In a previously published article on the same title [1], the results of the comparison of accelerated aging and natural aging performance changes around 14a were reported. Natural aging was then continued for another 16a. This article is the comparison result of oven accelerated aging and natural aging of about 30a.
1 Experimental part
1 1 1 Sample
The main components and vulcanization conditions of the practical formula are shown in Table 1. The sample is cylindrical, and its specification is 8mm × 10mm.
1 1 2 Aging conditions
The oven accelerated aging temperatures are 60 ℃, 70 ℃, 80 ℃, 90 ℃ and 100 ℃. The aging media are air, 8# lubricating oil, 10# hydraulic oil, 12# hydraulic oil, and л-1 hydraulic oil. The natural aging of the samples is carried out in the storage room. The storage room is not ventilated in the back of the sun, the windows are covered with curtains, and there is no heating in winter. The average daily temperature is 21 ℃ and the average relative humidity is 45%. Oil medium aging samples were placed in airtight glass containers filled with oil medium; air aging samples were placed in wooden frame drawers. The natural aging time is 30a for sample 1, 28a for samples 2 and 3, and 32a for sample 4.
1 1 3 Test method
For samples 1, 2 and 3, the accumulated long-term deformation ε and compressive stress relaxation coefficient σ/σ 0 were measured. Variation with aging time t; for sample 4 only the variation of ε with t was measured. The determination method is the same as the previous report.
2 Results and discussion
2.1 Comparison of 30a test value and predicted value
2.1.1 Establishment of ternary mathematical model
According to the accelerated aging, the P - T - t ternary mathematical model is established. The relationship between the rubber aging performance change P and the aging temperature T and aging time t can be described by the following P - T - t ternary mathematical model
y = B0 + B1x1 + B2x2 (1)
where: x 1 =1/T; x 2 = log t; B 0 , B 1 and B 2 are parameters to be estimated; y =log[ - log P/ B ] , for stress relaxation P = σ / σ 0 ;
For long-term deformation P = 1 - ε, here ε is expressed by coefficient instead of percentage; B is estimated by approximation method. Table 2 shows the model parameters calculated by using the experimental data of five oven accelerated aging temperatures, the variance ratio F of the regression equation and the residual standard deviation S of the curve.
2.2 Comparison of experimental value and predicted value
Use the established P - T - t ternary mathematical model to calculate the performance prediction value ^ P of different aging time periods at the equivalent temperature T 0 , and compare it with the experimental value P. And calculated the average deviation S of all test values and predicted values, as shown in Table 3, the calculation of S is carried out according to the following formula:
where n is the number of experimental points.
For the sake of clarity, Table 4 shows the comparison between the test value and the predicted value in the last 10 years.
2.1.3 Analysis of control results
Comparing Tables 2 and 3, it can be seen that the predicted average deviation S is less than 2 S curves for samples 1, 2 and 3, and only two cases of sample 4 are larger than 2 S curves but less than 3 S curves. That is, the predicted value is quite different from the experimental value. In addition, it can be seen from Table 4 that the test value changes faster than the predicted value, which is not due to the different mechanisms of natural aging and accelerated aging. Since this is the case both in the oil medium and in the air, the natural aging can be accelerated by the exclusion of moisture in the air. It may be caused by the following reasons: For sample 4, the aging fixture used in the natural aging test is double-layered, the upper layer is used for the test with a compression rate of 30%, and the lower layer is used for the test with a compression rate of 15%. All kinds of test samples are compressed by the same external force, so it is difficult to ensure that the samples are evenly stressed. Therefore, the experimental data fluctuates greatly, and there may be a systematic deviation.
