MSA Practical Guide
1. Overview of MSA
MSA (Measurement System Analysis, measurement system analysis) is a statistical study of the measurement system to understand the variation and bias in the measurement system, ensuring the reliability and accuracy of measurement data. MSA is one of the core requirements of the IATF16949 quality management system.
? Core Value: The measurement system is the foundation of quality control. If the measurement data is unreliable, tools such as SPC and CPK will lose their significance. MSA ensures that the "ruler" itself is accurate.
2. Five Characteristics of Measurement Systems
Bias (偏倚)
The difference between the measurement average and the reference value, reflecting the accuracy of the measurement system.
Linearity (线性)
The trend of bias across the measurement range, reflecting the accuracy of the measurement system over the entire range.
Stability (稳定性)
The stability of the measurement system over time, reflecting the long-term reliability of the measurement system.
Repeatability (重复性)
The variation in measurements taken by the same operator on the same part, reflecting the precision of the measuring instrument (EV).
Reproducibility (再现性)
The variation in measurements taken by different operators on the same part, reflecting the differences in operator techniques (AV).
3. Steps for Implementing GRR Analysis
1. Preparation
- Select 10 parts (covering the entire tolerance range);
- Select 3 operators (daily operators);
- Ensure the measuring instrument is calibrated;
- Prepare the GRR data record form.
2. Data Collection
- Each operator measures each part 3 times;
- Measure in random order to avoid memory effects;
- Record all measurement data;
- A total of 90 data points (10 parts × 3 operators × 3 measurements).
3. Data Analysis
- Calculate repeatability (EV);
- Calculate reproducibility (AV);
- Calculate GRR;
- Calculate %GRR and ndc.
4. Result Evaluation
| Indicator | Acceptance Standard | Evaluation |
|---|---|---|
| %GRR | <10% | Excellent (measurement system is acceptable) |
| %GRR | 10%-30% | Acceptable (based on importance) |
| %GRR | >30% | Unacceptable (requires improvement) |
| ndc | ≥5 | Acceptable |
| ndc | <5 | Unacceptable |
4. Bias Study
Implementation Steps
- Select 1 reference part (with a known standard value);
- Measure 15 times by 1 operator;
- Calculate the average and bias;
- Perform a t-test.
Bias Calculation Formula
Bias = Measurement Average - Reference Value
Bias% = |Bias| / Tolerance × 100%
Acceptance Standard: Bias% < 10%
Analysis of Excessive Bias
- Measuring instrument not calibrated;
- Measuring instrument wear;
- Incorrect measurement method;
- Environmental factors affecting the measurement.
5. Linearity Study
Implementation Steps
- Select 5 parts (covering the entire measurement range);
- Determine the reference value for each part;
- Measure each part 5 times;
- Calculate the bias for each part;
- Plot a linearity chart and calculate the slope and intercept.
Linearity Evaluation
| Indicator | Acceptance Standard | Description |
|---|---|---|
| Linearity% | <10% | Good linearity |
| R² | >0.8 | Significant linear relationship |
| Slope | Close to 0 | Bias does not change with the measurement range |
6. Stability Study
Implementation Steps
- Select 1 standard part;
- Measure 3-5 times per day;
- Measure continuously for 20 days;
- Plot an Xbar-R control chart;
- Analyze process stability.
Stability Evaluation
- No abnormal points on the control chart;
- The process is in a statistically controlled state;
- No obvious trends or periodic changes.
Causes of Poor Stability
- Aging of the measuring instrument;
- Environmental changes (temperature, humidity);
- Improper maintenance of the measuring instrument;
- Operator changes.
7. Classic Case in the Automotive Industry
Case: Improvement of a Measurement System GRR
Background: The GRR for a certain hole diameter measurement was 35%, which was unacceptable, leading to a nonconformity during customer audit.
Cause Analysis:
- Inadequate resolution of the measuring instrument (0.01mm, tolerance 0.05mm);
- Inconsistent measurement techniques among operators;
- Non-uniform positioning reference.
Improvement Measures:
- Replace with a high-resolution measuring instrument (0.001mm);
- Standardize measurement techniques and create a work instruction;
- Design dedicated fixtures to standardize the positioning reference;
- Retrain operators.
Effect: GRR reduced from 35% to 8%, ndc increased from 3 to 12, and the customer audit was passed.
8. FAQs
Q1: How often should GRR analysis be performed?
A1: It must be performed in the following situations: 1) First use of a new measuring instrument; 2) After repair of a measuring instrument; 3) Special customer requirements; 4) At least once a year.
Q2: How should %GRR exceeding the limit be handled?
A2: 1) Analyze whether it is a repeatability issue or a reproducibility issue; 2) Poor repeatability → Replace the measuring instrument, improve the measurement method; 3) Poor reproducibility → Train operators, standardize measurement techniques.
Q3: What is ndc? Why is it required to be ≥5?
A3: ndc (Number of Distinct Categories) is the number of categories, reflecting how many different groups of data the measurement system can distinguish. ndc ≥ 5 indicates that the measurement system has sufficient resolution.
Q4: What is the difference between variable and attribute MSA?
A4: Variable MSA analyzes continuous data (dimensions, weight) and uses %GRR for evaluation; Attribute MSA analyzes discrete data (conforming/nonconforming) and uses the Kappa value for evaluation.
Q5: What is the difference between MSA and calibration?
A5: Calibration ensures the accuracy of the measuring instrument (bias); MSA evaluates the variation of the entire measurement system (including people, machines, materials, methods, and environment). Calibration is a prerequisite for MSA.
Q6: Is MSA software necessary?
A6: For small batches, Excel templates can be used; for large batches and multiple measuring instruments, it is recommended to use professional MSA software (such as Minitab) for higher efficiency, automatic calculations, and automatic report generation.