70, Madison, WI: Department of Agriculture Forest Service, Forest Products Laboratory. [11] Ronca, P. Gubana, A. Mechanical characterisation of wooden structures by means of an in situ penetration test, Construction and Building Materials 12, 233-243, 1998