2 Relationship between accelerated test and outdoor test
2.1 Accelerated test design
After knowing the shortcomings of accelerated testing, you can design an accelerated aging test program instead of putting the sample into any available accelerated aging Test Chamber for exposure. The basic points of test design are as follows:
1) Accelerated test conditions and cycles should follow the principle of simulation first and acceleration second.
2) Determine the extreme values of the accelerated test conditions. For example, maximum and minimum temperature, maximum irradiance, range of relative humidity, etc.
3) Determine the response speed of the material: when the relative humidity increases from 50% to 100%, the rate of change of material moisture; and when changing in the opposite direction, the rate of change of material humidity.
4) Determine the minimum cycle time based on the test conditions under which the material reaches a steady state.
5) Consider an unbalanced cycle (eg SAE J2527) instead of repeated test conditions (eg 102/18 cycle).
6) If a test method can verify a certain conclusion, the test should be repeatable.
7) Harsh test, adding a test parameter to determine the limit of acceleration.
2.2 Acceleration
100% agreement can be expected when the simulated trials are not accelerated. As acceleration increases, it is clear that consistency decreases accordingly. In a satisfying test, one would expect a direct relationship between consistency and speedup, but typically there are limits. When the acceleration is within a certain limit, the consistency is very good; but outside the limit, the consistency will drop rapidly (as shown in Figure 3). This is because one of the influencing factors is beyond the critical point.
2.3 Outdoor Aging
To determine good data in a meaningful time frame, outdoor testing needs to be properly designed before testing begins. Outdoor weathering procedures also require some advance planning. Before exposure can begin, the purpose of the test needs to be clarified. Relevance can be better ensured by the following suggestions given.
1) Start testing as soon as possible.
2) Use many replicates.
3) Evaluate frequently.
4) At least 5 evaluations in 1 trial.
5) Use control samples or reference samples.
6) Determine the baseline for material aging.
7) Usually it takes 12 to 24 months.
2.4 Permutation and combination of samples
The variable arrangement of typical test materials is unbalanced, so a large amount of original data needs to be deleted and grouped to obtain the aging model. The correct way to do outdoor exposure testing is to expose as many samples as possible, and the faster the test, the better. If the correct experimental design is not done, the redundant experimental data collected after the experiment ends can make the test results look messy. An example of a prospective evaluation of a new test method is the testing of automotive finishes.
Color: red, blue, green, yellow, black, white, gray (7).
Substrate: steel, aluminum, plastic (3).
Adhesives: Adhesive A, Adhesive B (2).
Color coating: reflective, matte, glossy (3).
Outer Coating: Polyester, Silicone (2).
The number of samples is the product of all parameters, in this case 7 × 3 × 2 × 3 × 2 = 252. Outdoor exposure testing with 252 samples is considered a moderately large test, but for many studies, if the experiment Chamber accelerated testing, the number of samples in this group is too large. Therefore, it is necessary to consider reducing the amount of test samples. Select some samples from the above samples, the plan may be as follows: 3 colors, 2 substrates, 1 adhesive, 1 color coating, 2 outer coatings, the total number of samples is 3 × 2 × 1 × 1 × 2 = 12. This combination can be tested first, and if the test results are good, other samples can be further evaluated.
2.5 Balanced Design
The permutations of samples need to be balanced so that meaningful comparisons can be made. In order to obtain meaningful results, individual results should be grouped; for proper operation, all sample types should be identified in any test. Requirements are as follows:
1) each variable in each trial;
2) test the same number of samples in each trial;
3) exposure to the same test period in each test;
4) Use the same evaluation method throughout.
Evaluation times do not have to be evenly distributed. Several evaluations can be done in the early stages of the trial. Because at this stage, the sample performance may change significantly. The use of replicates is very important, as many experiments use only a single sample to represent each experimental variable. Avoid analyzing data at the end of the experiment and discovering that not enough samples were exposed.
