结论：复现成功才能确认原因，复现不出来也是重要信息

轮椅在使用中出现问题——车架开裂、制动失效、电池异常、部件脱落、使用者跌倒。

调查这类事件时，最关键的一步是确认原因。

而确认原因的可靠方式是：在实验室里把失效复现出来。

复现的意义在于：

验证假设。 如果假设的条件能复现失效，假设就得到支持。

排除其他可能。 复现过程中可以排除不成立的假设。

确定边界。 什么条件下会失效、什么条件下不会。

验证整改。 改进后在同样条件下不再失效，说明整改有效。

提供证据。 复现的过程和数据是有说服力的证据。

复现前的信息收集

复现的成败很大程度上取决于信息的完整性：

信息类别 具体内容
产品信息 型号、批次、生产日期、序列号
使用信息 使用时长、频率、使用者体重
环境信息 使用场所、路面、气候
事件信息 发生时间、当时动作、有无异常预兆
失效件 失效部件的实物
维修史 是否维修过、改装过
同批情况 同批次产品有无类似问题

「失效件实物」是最有价值的信息 ——断口的形貌能说明是疲劳、过载还是材料缺陷，而这些信息在照片上往往看不清楚。

建议在事件调查中尽可能保留失效件原状，不要清洗、不要拼接、不要二次破坏。

断口与失效面的分析

在设计复现方案之前，先分析失效件本身：

断口形貌。 疲劳断口有特征的贝纹线和光滑区，过载断口是韧窝或解理。

起裂位置。 从哪里开始，通常是应力集中或缺陷处。

扩展方向。 裂纹的走向。

变形情况。 有无塑性变形，反映载荷性质。

表面状态。 腐蚀、磨损、划伤。

材料检查。 材质、组织、硬度是否符合。

「断口分析先于复现」是正确的顺序 ——断口告诉你失效的性质（疲劳还是过载），而失效性质决定了复现方案该怎么设计。

复现方案的设计

确定假设。 基于信息分析提出可能的原因。

设计条件。 设计能验证假设的试验条件。

接近实际。 载荷、频率、环境尽量接近实际使用。

必要的加速。 用加速条件缩短时间，但要注意不改变失效机理。

对照组。 同时试验正常产品作为对照。

分级验证。 从最可能的假设开始。

记录充分。 全过程记录，包括未出现失效的条件。

「对照组」很重要 ——只做失效品的试验，无法判断结果是异常还是正常水平，有对照才能做出比较。

常见的失效类型与复现

疲劳断裂。 用疲劳试验复现，关注循环次数和载荷水平。

过载断裂。 用静载试验复现，关注失效载荷。

焊缝失效。 关注焊缝质量和应力集中。

紧固件松动。 用振动试验复现。

制动失效。 复现制动条件，关注磨损和调整状态。

电气故障。 复现工况，关注温升、振动、湿度的影响。

电池异常。 复现充放电条件和环境条件。

「紧固件松动」的复现需要振动加时间 ——单次试验可能复现不出来，需要模拟实际的振动谱和累计时间。

复现不出来时

复现失败很常见，这时候不应当直接下「产品没问题」的结论：

检查条件是否覆盖。 实际使用的条件可能超出了试验条件。

考虑偶发因素。 个别缺陷、偶然的极端载荷。

考虑组合因素。 多个因素叠加才导致失效。

考虑时间因素。 累计的时间效应难以完全加速。

考虑使用偏差。 实际使用方式与预期不同。

扩大条件范围。 加大载荷或延长时间看是否出现。

「复现不出来」本身也是信息 ——它说明在常规条件下产品是可靠的，问题可能来自个别缺陷或特殊使用条件，调查方向应当转向这两处。

整改后的验证

用同样的条件验证。 改进后在导致失效的条件下试验。

验证失效是否转移。 加强了一处，失效可能出现在另一处。

验证裕度。 改进后的裕度有多大。

批量验证。 不只验证一件，验证改进的一致性。

长期验证。 必要时做更长周期的验证。

「失效转移」要特别关注 ——把断裂的管子加粗了，载荷可能转移到焊缝或连接件上，新的薄弱点出现。

常见的准备不足

信息不全。 材料、工艺、使用环境的信息缺失，导致方案设计缺乏依据。

样品不具代表性。 试制样品或挑选样品，与量产状态有差距。

条件未明确。 测试条件没说清楚，数据的适用范围就不明确。

只看初始值。 忽略了环境和时间作用后的性能变化。

项目不完整。 遗漏了必要的项目，提交时才发现。

这五项在实际委托中出现频率较高，建议在项目启动时逐项确认，因为它们造成的返工成本远高于事前确认所花的时间。

数据的组织与保存

记录的完整性。 条件、方法、设备、样品状态、实测值。

可追溯性。 数据与样品批次、生产批次的关联。

保存期限。 覆盖产品的生命周期。

检索的便利。 需要时能快速定位。

版本的管理。 数据更新时的版本控制。

建议建立统一的数据台账，记录每组数据的来源、条件、适用范围和有效期。数据积累起来之后，后续的产品开发和变更评估都能受益。

我们的做法

接到失效分析和复现的委托时，我们会先做失效件的断口和状态分析，因为断口能说明失效的性质，这决定了复现方案的设计方向。

我们建议委托方尽可能提供完整的背景信息和失效件原件 ——使用条件、使用者信息、事件经过，这些信息对设计有效的复现方案作用很大。

复现试验我们会设置对照组，并完整记录包括未出现失效的条件，因为「什么条件下不失效」同样是有价值的边界信息。

整改后的验证我们会关注失效是否转移，而不只看原来的位置是否还失效。

如果你有轮椅的失效需要分析或复现，可以把失效件和背景信息一起送来，或者直接联系：132 4819 8029。检测能力见服务介绍，产品分类见手动轮椅检测与电动轮椅检测，案例见案例。

