Effectiveness and subjectivity of visual inspection as a method to assess bending stiffness and strength of chestnut elements

Besides the difficulty of assessing an existing timber structure on site, the efficiency and accuracy of visual inspection is often compromised by its subjective nature inherited by the level of expertise of the inspector. This often leads to conservative predictions of the mechanical properties, even with the use of specific visual grading norms. The main objectives of this work are to assess the effectiveness of visual inspection as to define different classes of strength and stiffness and to provide a statistical analysis on its subjectivity. For that aim, visual inspection using Italian standard UNI 11119:2004 and bending tests of 20 old chestnut beams (Castanea sativa Mill.) at different scales, were carried out. Comparisons and effectiveness of visual inspection is analyzed within and between different scales, and also regarding different level of expertise of inspectors. The results evidence similar percentages of segments classified with higher and lower visual grades and proved to be a good qualitative indicator of bending strength between sawn beams. For the most experienced inspectors, an overall 42% accuracy was found with a better differentiation between visual classes, whereas lower level inspectors scored approximately 5% less. Lower level inspectors evidenced higher concentration of values around a higher mean for each class denoting a more conservative approach. Regardless of the inspector level, knot size was the main limiting visual parameter with higher influence in small scales of the timber elements. When studying the characterization of a single knot, coefficients of variation of 15.7% and 21.8% were found for measuring the minimum and maximum diameter. Bayesian probability networks were considered as to individually assess the accuracy in stiffness prediction of different level of inspectors, and by combination of their information, evidencing that parallel combination for prior information may allow the increase in visual inspection accuracy. Introduction The result of visual strength grading is the attribution of indicative values or allowable stress levels, for prediction of key properties, which allow for the mechanical characterization of the material. However, besides the difficulty of assessing an existing timber structure on site, the efficiency and accuracy of visual inspection is often compromised by its subjective nature inherited by the different levels of expertise of inspectors. Measurement uncertainty arises from random or systematic errors made by different inspectors when identifying and characterizing defects that limit the performance of the timber element. To overcome those errors, guidelines for visual grading have been proposed in the past decades with subsequent national standards. Comparative studies for visual strength grading standards of different countries has been addressed (e.g. [1,2]). However without analysis of uncertainty between inspectors, with different expertise, when using the same standard. Visual inspection is subjective and may be inaccurate as evidenced in [3], when inspectors achieved only 68% of the expected result in recognizing, locating and identifying defects in grading red oak lumber. Also in [4], the quality yield for the automatic system was found to be between 52% and 75%, while manual grader only reached between 31% and 61%, comparing to the optimal grade defined by decision of several manual graders. Human visual inspection rarely achieves better than 70% performance in grading lumber [5]. While in [6] it was found that when grading boards into four grades, only 60% were assigned the same grade by two different expert inspectors. The main objectives of this work are to assess the effectiveness of visual inspection as a method to define different classes of bending strength and stiffness and to provide a statistical analysis on the evaluation of the subjectivity of this method, based on the effective wood performance. For that aim, visual inspection following Italian standard UNI 11119:2004 [7] and bending tests of 20 old chestnut beams (Castanea sativa Mill.) at different scales were carried out. Inspectors with different levels of expertise were considered in the subjectivity analysis and its accuracy was measured. Experimental Campaign Aiming to assess bending stiffness and strength using visual inspection, 20 chestnut (Castanea sativa Mill.) beams were visually graded and the results were compared to 4-point bending test results. The more than a century old timber beams were rescued from a building in Northern Portugal, where they served as structural floor beams. The experimental campaign was divided into three main phases, corresponding to different scales of the timber members. From one to the next phase, the timber elements were sawn into smaller sizes in order to isolate the influence and location of defects, and also to provide a better definition of the distribution of stiffness and strength along the length and height of the beam. The main experimental phases correspond to the members: (i) in the initial state of conservation as they were in the building; (ii) after being sawn to beams with 7×15×300 cm dimension; and (iii) after being sawn to boards with 7×4×300 cm dimension. In each phase, the members were visually inspected and graded on each 40 cm segment, using UNI 11119:2004 [7]. This standard establishes objectives, procedures and requirements for the diagnosis of the state of conservation and estimates nominal stiffness and strength values for structural wood elements. For strength grading of a single element, it considers three classes (I, II and III) regarding on site diagnosis. The wood element is from a given class if it fulfills all the imposed requirements, otherwise, in this study, graded as non-classifiable (NC). Although the standard bases its grading on the evaluation of the critical zone considered by the effective loading of the element, which would correspond in bent members to the tension side of the beam's central third, in this study all segments were considered as to provide a larger size sample with broader variety of defects. Therefore, visual grading was obtained by segment, rather than on an element level. Visual grading was performed by several inspectors divided by level of expertise and experience. Level 1 inspectors (L1), have prior experience in visual grading, both to softwood and hardwood species, with different grading norms and also experience in mechanical testing. Level 2 inspectors (L2) have experience in mechanical testing but have less practice in visual grading of hardwoods. Level 3 inspectors (L3) correspond to individuals with low previous experience in both visual grading and mechanical testing. A total of 336, 616 and 63 inspections to segments were made, respectively by L1, L2 and L3 inspectors. The sawn beams and boards were submitted to 4-point bending test according to EN408:2003 [8], obtaining local (Em,l) and global (Em,g) modulus of elasticity in bending and bending strength (fm). Further detail on the experimental campaign sequence and results has been presented in [9]. Effectiveness and Accuracy Effectiveness and accuracy of visual grading was considered by analysis of the percentage of segments attributed to a given visual grade comparing to the mechanical tests results. This analysis compiled the comparative study between visual inspection of L1 inspectors in different scales (test phases) and the visual inspection of L1 and L2 inspectors made to a sample of sawn boards. Different Scales. For the old beams, visual inspection was essential for determining the most damaged zones which were mainly concentrated in the extremities of the beams (with biological decay and cracks), and/or in the central span zones (knots with relevant size and drying cracks). With the purpose of a more adequate comparison with the following test phases, the possibility to discard the parameters influenced by external damage due to structural or environmental reasons was considered and a residual cross section was assumed. This allowed to obtain the mechanical properties related to the material itself and its defects rather than to the state of conservation. The results of visual grading in different scales are presented in Fig. 1, with consideration to the percentage of segments classified in each visual class. The disregard of external damage in old beams led to a significant higher percentage of class I segments and decrease of class III and NC segments, evidencing the need to define properly the residual cross section. The following phases present similar percentages for intermediate classes, however different in extreme classes (lower class I percentage and higher NC percentage). This indicates that decreasing the cross-section size produced a decrease in visual grading, especially when the size of knots was the limiting parameter. Based in a qualitative analysis, visual inspection proved to assess fm efficiently in sawn beams. Considering the sample of sawn beams taken to failure, a higher fm was obtained for the beam with higher percentage of class I and II segments, while a lower fm was obtained for the beam with higher percentage of NC segments. In all cases, beam failure initiated in a segment previously identified as a critical section where higher percentage of defects were accounted (see also [9]).