中文摘要:
本研究以抗折試驗方式,對高純度的氧化鋁陶瓷材料所製作的兩種不同尺寸試片進行四點抗折強度試驗,與以四點及三點抗折方式承受靜力與週期負荷下的疲勞試驗進行研究。由韋伯分布強度與破壞機率的關係式中提出尺寸變數,以探討不同試片尺寸間抗折強度的尺寸效應,並利用韋伯分布結合慢速裂縫成長模式所推導的壽命關係式,探討在不同試片尺寸間疲勞壽命預測的尺寸效應。結果顯示,大小尺寸試片可以由體積缺陷之韋伯分布預測抗折強度的尺寸效應;由大尺寸試片靜力疲勞壽命對週期疲勞壽命的良好預測,顯示大小尺寸試片同樣無明顯週期疲勞機構。由於是以抗折強度實驗決定離散度的大小,當試片邊界晶粒粗大化的一側為張應力面時,很容易由這些大晶粒邊界處發生裂縫的起始而造成試片破壞,所以小尺寸試片具有較大的離散度;以離散度較大的數據關係預測離散度較小者容易將其實驗數據涵蓋。此外,小尺寸試片的頻率效應研究顯示週期疲勞壽命只與壽命時間有關與週期數(或頻率高低)無關,再度顯示小尺寸試片並無額外週期疲勞機構。根據破斷面的觀察小尺寸試片除了部份試片的邊界晶粒粗大化影響離散程度外,大尺寸試片比小尺寸者具有較差的微結構,因此在研究尺寸效應時,大尺寸對小尺寸的強度與疲勞壽命預測較保守而實際值較高。此外,統計疲勞試片的破斷位置,可解釋本研究之結果以單一或多個裂縫成長模式進行壽命預測,二者差距不大的原因。
Abstract
Delayed fracture behavior of polycrystalline alumina was investigated under long-term static and cyclic loading by four- and three- point flexure tests. The stress-probability-lifetime relationships in both single- and multiple-crack growth models were obtained with the merging of the concepts of statistical variations in strength and subcritical crack growth leading to delayed fracture. Specimens with two different sizes were prepared to evaluate the size effect on the strength and fatigue life distribution. Experimental results showed that the size effect of strength could be predicted by the volume-distributed-flaw Weibull function. Both single- and multiple-crack growth models provided good descriptions of static and cyclic strength-probability-lifetime for large and small specimens and no significant cyclic fatigue effect was detected. The predictions of fatigue lifetime from small to large specimens with different flexure modes agreed well with the experimental results due to the greater scatter of small-specimen data. The existence of larger grains on the outer edges of small specimens makes a greater scatter of testing results. However, the same predictions from large to small specimens underestimated the testing results. This could be explained by the less scatter of data and the worse microstructure of large specimens. The frequency effect was investigated with small specimens under several frequencies. Comparison of experimental results and theoretical predictions provided further support that no true cyclic fatigue effect was found in this particular alumina, as the total failure time in cyclic loading was independent of frequency. In addition, the statistics of fracture positions in fatigue specimens may explain the slight difference between single- and multiple-crack growth models adopted in this study.