焊接构造细节疲劳试验数据统计特征分析

Analysis of Statistical Characteristics of Fatigue Test Data for Welding Details

疲劳试验是钢结构疲劳设计的基础,由于疲劳试验数据的高离散性,应结合数理统计分析建立具有可靠度保证的失效概率-可靠度指标-应力幅-疲劳寿命(P-α-S-N)曲线.本文对现有国内外主流规范中疲劳试验数据统计分析方法进行汇总,并对容限系数k的计算方法与概率假设进行对比,分析容限系数k随可靠度指标α及失效概率P的分布情况以及有效数据个数n对其影响情况.对现有文献中对接焊缝、传力十字接头及非传力十字接头3类典型焊接构造细节疲劳试验数据进一步开展了统计分析,建立了可靠度指标α分别为0.95、0.75和0.50时的焊接构造细节P-α-S-N疲劳设计曲线.结果表明:现有规范中3种焊接构造设计曲线均偏于安全且与本文数据统计分析结果较为接近;相较于样本数量充足的情况,当试验样本数较小时可靠度指标α及失效概率P对容限系数k的影响更为显著;焊接构造细节疲劳试验数据拟合标准差SD集中在0.12~0.19,随疲劳试验样本数量减少,数据整体标准差SD倾向于增加,单侧容限系数k_((P,α,v))及疲劳设计曲线与中值S-N曲线偏移量k_((P,α,v))SD均呈上升趋势;当样本数量充足时,可靠度指标取值对焊接构造细节疲劳设计曲线影响较小,而当样本数量较小时(小于272),可靠度指标取值对焊接构造细节疲劳设计曲线影响增大.本文所建立的3种焊接构造细节P-α-S-N曲线将对确定焊接钢结构疲劳设计与安全评估可靠性及其他疲劳试验的规划和开展提供有益参考.

Fatigue tests are the basis of fatigue design for steel structures,and due to the high discreteness of fatigue test data,a failure probability-reliability index-stress amplitude-fatigue life(P-α-S-N)curve with the assurance of reliability should be established by combining with statistical analysis.In this paper,the existing mainstream statistical analysis methods for fatigue test data at home and abroad are summarized.Then,the calculation methods and probability assumptions for tolerance factor k are compared,and the distributions of k with reliability indexαand failure probability P and the influence of the number of effective data n on k are analyzed.Furthermore,the fatigue test data of three typical types of welding details(i.e.,a butt weld,a load-transferring cruciform joint and a nonload-transferring cruciform joint)in the literature are analyzed,and the P-α-S-N fatigue design curves for these welding details with different values of reliability indexα(i.e.,0.95,0.75 and 0.50)are established.Analyses show that the design curves for the three types of welding details specified in the existing codes are on the safe side and close to the statistical results obtained in this paper.Compared with those in the case of a sufficient number of test samples,the influences of reliability indexαand failure probability P on tolerance factor k are more significant when the number of test samples is small.The standard deviation(SD)of fatigue test data of welding details is concentrated in the range of 0.12—0.19.With a decrease in the number of test samples,the SD of the overall data tends to increase,and both the one-sided tolerance factor k_((P,α,v))and the offset k_((P,α,v))SD of fatigue design curves and median S-N curves are on an upward trend.When the number of test samples is sufficient,the value of reliability indexαwill have less influence on the fatigue design curves for welding details.In comparison,when the number of test samples is small(e.g.,less than 272),the corresponding influence will be more obvious.The established P-α-S-N curves for the three types of welding details provide a helpful reference for the welded steel structure fatigue design and safety assessment reliability,as well as for the planning and carrying out of other fatigue tests.