Using the extrapolation calculation method, the prediction error is very large. There are two types of methods for error estimation: (1) safety factor method, and (2) mathematical statistics method. The former does not consider the specific circumstances of the test, and puts a safety factor in the same way; the latter gives the prediction interval with a certain probability according to the specific conditions of the test. It can be considered that the latter is more reasonable than the former, but there are certain problems, that is, only one of the two statistics is analyzed, and whether this simplification is appropriate remains to be further studied. According to the specific situation of this experiment, for the convenience of discussion, ±3 S-curve is temporarily used as the estimation error, and the performance change adopts the following formula
按此式可以计算预测的平均值、最大值和最小值 ,由表 3 可知对于所有情况下的 S 均小于 3 S 曲 , 因此就急体而言式 (3) 是成立的。由于自然老化下一年四季温度波动较大 , 测试季节的不同 , 测得的数据分散性较大 , 就个点而言不一定能全部满足式 (3) 。把表 3 中 S 较大的 3 种情况做 1 - ε对log t 关系图 , 如图 1 ～ 3 所示。图 1 ～ 3 中实线表示按公式预测的平均值 , 上面虚线为最小值 , 下面虚线为最大值 , 点为试验点。从图 1 可以看出全部试验点均处在 2 条虚线之间 , 对于图 2 和 3 中 ,绝大多数点也处在 2 条虚线之间 , 只有个别点处在最大值的下方 , 但距离不远。
综上所述可以认为对于应力松驰和积累长久变形 2 种老化性能而言 , 烘箱加速老化和室内自然老化机理是相同的。利用烘箱加速老化建立的P - T - t 三元数学模型外推计算自然老化下在30a 左右时间内的性能变化 , 其预测值与实测值基本上是吻合的。
2.2 长时间应力松驰行为的考察
应力松驰是橡胶密封材料一种重要的技术性能指标 , 研究它的长时间行为很有必要。在先前的文章内笔者曾说过研究二三十年自然老化行为就足够了 , 那时的认识有局限性。笔者以及国内绝大多数老化研究者都是针对航天航空等现代技术产品中的橡胶制品而进行的橡胶老化研究 ,这些产品中橡胶制品的保管使用期最长不超过15a 。现在随着工程建设的发展 , 橡胶密封制品广泛的应用 , 在地铁隧道 , 过海或过河隧道以及电缆隧道等工程中 , 对橡胶制品的使用期要求很长 ,例如地铁隧道工程中用的橡胶密封垫的使用期为100a 。
为了研究试样 1 、 2 和 3 常温下长时间应力松弛行为 , 做出了σ / σ 0 随时间 t 变化的关系图 , 如图 4 ～ 9 所示。图 4 ～ 9 实线是按According to加速老化建立的 P - T - t 三元数学模型 , 计算在等效温度下σ / σ 0 变化的预测线 , 虚线表示按 ±3 S 曲 计算的偏差线 , 点为实际自然老化数据。从图 4 ～ 9 中的实线可以看出 , σ / σ 0 随时间 t 的变化是非常缓慢的 , 特 别 是 对 于 试 样 2 和 3 。如 果 按 照ГОСТ 9 1 7143 - 86 规定以σ / σ 0 = 0 1 2 做为临界值的话 , σ / σ 0 变化到此值的时间非常长 , 如试样 3在空气中老化时 , 经万年后 σ / σ 0 才能变化到0 1 2 , 这是令人难以置信的。从实用角度出发 , 我们只考察 100a 内的应力松驰行为。为此计算了在等效温度下在 t = 100a 时σ / σ 0 的预测平均值及其上下限如表 5 所示。
100a 时σ / σ 0 的预测值的准确程度如何 , 无法通过试验加以验证 , 只能通过对图 4 - 9 的分析加以考察。因为已经有 30a 左右的试验数据 , 虽然 30a 与 100a 相比相差很大 , 但在对数座标上 2者相距并不大。如以天做为计算单位 ,30a 的对数座标为 log t = 4 1 04 , 而 100a 的 log t = 4 1 56 , 即只需考察 0 1 52 这一小段预测曲线与试验值的可能偏差问题。According to 30a 以前试验数据的变化趋势来看 , 可以预言在 log t 座标 4 1 04 到 4 1 56 这个区间试验数据与预测曲线偏差不会太大 , 特别是对于试样 2 和 3 情况会更加好一些。至于试样 1 偏差可能大一些 , 为了考察它的大小 , 把试样 1 自然老化试验数据σ / σ 0 对 t 关系做为直线图 , 如图10 ～ 12 所示。从试验点的变化趋势来看是直线 ,可建立直线方程 , 并According to此方程外推计算 100a 时的σ/ σ 0 值 , 见表 6 。比较表 5 和表 6 可以看出对于在 10#油中的σ / σ 0 值仍在预测偏差范围内 , 对于在空气和 8#油老化情况下虽然超出了预测的下限 , 但仅差 0 1 01 和 0 1 02 。因此仍然可以认为预测与实际变化相差不会太大。
我国橡胶工业发展的历史较短 , 关于橡胶工业制品实际使用最长期限的数据未见报道。对于橡胶制品究竟能使用多长时间 , 缺少感性认识。据国外报道 100a 前设置的橡胶桥梁支座在至今仍在使用, 虽然橡胶不断受到光、热、氧、臭氧、水、环境污染以及应力的作用。如果 100a 前配合的橡胶材料在如此恶劣的环境下工作 100a 尚未失效时话 , 那么具有良好耐老化硫化体系和防老体系的试样 2 和 3 制成橡胶密封件在比较温和的老化环境 ( 仅受热、氧、机械应力和油介质作用 ) 工作 100a 应当是可能的。
3 结论
The test results show that the four kinds of nitrile vulcanizate samples studied, in air or various oil media, the accelerated aging under stress state and the performance change mechanism of indoor natural aging for about 30a are the same. The natural aging temperature is characterized by equivalent temperature. According to the P - T - t ternary mathematical model established by accelerated aging, the predicted value of the natural aging performance change by extrapolation is consistent with the experimental value.
