Panic in the Face of A Quality Issue - Fuk

Yesterday I posted about the incorrect report of a huge radiation spike at the Fukushima plant.

Now I am reading some accounts from Japan and in addition to the handling of wrong data I am also struck by the common reaction to a quality problem. Shigeto Tanaka of the Yomiuri Times has an account from the area while the alerts were broadcast.

While this reporter was covering the damage by Friday's series of tsunami at a settlement along the coast in Minami-Soma, Fukushima Prefecture, about 25 kilometers north of the nuclear plant, a message was sent around 4 p.m. by disaster response radio broadcast from the Minami-Soma municipal government. The message said: "We have been informed that the Fukushima No. 1 nuclear plant has exploded. Don't go outside."

Listening to the message, about 10 fire company volunteers, who were on alert at the settlement, became panicked and hurriedly left by car, shouting to each other, "Get out of here now!"

This reporter was unable to make a quick decision whether to evacuate.

Today the BBC reports that "The Japanese government has attacked the operator of a crippled nuclear plant for "unacceptable" mistakes."

I have seen panicked reports of a quality issue from both the customer and the supplier side. Panic never really helps the situation. We all know this, but in the heat of the crisis if you do not have some experienced, strong, players you can get into a bad situation. A lot of time can be wasted trying to react to false information.

By the way, one quote got me, "The government understood the workers were overworked and under stress, he added." That reminds me of the saying, "you are only as good as your last f*ck up."

Measurement Error at Nuke Disaster

Here is another "ah ha" moment. Earlier this week officials in Japan reported a radiation spike at the Fukashima nuclear plant that caused workers to flee and international alarm. Today the officials reported that the data was a mistake:

"The number is not credible," said Tokyo Electric Power Co. spokesman Takashi Kurita. "We are very sorry."


How often does this happen in normal factory operations? During PPAP or other quality data reviews we often see odd or out of place data. Sometimes the error is due to simple transposition or data entry errors. Sometimes the data is taken in an incorrect way. So, what to do?
First, data that looks out of place, or extreme, should be reviewed immediately. If the data is correct, then react appropriately. The point here is to treat unexpected data urgently. Verify it. If it is true you are going to want to problem solve it. If the data is wrong you want to correct it as close to the time it was taken to be able to capture the samples and avoid costly inappropriate response.

The Hollowing of the Middle

David Leonhardt at the NYT has as article about a recent paper by economist David Autor. I cannot read Autor's original paper, but Leonhardt's article is worth a read.
Over the course of the past 20 years the middle class has seen declines in opportunities while low and high paid jobs have increased.
In the last two years especially I think this is pretty clear. We continue to lose good factory jobs with good wages for line workers and salaries for shop management. The costs for factory jobs are just too low outside the US/Canada. HOWEVER, Autor's data indicates that while middle wage workers did fare worse than high wage workers, the middle wage workers fared better than low wage workers.

On wages — the most important factor for most workers, since the vast majority of workers remain employed even when unemployment is high — middle-wage workers have fared worse than high-wage workers but better than low-wage workers. Rising inequality, rather than a hollowing out of the middle, seems to be the best summary of recent wage trends.

Consider the change in inflation-adjusted wages by education attainment from 2007 to 2010:

Magnified Inspection

One of the countermeasures to weak PFD/PFMEA/PCP that I showed in an earlier post is magnified inspection.
The intent of specifically adding an exercise like this to to help highlight the need for an internal non-production department close inspection of the parts being produced. I am not talking about 100x inspection, really just 10X or slightly higher. The goal is to find obvious defects. The inspector should not be targeted toward specific "high risk" areas. I find that the high risk areas, or areas that the inspector would be told to look at, will already be in good control. The concern is the "unknowns". Have the inspector slowly scan around the part looking for "issues". The issues can be anything unusual. I know that this all sounds vague, but there can be some good learnings from this type of exercise.



In some cases, if you have a good magnifier with a display screen a small team can participate in the review. Treat it like a brainstorming session. Things like "discoloration" can be noted, for example. In a real world example, the discoloration was due to feed and speed on a drill. The discoloration was caused by heat and the heat caused micro cracks. the micro cracks probably could have also been identified in this inspection. But noting the discoloration could have driven the machining team to identify the cause and the fix. The damage would have also been more likely to have been identified at that time.


Picture of "Overlapping semi-circular markings and a wrinkled surface texture adjacent to the fracture surface were indicative of cold flow"


Remember, the focus is on the finding the unknowns before your customer does.

