DFMEA Completed, but Verification Plan Starts from Scratch? —— A Five-Step Approach to Linking Failure Causes to DVP&R

1. DFMEA Filled with Measures, but Sample Verification Only Conducts General Tests

A car parts manufacturer discovered a design issue during a post-production review. The connection clips of an injection-molded casing were found to break in batches during winter in northern customers' locations, with the fracture surfaces turning white, a typical case of low-temperature brittleness.

Checking the DFMEA records: The relevant row listed the failure mode as "clip fracture, loosening after assembly," the failure cause as "insufficient material toughness at low temperatures," the preventive action as "select a toughened grade and confirm the material data sheet," and the detection action as "design review to confirm the material grade and low-temperature performance." The risk assessment was S=8, O=4, D=5, with the AP being M, indicating that all necessary steps were taken.

Checking the DVP&R (Design Verification Plan and Report): There were only four entries — high and low-temperature storage, temperature cycling, random vibration, and appearance and dimension inspection. The tests were conducted according to industry general standards, with three samples for each test, and all concluded as passed.

Each record makes sense on its own, but together they reveal a problem: The failure mechanism identified in the DFMEA is "the material's toughness decreases at low temperatures, leading to fracture under structural stress," yet the verification plan does not include any tests that apply assembly and disassembly forces to the clips under low-temperature conditions. Low-temperature storage can only prove that the parts do not deform; it cannot prove whether the clips can still hold at minus twenty degrees. The risk was identified but not verified.

This is not an isolated incident: DFMEA and DVP&R are often completed by different people in different phases. The former is archived once submitted, and the latter is created based on a standard checklist, breaking the critical link between failure causes and verification items.

2. The Source of DVP&R Items is the Failure Causes in DFMEA

To understand the relationship between the two, remember this: DFMEA is responsible for explaining "how it might fail," while DVP&R is responsible for proving "it indeed won't fail that way." The output of the former is a list of risks, and the output of the latter is a list of evidence, with the connecting piece being the two types of actions in DFMEA.

Preventive actions answer "how to prevent it from happening." The nature of these actions is to reduce the occurrence (O), and the corresponding verification includes analytical and review-based evidence: simulation analysis, tolerance chain calculation, material performance data, historical data comparison, and design reviews with decision criteria. The most common issue here is the phrase "design review confirmation" — the review itself is not evidence; the conclusions and their basis are.

Detection actions answer "how to detect it if it happens." The nature of these actions is to reduce the detection (D), and the corresponding verification includes physical tests: bench tests, environmental and life tests, limit and destructive tests, and fault injection tests. If these actions are not translated into specific test items, the D score is merely theoretical.

From this, we can derive a mapping relationship:

It is evident that DVP&R items should not be copied from standard checklists but should grow from the failure cause list in the DFMEA. Standard tests verify general environmental adaptability and do not cover the specific failure mechanisms of this product.

3. A Five-Step Approach to Linking DFMEA to DVP&R

Step 1: Extract the verification list. From the DFMEA, select two types of rows: those with a severity (S) ≥ 8 or an AP of H, and all rows with detection actions. Copy the failure modes, failure causes, preventive actions, detection actions, and current verification status into a simple table. This table serves as the input for all subsequent work and typically contains only a few dozen rows.

Step 2: Categorize actions into three types of verification. Analytical (simulation, calculation, tolerance analysis, material data), review (design review, expert review, historical data comparison of similar products), and test (bench, environmental, life, limit, fault injection). Two rules apply: each action must have one and only one verification category; statements like "confirmed through design review" or "confirmed with supplier" that lack decision criteria and evidence links must be rewritten.

Step 3: Define the five elements for each test item. Test object (sample status, batch, quantity), test conditions (load, temperature, number of cycles, must include specification boundaries and beyond), criteria for judgment (quantified failure thresholds, derived from specifications or design goals), sample size basis (determined by severity and risk, for items with S ≥ 9, at least three batches are recommended, with five samples per batch, including extreme value samples for parameter upper and lower limits), and record requirements (raw data, failure modes, failure times). Missing any of these five elements makes the test non-reproducible.

Step 4: Establish a traceability matrix. Each row of DFMEA failure causes should correspond to at least one DVP&R project number. The DVP&R report should backfill the conclusions and raw data. Any failed items must return to the DFMEA for updates — design changes, material changes, or detection method changes are all acceptable, but the verification conclusion cannot be marked as "pending observation." This matrix can be built using project numbers for bidirectional referencing without the need for complex software.

Step 5: Link review and change. After design changes, material substitutions, or supplier changes, first determine "which failure causes are affected," then decide whether to add verification items to the DVP&R. Failed or pending items in the DVP&R must be included in the production part approval process as part of the design records, and they cannot enter mass production with a "pending verification" status. The closed-loop check involves three questions: For each failure cause with S ≥ 8, is there a corresponding verification item? For each verification item, is there traceable raw data? For each failed item, is there an update record in the DFMEA?

4. Three Common Misconceptions

Misconception 1: Treating general test standards as the verification plan. "Conducting high and low-temperature, temperature cycling, and vibration tests according to standards" verifies environmental adaptability, answering "what environments the part can withstand," not "whether the specific failure mechanisms of this design will occur." The former is the baseline, while the latter is the primary task of DVP&R.

Misconception 2: Closing high-risk actions with design reviews. Reviews can close review-based actions, but failure causes such as material properties, structural strength, and tolerance accumulation must be supported by analytical or test data. Closing risks with S ≥ 8 with a simple "review confirmed" is equivalent to moving the risk into mass production.

Misconception 3: Conducting DVP&R only once. Archiving the verification plan after the sample stage and not triggering additional verification for design changes, material substitutions, or process changes during mass production is a common path for design issues to concentrate and emerge. Equally common is the mismatch between sample size and severity — a failure mechanism with S = 9 is verified with only one sample.

5. In Summary

DFMEA delivers a list of "how it might fail," while DVP&R delivers evidence that "it indeed won't fail that way" — when the numbers align, design verification is not just a formality.

Whether verification items are copied from standard checklists or derived from failure causes determines whether DFMEA is a genuine analysis or just a form-filling exercise.

Knowledge code: 8.2.1

Version: v20260913

Author: QTank QTank is dedicated to providing systematic professional knowledge, methodologies, and practical tools for quality management practitioners, helping companies continuously improve their quality capabilities.