Advancing QE Skills (23) | The Synergy of DFMEA and PFMEA: How Design Changes Affect Processes

A wire harness connector produced by an automotive parts company received a batch of complaints about pin withdrawal in the third month after mass production. Upon disassembly by the customer, it was found that the actual terminal retention force was only 78-86N, below the lower limit of 95N specified in the drawings. The investigation was awkward: six months ago, due to a failed insertion and withdrawal durability test, the terminal retention force requirement was increased from 80N to 100N, and the terminal supplier and crimping window were changed. The design change notification (ECN) went through the process, and the DFMEA was updated; however, the crimping process line in the PFMEA remained unchanged, and the upper and lower limits for crimping height monitoring in the control plan were still set according to the old window. The work instructions, first article inspection records, and inspection tool parameters were not updated. The production line followed the old standards diligently for half a year, and inspections also adhered to the old standards, deeming the products as qualified. The design requirements changed, but the process was completely unaware—when the documents don't match, the on-site operations will inevitably go wrong.

1. The Division of Labor Between the Two FMEAs and the Interface That Must Be Welded

DFMEA and PFMEA are not two parallel documents but the upper and lower segments of a failure chain.

DFMEA answers "why the design would lead to failure": the failure effects are consequences at the vehicle/system level, the failure modes are the specific failure manifestations of the part itself (such as fracture, pin withdrawal, leakage), and the failure causes are located in design parameters (material grade, structural dimensions, tolerance allocation, surface treatment). PFMEA answers "why the process would produce defects": its failure modes should be the specific manifestations of DFMEA failure modes during the product manufacturing process, and the failure causes must be located in process parameters (crimping height, injection molding pressure, drying temperature, welding current).

The interface between the two can be explained in one sentence: The failure modes of DFMEA are the failure effects of PFMEA; the design characteristic requirements of DFMEA are the process characteristics that PFMEA must ensure. Once this interface becomes vague, DFMEA becomes a design analysis that the process engineers cannot understand, and PFMEA becomes a procedural review that is disconnected from the design.

The entity that carries this interface is the special characteristics table. Functional characteristics in DFMEA with a severity (S) ≥ 8, as well as safety characteristics (CC) and critical characteristics (SC) identified by the customer or regulations, must be passed to the special characteristics table in their original form. The special characteristics table then assigns these to the corresponding processes, control plans, work instructions, and inspection standards in the PFMEA. This chain is called the "characteristic transfer chain," and any break in it will expose itself in the form of complaints in a certain batch.

2. The Conduction Path of Design Changes and the Rules for Graded Re-evaluation

A change is not just a signed ECN; its complete path involves seven steps:

Design change proposal → DFMEA update → Special characteristics table update → PFMEA update → Control plan update → Work instruction and inspection standard update → Training, line changeover verification, and first article confirmation.

In practice, the most common breakpoints occur in the last five steps. To avoid treating every change generically, changes can be classified by severity, with specific re-evaluation scopes and timelines defined:

Level 3 changes are the easiest to overlook. Replacing material suppliers or changing auxiliary materials (such as release agents, fluxes, lubricants) may seem unrelated to function, but they can alter friction coefficients, wettability, or weldability, indirectly shifting the process window. Therefore, it is more prudent to treat all such changes as Level 2.

3. Practical Steps and Quantitative Criteria

Step 1: Establish a "Design Characteristics—Process Characteristics" transfer matrix. Using DFMEA as the source, list all items with S≥8 and SC/CC items individually, and mark in the matrix which process and process parameters ensure each item. Criteria: This matrix must cover 100% of the high-risk items in DFMEA, without missing any row; any characteristic not covered is considered to have no control means in the current process and must be suspended or new controls added. After the matrix is established, each change only requires checking this table to identify the affected processes, eliminating the need to review the PFMEA from the beginning.

