Does Honda Use SPC? (With Kevin Hop)

20 years ago, Honda stood out through its reputation for quality. Outsiders were studying Honda’s approach and Youtube now offers several videos shot at that time about it. Today, quality is no longer the differentiator among carmakers that is used to be but the practices of a company like Honda — past and present — remain a worthwhile object of study.

Industry Developments Since 2000

Carmakers watch the JD Power Initial Quality Studies (IQS) closely as benchmarks on the latest cars, primarily about manufacturing quality. They follow in particular the number of problems experienced in the first 90 days of ownership per 100 vehicles, called “PP100.”

Infotainment As A Source of Quality Problems

Of interest in the IQS are both its absolute value and the rank for a model or a brand. In absolute terms, the IQS is worse in 2020 than in 2000. The top-performing model was then the Toyota Cressida, with 63 PP100. Today, it’s the Chevrolet Sonic with 103 PP100. Most of the problems reported in 2020, however, are with Infotainmemt features that did not exist in 2000 – that is, voice recognition, connectivity, touchscreens, navigation systems, or Bluetooth.

Quality No Longer A Differentiator

In rankings, American, European, and Korean makers have clearly caught up with Toyota, Honda, and other Japanese automakers. Press accounts confirm that Honda has slipped out of the “top 10” car makers in quality ratings from J.D. Power and Associates and that Honda’s top management has acknowledged facing a crisis. CEO Takahiro Hachigo blames the worldwide proliferation of products and options, and has a strategy of reducing the mix to have more common components. It suggests they are working to make the challenges easier but don’t plan to change the company’s approach to managing quality.

A Video about SPC

In the LinkedIn discussion of More About Toyota and SPC, I pointed out that Honda did not use SPC. This was based on my experience of providing training at their engine plant in Anna, OH, and personal communications with both American employees and Japanese expatriates.

Kevin Hop about SPC at Honda

According to my colleague Kevin Hop, who spent 11 years at Honda in engineering and management positions including supplier quality:

“You might see a little ‘SPC’ being used on a kind of personal level by engineers – their Japanese mentors would have encouraged them to try various approaches – or at least not gotten in their way. But as a company systematic approach, you saw little if any being used.
Honda’s philosophy was to get processes under control with C_{p_k} \geq 1.3 and let it run. Then they would just control input parameters of machines using PQCT– Process Quality Control Tables – quality operation standards for the process derived from an FMEA.
For example, to control a machined hole size, you initially prove that the machine can give you C_{p_k} \geq 1.3, and then you control the 4 key parameters: feed, speed, tool change frequency, and coolant ratio.”

The McGraw Hill video from 2000

As evidence of SPC at Honda, Allen Scott referred me to the following video:

It’s in Vol. 7 of an Operations Management Video Series developed for Irwin McGraw-Hill. It’s been available on YouTube since 2014 but the copyright is from 2001. Based on the Civic models in it, they shot it in 2000.

McGraw-Hill offers manuals and videos for an Introductory Operations Management Course, but does not include the videos from 2000 currently on YouTube in their current products.

This video is not a Honda document but a document Honda helped with. They let the author shoot in East Liberty facility and granted access to employees. The setting is Final Inspection of cars and the main narrator is Sam Kennedy who was then Associate Chief Inspecting Engineer.

SPC at Final Inspection?

Final Inspection is an odd place to apply SPC, because it takes place after the entire process is completed. Theoretically, SPC is supposed to happen in-process, based on observations of workpieces right after they complete an operation. Then it prompts immediate action as needed.

Per Kevin Hop:

“This video is not real SPC – it is what we called Product Inspection. Honda did have a policy of sampling products coming off the line just as a kind of audit. The purpose was only to catch a catastrophic running repeating failure that was somehow escaping the normal quality control system. A kind of insurance policy against a huge amount escaping.”

What Final Inspection is For

Carmakers generally don’t use Final Inspection as a substitute for in-process controls as suggested in this video. Final Inspection is too late to discover that a fixture broke. When it happens, the team in charge of the operation should notice it and fix or replace it.

The primary purpose of Final Inspection is to prevent defective goods escaping into the market. As it is impossible to design non-destructive tests for every possible use of a product, it can never be perfectly thorough, and the tests need to be structured as a decision tree that leads to fast identification of defectives. The goal is to ship as quickly as possible all  the units that Final Inspection “fails” to find defective.

Once Final Inspection brands a unit as defective, it becomes the object of Failure Analysis, with the purpose of identifying all its defects and their origins in the process or the supply chain. It results in quality problem reports (QPR) sent to the appropriate departments or suppliers to take immediate countermeasures, identify and eliminate root causes. See Final Inspection and Failure Analysis for a discussions of methods.

Along the way, Final Inspection and Failure Analysis glean data about the plant’s quality performance overall and process capability in particular. While this data is useful, it is a by-product.

The Door Closing Case

The example shown in the video is the measurement of the minimum speed required for a door to shut properly. The operator manually closes the door, with a sensor recording the closing speed and observes whether it closes. In 10 to 15 tries, the operator estimates the minimum closing speed to within ±0.1 km/hr and punches the numbers into a handheld data acquisition system.

Customers open and close car doors frequently, and defects in this area hurt the perception of a car’s quality. It is not a minor issue and, in fact, the next international conference on Doors and Closures in Car Body Engineering is scheduled for 11/20 in Germany.

The video was shot in 2000. The first question is whether they still use the same method in 2020? The data is valuable but its collection is labor-intensive and manual data entry is error-prone. As of 2017, Nissan used a robot to test the durability of car doors which, presumably they can use to measure and automatically record the minimum closing speed with fewer tries and more consistency than a human:

Kevin also found a device that can be used in assembly to control the location of this striker plate:

Is the Door Closing Case an Example of SPC?

Back at Honda in 2000, the video shows software plotting the sequence of minimum closing speeds on an X-chart and in a histogram of the last two weeks of data. The chart in the video shows one value out of control limits. The  search for an assignable cause then turns up a broken fixture used to mount the striker on the car body.

The narrator doesn’t say anything about sampling. We don’t know how many of the cars theymeasure nor how they select them. There are, however, clues in the video itself:

1. Data collection does not look set up for one car every minute, and an inspector who would repeat the motions seen on the video at this pace would end the day with back pain.

1. The histogram of the values of the past two weeks looks based on tens of points, not thousands.

1. The story of the striker mounting fixture is not consistent with 100% inspection. Assume an assembly line with 200 stations, with the striker mounted at operation 180. If the fixture breaks and no one notices, 20 cars will go through with a wrongly mounted striker before the first defective rolls off the line. Why then does the X -chart show only one defective? If they measured all cars, you would expect the 19 cars made between the breaking of the fixture and Final Inspection on the first one to fail as well.

Per Kevin’s recollection:

“The 80 or so data points on the chart represent one point every 2 hours. That would be a typical way they would have sampled. There would be 160 hours of production for a 2 shift operation which they were. So 80 data points for 2 weeks.”

The Video’s Message

The soundtrack ends with “By teaching their employees to understand statistical process control, and immediately counteract out of control situations, Honda prevents problems from recurring in manufacturing.” These words don’t match the above case, which shows measurements on a sample during Final Inspection as triggering a delayed reaction.

In this video, the authors borrowed Honda’s credibility to make their teaching points, and Honda played along. It’s not a video by Honda to train its own associates. The same volume from McGraw-Hill contains another video of the same vintage about Honda Quality in general, which does not mention SPC once.

#honda, #spc, #quality, #finalinspect