2.6 Reference material
Reference materials, also known as control materials, are useful tools for comparing accelerated and field tests. The properties of a reference or control material are known and can be used to verify that accelerated testing produces the same type of aging as outdoor testing. Use reference materials to compare different tests or exposures. Accelerated test results are only valid if the accelerated test produces the same type of aging as the outdoor test. Taking the aging degree of the reference material as the end point of the test, the pros and cons of other materials can be judged.
The test results of reference materials can be used as standard results. If you know the time required for a reference material in outdoor testing to achieve a certain degree of aging, then you can determine the time required to achieve the same degree of aging for laboratory equipment. When the time required to achieve a change varies from test to test or exposure, the response of the reference material can also be used to reset the length of the test.
另外一种最实用的方法是同时曝晒1种好的和1种差的参照材料。当试验中只使用1种参照材料时，测试终点时它的老化性能应该位于1组材料的中间位置。即参照材料的性能不应该是很好的或最差的，有时候很难预测，所以更可靠的方法是选用2种参照材料。如果选用1种相对较好的和1种相对较差的参照材料，就会较容易地确定它们的测试结果介于什么位置，而且还可以了解2种参照材料在不同情况下的差别。
讨论用于加速或户外老化测试中的控制或参照材料时，这种材料不是指标准参照材料，如蓝羊毛或聚苯乙烯薄片。参照或控制材料是产品，这种材料经常被测试，或在最终使用环境中这种已知性能的材料已积累大量的试验数据。
2.7 重复试验
通过确保试验的可重复性以证明试验方法的正确性，需要重复试验的数据来显示试验本身内在的可变性。在确信2种材料的性能存在差异之前，首先确认它们之间的性能差异比试验本身内在的差异大。其次，证明试验可以重复。最后，验证另外一个实验室或技术人员是否可以重复操作这个试验是很重要的。
2.8 未知因素
确定和测量在试验中会出现的未知因素。老化测试的一个现实情况是，试验中未知因素可能发生。因此当意外事件发生时，需要作记录并储存数据。有的时候奇怪的数据会显示一种模式。如果在一个试验和另一个试验，或户外与加速试验之间有不同的情况发生，可能意味着发生了不同的老化模式。
2.9 收集实验数据
正确进行试验，在试验过程中最少保证评估5次。如果最初条件算作1次评估，那么还需要另外4次试验结果，是正确追踪老化进程所需进行的最少次数的评估，遵守以下3个准则。
1）产品的老化模式。这会保证测试结果的可预测性，只有正确的老化模式才是被接受的。
2）老化结果需要相同。如果不同的试验中得到不同的老化结果，试验是无效的。如果在户外测试中发现开裂现象，而在加速试验中发现起泡现象，那么就表明发生了错误的老化模式。
3）需要评估所有已知的老化模式。只评估1种老化模式是不够的。在大多数情况下，多种老化模式并存。保存所有初始数据，特别是当您使用仪器测试颜色变化时。
选择1种符合需要的评估类型。表5是不同评估类型优点之间的比较。表6中的测量方法是类型分析的一个例子，是针对老化样品的。非破坏性测试适合用于评价表面性能，优点是需要的样品数量少。破坏性测试评估材料的内部性能，但是因为在评估过程中样品受到破坏，所以在试验开始时需要多准备一些重复样品。
2.10 加速因子
在一个确定的试验中针对确定的样品来了解它们之间的关系，首先确定加速因子。确定加速因子的步骤如下：
1）使用时间-老化的曲线图；
2）比较加速和户外测试；
3）检查达到相同老化程度所需时间；
4）用排序法验证或比较平均值；
5）如果结果一样，计算加速因子（AF）；
6）加速因子=户外老化时间/加速老化时间。
2.11 数据分析
在老化领域，有2种好的方法来分析数据。
2.11.1 时间-老化曲线图
