Quality Function Deployment With a Worked Matrix

Quality function deployment (QFD) is a planning method that translates customer requirements into measurable technical characteristics and target values, then carries those targets into part design, process planning and production. Its best-known tool is the House of Quality, a matrix that links what customers want to how the product delivers it.

John Hauser and Don Clausing traced the House of Quality in the Harvard Business Review in 1988 back to Mitsubishi's Kobe shipyard in 1972, from where Toyota and its suppliers developed it further. iSixSigma credits Yoji Akao with developing QFD in Japan in the 1960s.

The House of Quality, Part by Part

The Institute for Manufacturing at the University of Cambridge describes the House of Quality as six connected parts.

On its decision support page, the IfM adds that the customer weights together add up to 100 %. It lists five steps: establish the customer attributes and their weights, identify the design parameters, fill in the relationship matrix, compare customer perception with competing products, and analyze the roof.

Work Through a Weighted Relationship Matrix

The housing, weights and scores below are made up to show the arithmetic. A housing for an electronic control unit is fitted by hand on the customer's assembly line. The team has three requirements with weights adding up to 100 %: easy fitting 40 %, reliable sealing 35 % and quick inspection 25 %. It chooses three engineering characteristics: insertion force in newtons, seal compression in percent and inspection clearance in millimeters. The example scores strong, moderate and weak relationships as 9, 3 and 1, and no relationship as 0.

Each column multiplies the relationship scores by the weights and adds them up. Insertion force: 40 × 9 + 35 × 3 + 25 × 0 = 465. Seal compression: 40 × 3 + 35 × 9 + 25 × 1 = 460. Inspection clearance: 40 × 1 + 35 × 0 + 25 × 9 = 265.

Insertion force and seal compression differ by 5 points out of 1,190. Swap the weights of fitting and sealing to 35 % and 40 %, and the ranking flips: insertion force falls to 35 × 9 + 40 × 3 = 435, seal compression rises to 35 × 3 + 40 × 9 + 25 × 1 = 490. The IfM warns that QFD can become too analytical, with a number treated as right when it is not. When a small weight change reverses the order, treat both characteristics as top priority and check the weights against customer data before you commit design effort.

Use the Roof to Find Conflicts

Insertion force and seal compression pull against each other. More compression improves sealing but raises the force the line worker needs. In the roof, this is a negative correlation. The team therefore sets a target pair instead of two separate targets, for example 20 to 30 % compression with no more than 40 N insertion force. A leak test and an assembly trial on prototypes confirm or move both values; the matrix alone cannot set them.

Inspection clearance has no conflict in the roof and a clear lowest priority. It still gets a target, because the customer's line has to inspect every housing it fits.

From Design Targets to the Shop Floor

After the House of Quality, many teams chain further matrices. iSixSigma lists four, with the House of Quality as the first; the later ones carry the targets down to part characteristics and into process and control planning. For the housing, seal compression becomes a gasket hardness and a groove depth, and then a press-fit force on the assembly station. The targets feed the process FMEA and the control plan, which in automotive supply chains belong to APQP; the FMEA page explains that step.

Production then tests the assumptions. If housings fail the end-of-line check in your own plant mostly for sealing reasons, the seal compression target needs another look. Rework booked by reason and the first pass yield of that station show whether a change to the target helped.

What SYMESTIC Contributes to QFD

The matrices are built in a spreadsheet or a QFD tool. In the Cloud MES SYMESTIC, you see how the chosen targets behave once the product is in series production. Quality inspections are recorded at the workstation per part with serial number as OK or NOK, with the defect location on the product image and a reason from the defect catalog. Scrap and rework are evaluated by reason, with the top defect reasons, the defect rate in percent or PPM and the first pass yield for serialized parts.

More on the pages on quality control software and product quality.

Sources

• University of Cambridge, Institute for Manufacturing: Quality Function Deployment (decision support tools)
• University of Cambridge, Institute for Manufacturing: Quality Function Deployment (QFD), tools and techniques
• John R. Hauser, Don Clausing: The House of Quality, Harvard Business Review, May 1988
• iSixSigma: Quality Function Deployment (QFD)