The Cost of Manufacturing in the US

I had a link to an article about photovoltaic cell production in the US and how the costs were just too high to compete with the surging supply base in China.
I am not enough of an expert to understand exactly what the cost differences are between the US and China. I know labor cost is a chunk of the cost, but what percentage of the production cost is labor and what other big differences are there? My interest is not that I am focused on producing in the US, rather it is on what are the competitive pressures to overcome whether in the US or anywhere.

The Seeking Alpha website, which is a stock analysis site, recently had an article about the recent ISM survey regarding the US manufacturing status. The author, Ronald Rutherford, walks through some analysis of the survey and other data. It is a good article, worth reading.

Some key points: China manufacturing growth is slowing, some consider that a bearish sign for the world economy. Ronald ponders whether that is true. US manufacturing is working hard, costs of overseas components are rising (partly due to weaker $).

The actual survey ressponses quoted are very interesting. I wish I was an ISM member:

Manufacturing: # "Our plants are working 24/7 to meet production demands." (Fabricated Metal Products) # "Capital projects moved from inactive to active" and "Award of long-awaited contracts."

ISM is the Institute for Supply Management:

ASQ Automotive Division Symposium

The ASQ Automotive Division is hosting a symposium on April 4, 2011. If your live around Detroit you may be interested. I went to the event last year and had an enjoyable and informative day.

Here is the speaker list:
Dan Nicholson, Vice President Global Quality, General Motors
Keynote Presentation

Mike VanNieuwkuyk, Executive Director Global Vehicle Research, J.D. Power and Associates
Consumer Aligned Product Development in a Quality Equality Market

JD Marhevko, Lean and Quality Independent Consultant and Chair, ASQ Quality Mgmt. Division
Sample Size of One: Quality & Lean Tools with a Kid on the Autism Spectrum

Vince Mastrangelo, Vice President Quality, Continental Automotive
Assuring the Quality of Innovation, Continental Automotive Group: Delivering Invention to our Customers

Bill Hampton, Publisher, AutoBeat Group
Where is our Industry Headed?

Jane Hoying, Consulting Manager, Shainin, LLC
TransaXional Problem Solving – A New Paradigm


Here is other information:
Date: Monday, April 4, 2011
Time: 7:30 am – 8:30 am: Registration & Breakfast
8:30 am – 5:00 pm: Symposium & Networking

Location: Macomb Community College (Center Campus)
University Center UC 1 - Assembly Hall
44575 Garfield Rd, Clinton Township, MI 48038-1139
Cost: Before Mar 25, 2011 - $30 for ASQ Automotive Division Member; $45 all others. After Mar 25, 2011: $65 for All
Food: Continental Breakfast, Lunch and Snacks included
RU’s: 1.0 RU (Recertification Unit) for ASQ Recertification
Register: www.Regonline.com/ASQAuto2011

Contact: Jennifer Schneider, Symposium Chair, Phone: (248) 705-3216
E-Mail: Jennifer.Schneider@continental-corporation.com

New Product Development - Additional Protection From Unknowns

I thought I would lay out some of the things that I think are needed to have a chance for a clean launch of a new product.
I have started to discuss PFD and PFMEA and there is more to come on that. In addition to those two essentials, there are some other tools that can reduce the risk of an unknown problem from slipping through to the customer.
Remembering that the PFD defines the intended process and the PFMEA picks apart things that might go wrong, and why, we have the fact that humans are fallible and mistakes and omissions will be made. We will not be able to 100% determine all of the prevention and detection controls in advance. How do we protect ourselves from these "unknown"risks?

Here is a list.

High Tech Production Still Too Costly in US

An interesting article from the Metrowest Daily News about a tech start up company. Evergreen Solar had an innovative way to produce solar panels. The cost per watt was still higher than the market would bear, but with incentives and subsidies in 2008 they could make it:

Back then, photovoltaic panels - or "modules" - sold for an average of $4 per watt. Still spending $4.35 per watt to make the panels, Evergreen knew mass production was the key to lowering costs. The race was on: Make more. Sell more.

The production strategy worked. Production costs dropped to $2 per panel by the end of 2009. Shipments rose 84 percent to 104 megawatts. By then, the company was supposed to be turning healthy profits, spokesman Chris Lawson says.

But then, competition from China.

As Chinese photovoltaic companies moved into the market, panel sale prices have plummeted to record lows in two short years.
Chinese companies manufacture panels for $1.20 per watt and sell them for $1.60 to $1.70. Evergreen spends $1.90 per watt to make the panels.

