數位影像相關法分析木材橫向拉伸力學行為

Abstract

木材是低環境衝擊之材料，然而其力學行為因構造與生成方式，顯得複雜而具有高變異性。為解決材料特性難以掌握的問題，本研究目的在於藉由木材的應變分布與構造破壞型態預測強度。試驗以杉木（Cunninghamia lanceolata）之橫向拉伸為主軸，採用數位影像相關法（Digital Image Correlation, DIC）的非接觸式全域應變量測技術進行，所得之應變分布圖再以「應變面積累加曲線」和「基準應變定位曲線」量化應變集中的程度。 不同取位之試材會有構造上相對弱點，依重大性排序包括髓心、春秋材交界、春材部到秋材部，髓心最大應變達0.36mm/mm，遠大於僅0.002mm/mm的秋材。此外，當缺點並存時，最重大的缺點會使其餘缺點的應變集中情形不明顯，也因此缺點若包覆於內部則難以觀察。「應變面積累加曲線」所計算的「應變積分面積比」隨時間的變化圖，因無法考慮空間相對位置造成的構造細節差異，依取位分組後組間曲線趨勢個別差異大，僅含髓心者可歸類明顯圖形樣式。若排除缺點不在表面之試材，可得到最終「應變積分面積比」與強度成線性正相關。「基準應變定位曲線」應變將縱斷面相對位置的落差更清楚描繪出各組差異。計算瀕臨破壞時最大應變區與破壞位置的「應變梯度」與強度迴歸，得到指數相關且R-square值分別為0.78和0.68。最大「應變梯度」隨時間為近似線性成長，取得對時變化率與強度亦為指數負相關，R-square值高達0.82，具有在試材破壞前預測強度的潛力。「應變梯度」與強度的關係，驗證構造過渡造成的應變集中，是決定破壞形態與時機的重要因素。 基於DIC克服傳統難以取得瀕臨破壞之應變及全域應變分布的量測範圍拓展，取得木材強度預測的相關資訊。未來若進行三維技術及內部透視的應變觀測，能更精確地探討木材力學行為的全貌。

Wood is the low-environment-impact material. However such a biological material is hard to well use on engineering because of its complicated and highly variable mechanical behavior. To solve the problem, this study attempt to predict wood strength through analysis of strain distribution. Transversal tension test with specimens of China fir（Cunninghamia lanceolata）was conducted and global strain was measured by DIC（Digital image correlation）technology without contact on material surface. After then we used the data to draw Strain-area cumulative curve and datum strain location curve that could quantize strain concentration. There are a few defects of wood structure ordered by relative severity that lead to strain concentration, including pith, growth-ring transition, earlywood and latewood. The maximum strain at pith is 0.36mm/mm, far greater than that at latewood 0.002mm/mm. Besides the most severe defect would deteriorate the strain concentration of coexist ones so that to observe the strain concentration of defects beneath surface was difficult. For the first analysis method, it was hard to categorize the characteristic of each specimen group from different cutting location by area ratio of strain integral-time curve calculated from strain-area cumulative curve. In fact the only obvious curve type that could be found was specimens with pith. Because of ignorance of effect on relative position, individual variation led to no significant positive linear correlation between ultimate area ratio of strain integral and strength unless we excluded the specimens with defects beneath surface. The second method of datum strain location curve was used to calculate ultimate strain gradient at area near the maximum strain and the crack that regress to strength with R-square values of 0.78 and 0.68. It was also potential to predict strength before failure by rate of maximum strain gradient-time curve which was linear according to the high R-square value of 0.82 of curve rate-strength regression. Thus, the results indicated that strain concentration owing to structure transition determined the fracture type and failure time. Consequently DIC measures global strain distribution, even close to damage that is hard to acquire on conventional measurement. This study prove that wood strength were possible to predict by strain concentration analysis. Strain observation by 3D technology and fluoroscopy will lead to much more accurate analysis on wood mechanical behavior in the future.