English version

Conclusion. Wheelchairs develop problems in service: frames crack, brakes fail, batteries behave abnormally, components detach and users fall. The critical step in investigating such events is establishing the cause, and the reliable way to do that is reproducing the failure in the laboratory. Reproduction serves several ends. It tests hypotheses, since a hypothesis supported by conditions that reproduce the failure gains weight. It eliminates alternatives that do not hold. It establishes boundaries, showing under what conditions failure occurs and under what conditions it does not. It verifies corrective action, since a modified product that no longer fails under the same conditions shows the action worked. And it provides evidence, since the process and data of reproduction are persuasive.

Gathering information first. Success depends largely on how complete the information is. Product information covers model, batch, production date and serial number. Usage information covers duration of use, frequency and the user's mass. Environmental information covers the setting, surfaces and climate. Event information covers when it happened, what was being done and whether there were warning signs. The failed part itself. Repair history covers whether it was repaired or modified. And batch experience covers whether similar problems occurred in the same batch. The failed part is the most valuable of these: fracture surface morphology distinguishes fatigue from overload and from material defects, and photographs frequently cannot show it. Preserve failed parts as found during investigation, without cleaning, reassembling or further damage.

Analysing the fracture surface. Analyse the failed part before designing the reproduction. Fracture morphology distinguishes fatigue, with its characteristic beach marks and smooth region, from overload with its dimples or cleavage. Initiation location shows where it started, generally at a stress concentration or a defect. Propagation direction shows the crack's path. Deformation indicates the nature of the loading through the presence or absence of plasticity. Surface condition covers corrosion, wear and scoring. And material checks cover composition, microstructure and hardness against specification. Fracture analysis before reproduction is the correct order: the fracture surface reveals the nature of the failure, whether fatigue or overload, and that nature determines how the reproduction should be designed.

Designing the reproduction. Frame the hypothesis from the information analysis. Design conditions that test it. Stay close to reality in load, frequency and environment. Accelerate where necessary to shorten the time, taking care not to change the failure mechanism. Include a control, testing normal product alongside. Verify in order of likelihood, starting with the most probable hypothesis. And record fully, including conditions under which failure did not occur. The control matters: testing only the failed item leaves no way to judge whether a result is abnormal or normal, and comparison requires a control.

Common failure types and their reproduction. Fatigue fracture is reproduced by fatigue testing, with attention to cycle count and load level. Overload fracture is reproduced by static loading, with attention to the failure load. Weld failure needs attention to weld quality and stress concentration. Fastener loosening is reproduced by vibration testing. Brake failure is reproduced by recreating braking conditions, with attention to wear and adjustment. Electrical faults are reproduced by recreating operating conditions, with attention to temperature rise, vibration and humidity. And battery anomalies are reproduced by recreating charge-discharge and environmental conditions. Reproducing fastener loosening takes vibration plus time: a single test may not reproduce it, and the actual vibration spectrum and cumulative duration need simulating.

When reproduction fails. Failing to reproduce is common, and it does not warrant concluding that the product is sound. Check whether the conditions covered the case, since actual use may exceed the test conditions. Consider isolated factors such as an individual defect or an occasional extreme load. Consider combined factors, where several must coincide. Consider time effects that acceleration cannot fully capture. Consider deviation in use, where actual practice differs from expectation. And widen the conditions, increasing load or extending time to see whether it appears. Failure to reproduce is itself information: it indicates the product is reliable under ordinary conditions, and the problem may come from an individual defect or unusual use, which is where the investigation should turn.

Verifying corrective action. Verify under the same conditions that produced the failure. Check whether the failure has migrated, since strengthening one location may move it elsewhere. Check the margin the improvement provides. Verify across a batch rather than a single item, to confirm consistency. And verify over a longer period where necessary. Migration deserves particular attention: thickening a tube that broke may transfer the load to a weld or a fitting, where a new weak point appears.

How we handle it. For failure analysis and reproduction work we analyse the failed part's fracture surface and condition first, since it reveals the nature of the failure and determines how the reproduction should be designed. We ask clients to provide background information and the original failed part in as much completeness as possible, covering conditions of use, user information and the sequence of events, since those contribute a great deal to designing an effective reproduction. We include a control in reproduction testing and record fully, including conditions under which failure did not occur, since knowing where failure does not occur is valuable boundary information. And in verifying corrective action we look at whether failure has migrated rather than only at whether the original location still fails.

If you have a wheelchair failure needing analysis or reproduction, send the failed part with the background information: +86 132 4819 8029.