Step 2: Determine the re-evaluation scope according to the grading rules. The change order must be annotated with the level, the S value or characteristic identifier, and the list of documents that need to be updated. Criteria: Change orders without level annotations will not be approved by the quality department; document updates must be reflected in version numbers, and the version correspondence of the four documents (DFMEA, PFMEA, control plan, work instructions) must be recorded in the change log.

Step 3: Verify whether the failure chain has truly been transmitted. For each process involved in the change, verify the following: whether the failure causes in DFMEA point to design parameters, whether the failure causes in PFMEA point to process parameters, and whether the physical mechanisms are consistent. Also, check whether the detection capability has kept up—after the change, if the required resolution is higher, the existing inspection tools may no longer be suitable, and a new measurement system analysis must be conducted. Criteria: %GRR < 10% is acceptable, 10% to 30% is conditionally acceptable (with a written explanation that it does not substantially affect the judgment), and > 30% is unacceptable, requiring improvements to the measuring tools or methods. For attribute detection (such as visual inspection of appearance), use consistency analysis for verification. Criteria: inspector consistency with the standard Kappa ≥ 0.75.

Step 4: Line changeover verification and closure. The first batch of production after the change must be verified according to the following criteria: one full-size first article + critical characteristic data from three consecutive batches, with no fewer than 50 pieces per batch; critical characteristics Cpk ≥ 1.67, important characteristics Cpk ≥ 1.33, and non-compliance is not allowed, with no substitution of "overall impression of qualification" for data. The change closure requires signatures from design, process, and quality, and the verification data, document version correspondence, and training sign-in records must be archived to ensure traceability to specific production batches.

4. Common Pitfalls

Pitfall 1: Copying design causes as process causes. DFMEA states "insufficient terminal retention force, the cause is low material elastic modulus," and the same is copied into the PFMEA—this is not synergy, but duplication. The failure causes in PFMEA must be process parameters that can be adjusted by operators and process engineers, such as crimping height, crimping die wear, and stripping length. Otherwise, corrective actions cannot be implemented.

Pitfall 2: Updating only DFMEA while leaving downstream documents unchanged. The pin withdrawal case at the beginning is an example of this. To determine if a company has this issue, simply review 5 ECNs from the past 12 months and check if each has corresponding PFMEA and control plan version update records.

Pitfall 3: Believing that small changes do not require re-evaluation. Especially for material and auxiliary material changes. A small change in quantity does not equate to a small risk. Failures due to "design changes without process changes" are often more hidden than obvious major changes because they do not surface during first article inspection.

Pitfall 4: Ignoring the transfer of detection (D). The design side has reduced the occurrence (O), but the process side still uses the old detection methods, with the D value retaining the old score. The risk score may look good on paper, but it is just a number. The score should be re-evaluated based on the actual changes in detection methods.

Pitfall 5: Using "documents updated" to replace "process verified." Updating documents only completes half of the closure. Closing a change without supporting data on process capability and measurement systems is just a paper closure.

5. Self-Check List

• □ Are all items with S≥8 and SC/CC in DFMEA individually listed in the transfer matrix, PFMEA, and control plan, with a coverage rate of 100%?
• □ Does each design change order from the past 12 months have corresponding version update records for PFMEA, control plan, and work instructions, or a written "reason for no change"?
• □ Have the key processes involved in the change completed process capability verification (Cpk ≥ 1.67 for critical characteristics, Cpk ≥ 1.33 for important characteristics) and measurement system confirmation (%GRR < 10%, attribute consistency Kappa ≥ 0.75)?
• □ Are the failure causes for the same characteristic in DFMEA and PFMEA located in design parameters and process parameters, respectively, without mutual duplication?
• □ Does the change closure record include signatures from all three parties, document version correspondence, and verification data, allowing traceability to specific production batches?

Design changes should not stop at the drawings; they must be transmitted all the way to the production line.

Knowledge code: 8.2.1

Version: v20261003

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.