这种方法只要求绘制出老化结果相对时间变化的曲线图。According to图中绘制的点，可以看出曲线的形状。不是所有的老化都是线性的，事实上大多数老化曲线显示有一个稳定阶段，然后老化明显加速。通过比较相同材料的加速和户外测试的老化曲线，可以确定以下几点：1）曲线形状是否相似；2）可以知道老化机制是否相同；3）试验过程中样品的相关性能是否改变：4）测试过程是否正确。在确定性能差异的试验中最称心的情况是：样品之间的差别较大。样品之间的差别越大，就越容易对它们的相关性能进行排序。
图4和图5中的时间-老化曲线可用于图6中的加速因子的计算。图4、图5和图6中的例子只是说明如何计算加速因子。另外，如果所有样品的加速因子大致相同，那么说明这个比较测试是可取的。如果加速因子是不同的，就说明加速测试不能很好地模拟户外曝晒。
2.11.2 统计分析
所有测试结果都应进行某种形式的统计分析。使用正确的统计分析可以获得如下重要信息：1）测试方法是否正确；2）测试结果是否可靠；3）加速因子的大小。
尽管有很多种统计方法，基于多年老化数据的应用，建议2种简单但很有效的方法：平均值比较方 法（Student t），见 公 式（1）；相 关 系 数 方 法（Spearman’s rho），见公式（2）。二者都是 ASTM 的
G03标委会建议的，可在ASTM G169标准中查到 [4] 。
1）平均值比较
直接比较2组数据：
一个试验与另一个试验之间的不同之处？
两个样品性能有何不同？
产品比较试验：
新的产品比旧的产品好？
是否有重大改进？
Student t试验：
确定2个平均值x1，x2
确定2个标准偏差S1 ，S2
样品个数N 1 ，N2
这种统计方法的主要优点是，它的应用不受样品数量的限制，所以没必要要求2次试验的样品数量需要相同。在每个试验中，当用于比较1种材料、1种颜色，或1种类型时，这种统计方法是很好的的。
2）排序数据
按照性能等级排列样品。
选择可测量目标数据：失光率达到50％的时间、按最大差别排列、按试验结束的时间排列、用于确定加速测试是否可以很好预测真实时间。
Spearman相关系数：
2组排序数据相减；
将差值平方，然后再求所有平方的和；
计算样品的个数N 1 ，N 2 。
在这种统计方法中，允许评为相同等级，但是需要使用半级数值进行计算。这种统计方法只有在2个试验中的曝晒样品个数相等的情况下才适用。当评估2个试验中的多种颜色或多种材料时，这种统计方法是很好的的。
2.12 很好的测试
老化很好的的测试方法是进行几个户外试验和几个加速试验，利用上面提到的统计方法确定结果最真实的测试方法。表7是户外和加速测试现行技术可能性的列表。
不同的户外曝晒场用进而保证测试出所有可能发生的老化模式。当材料发生变化了，它对曝晒条件的反应也随之发生变化。在曝晒新材料时要使用控制样品，并重复曝晒。
Table 8 shows a possible consistent outdoor and accelerated test procedure. In this case, the material is an automotive exterior finish that exhibits gloss and color variation. The manufacturer expects a shelf life of material properties of 5a. Initial testing of the material showed that the aging mode could be simulated by a modified SAEJ2527 test. The new idea is to reduce the cost of coating production, determined to use laboratory tests to see whether the new cheap formula has the same weather resistance as existing products.
3 Conclusion
Outdoor testing is the basis for verifying accelerated testing. Outdoor testing and accelerated testing can be carried out at the same time, coupled with reasonable evaluation and statistical analysis, can increase the reliability of accelerated testing results. Laboratory accelerated testing is one of the effective methods for researchers in the field of developing new products. But to reduce the risk of erroneous conclusions, real outdoor benchmark data is also needed. Accelerated testing needs to be verified with field testing.