Is the difference between the cost in China and the cost in the US due only to labor costs? In a high tech production system, how much of the cost per watt is due to labor?

Read more: The eclipse of Evergreen Solar - Framingham, MA - The MetroWest Daily News http://www.metrowestdailynews.com/archive/x945640045/The-eclipse-of-Evergreen-Solar#ixzz1FsRQlnjO

Process Flow Diagram

While discussing the PFMEA we are inevitably drawn to the topic of the process flow diagram. Where I work we and our suppliers are supposed to have a process flow diagram. The requirement is somewhat vague, I guess. And when I say "have a PFD..." I mean, we should have and UTILIZE an effective PFD. The PFD shown below would fit the letter of the law. In the illustrated case the manufacturer is just describing, for sales purposes, the process steps that their piezo actuator goes through. The instructions from the AIAG APQP Manual state:

The process flow chart is a schematic representation of the current or proposed process flow. It can be used to analyze sources of variations of machines, materials, methods, and manpower from the beginning to end of a manufacturing or assembly process. It is used to emphasize the impact of sources of variation on the process. The flow chart helps to analyze the totalprocess rather than individual steps in the process. The flow chart assists
the organization’s product quality planning team to focus on the process when conducting the PFMEA and designing the Control Plan.

There is a process flow chart checklist in the appendix. The checklist is focused on including steps in the process, but not process description

However, when a NPD team is working to develop a manufacturing process they need to define the process in detail, including inputs and outputs. A complete PFD is essential for the PFMEA discussion.


This PFD is shown at noliac.com

The Fight is Against Single Piece Errors

The series of posts I have recently begun on PFMEAs are rooted in my frustration at single piece defects getting to the customer factories. By this, I mean "one body missing a machined hole" or "one assembly missing a retaining pin".

Part of the manufacturing/quality struggle is to produce all parts within print specification. This struggle is addressed by dimensional controls and SPC. The PFMEA usually, almost exclusively, addresses dimensional issues. But the problema that I see passing through to customers are not that "our process shifted 10 microns over size", it is that that one machining operation was missed on one piece.

In the attached video the PFMEA is mentioned. I liked this video. The PFMEA comment was accurate.
"PFMEAs, although a great tool, are only as good as the knowledge put into them."



The video author admits that often failure modes and causes are missed in the PFMEA. What is the countermeasure to this failure mode in the NPD (new product development) process?


Thanks to TheUjigamiGuy on YouTube

PFMEA Training in China

Note: The video link is not always working, even directly to the site.

Here is a little video of a PFMEA training session in China. The banner reads,

"Failure Mode and Consequences Analysis"

The company is Konka Television, apparently pretty famous in China. When I work in China I see training like this quite often. It strikes me that I do not see this kind of thing in the US. Is it because it is assumed everyone in the US knows how to do a PFMEA? Don't get me wrong, I am not saying that this kind of training will be extremely effective. I am saying that the focus on "attempting" to improve the process is refreshing.

Where is the effective use of a PFMEA? Anywhere.

I am getting far enough in my career so that I dread starting work with a new product development team and the inevitable disappointment I will feel when I see how ineffective they are. The team is responsible for launching the new product without any glitches, defects or flaws. This is a tall order, right? Not easy. Yet these teams sleepwalk through advanced quality planning and walk into launch with their pants down.

Let's look at the use of PFMEAs. Can anyone reading this blog tell me of a time they worked in a cross=functional team that actually worked to develop the PFMEA as an honest attempt to list out potential failure modes and causes?

What I find is:
A) The PFMEA is a stale document handed down from project to project. Always, it never was that good in the first place and, even for new aspects of the new product there is no attempt to freshen it up.
B) The PFMEA was written by one person, not a team.
C) The PFMEA was written for the sole reason of checking the box: Is there a PFMEA? Yes or No
D) When confronted with the fact that the PFMEA is a stinker the author, or the project manager 1) denies it, and 2) challenges me to produce good examples (in other words, do it for them)
E) When obvious failure modes are brought to their attention the response is, "oh, that cannot happen because..." Without the PFMEA exercise holes in the prevention and detection process are not effectively reviewed.
F)Several times the new product development team will start to use some kind of PFMEA software to show they are now serious about doing a PFMEA. However, the software is difficult to use and somehow seems to distract from the goal.

Time and time again we see defects arriving at customer plants that are well known failure modes that should be expected if there are inadequate controls. Yet, they happen. Again and again.

The next article will be focused on some of the best practices I have seen.