Wednesday, November 23, 2011


Gemba : In Japanese refer to the place where an event happen. The term can be easily understood if translated into Chinese or Kanji.

Gemba : 现场

I always believe Gemba visit is an effective way for root cause finding. To understand a problem, engineers must jump into the site when doing root cause finding for the problem, by then only he can fully understand the impact of the problem to others and propose precise corrective action.

I like to carry out Gemba walk when I have project activities executing in the production floor. I do believe this is the only way to get first hand information and react accordingly when something went wrong.

Gemba walk is easy, you no need any special tool. What you need is your initiative to walk into the site to have a personal feel of a working condition that other person is facing. By then, only you know what actually went wrong, how serious the problem and the impact to others.

Sound simple isn’t it ? but in reality, how many of us really like it or practicing ?

Tuesday, November 1, 2011

Why ???

Why do it this way...
If you have a chance to simplify your process in this way ?

A Simple Change

This is the outcome of the Gemba walk I have in one of the production floor. Red color line indicates the material flow from one process station to another. Don’t be surprise, peoples working in the production floor has been living in such production floor plan for several years and they don’t see any problem with it.

They can make their process flow leaner if they willing to change. I do send out my proposal as below. Hopefully they will accept.

Wednesday, October 19, 2011

You Are The Real Boss

A colleague silently expresses his disappointment to me in the last Friday meeting. From the short conversation, I can sense that he is quite unhappy with his superior and the management. He has been with the department for more than 6 years but he is not given a change for his career advancement.

I feel sorry to hear that. From what I know, he is a very experienced engineer who gains lot of respect from his technical achievement.

He told me he is considering of inter departmental transfer but he worry he cannot get the green light from his “BOSS” because there are still plenty of cases need his attention…..

I don’t want to elaborate the case further since the answer is straight forward….

I have seen many similar cases like this, most of the time peoples suffer in such situation prefer to keep silent even they know they were ignored by the management. No one dare to walk into their manager’s office to voice up their feeling and seek for opportunity to fix the problem.

Honestly, I do face the same problem; I really walk into the GM’s office and ask for discussion about my career path, but I was chase out by him. I am quite disappointed when I asked to get out from the office, it is my bad experience. Anyway, I learn something from the lesson.

What for working with the management who don’t appreciate your talent?

Since then, I never treat my employer, managers as my “BOSS”. They are not my boss, they are just one of my customer. I M THE REAL BOSS.

I treat myself a service provider selling a service to the company at the price I think fair for both. I am responsible for the work I deliver. I have the right to ask for the fair price and I have the right to choose my customer who offer me best return and opportunity to growth.

Try to think this way, you may see thing differently….

Originally post on 2007

Thursday, October 13, 2011

The Most Difficult Item In Any Change Project

If you ask me, what is my biggest challenge in any project for continuous improvement?

I always gives a standard answer, i.e. : To change the people’s behavior.

Per my past experience, I always start an improvement project with lengthy discussion with a group of manager. I have to keep on repeat the same story again and again, for month or even for years, in order to convince them to approve my proposal to make a change.

When come to the change execution, actually, this is not so difficult to me, since I m only dealing with machine most of the time. I can simply make a change, do some data collection. And usually it only takes a few weeks for me to complete the task.

Now, come to the full implementation stage, I need to ensure peoples in the shop floor following the new instruction, make use of the new facility given to them and not converting the process system to the old way that they used to it. Dealing peoples in the shop floor always a big challenge to me. People are just simply refused to change; even they knew that they will be benefited in the new system. In my past experience, I ever watch a production operator to perform his duty for more than 6 month, in order to tune him to follow the new work procedure.

This is the reason why many improvement activities takes years to complete, I would say, 50% of the time wasted dealing with peoples on non value added topic.

Saturday, October 1, 2011

A Big Project I Ever Handle

Having a GB project review together with a group of Deployment Champion. To a company that new to Six Sigma, in general, I believe they are impressed with the result presented by the GBs, I m glad to see all the GBs presented their project successfully and they will be reward with an official Green Belt certification in a graduation ceremony.

This is the 1st batch of Green Belt being trained and certified with the classical Six Sigma certification approach. i.e. the candidate have to go through 8 days of class room training to cover all DMAIC tools, work on a project and must be able to demonstrate the capability of using the tool.

Since this is the first time the company adapt the Six Sigma program, and my 1st experience being the key driver of the deployment, some imperfection is expected, but this is not a blocking for continuous improvement.

In fact, we started the discussion about the deployment in July 2010, and we decided to start off a small scale program with the support of three internal staff that has the background of Six Sigma DMAIC methodology. We started the project with Six Sig Deployment strategy set, followed by organization structure, training material and finally training program.

This is not an easy journey to me in the past one year, but it is definitely a great success to me. You know, I m working with a company that do not believe in Six Sigma program once upon a time. People only start to talk about the program after four years since I join them as Project Leader. Anyway, they willing to open up an opportunity to try something new to them, the move is really encouraging.

Well, this is the time I need to prepare for the Wave II training. The demand for Green Belt training is overwhelming, again, this is another exciting challenge waiting for me, I never had the experience to handle such a big group of trainees, I have to think away to overcome the problem.

Let’s see what will happen to the next Wave of Green Belt training.

Sunday, September 18, 2011

Alpha Risk & Beta Risk

Alpha Risk : Also name as Type I Error, is the risk of rejecting the Null Hypothesis when in fact it is true.

Example of Type I Error;
Production operator is making a wrong judgement to reject a good part. Therefore producer is facing some level of risk to loss due to the Type I error commit by the production operator, because of this, Type I Error ( Alpha Risk ) also name as Producer Risk.

Better Risk : Also name as Type II Error, is the risk of accepting the Null Hypothesis when in fact it is false.

Example of Type II Error;
Production operator is making a wrong judgement to accept a rejected part. Due to rejected parts was accepted and deliver to the market, consumer is facing some level of risk to accept the rejected part. Therefore, Type II error also name as Consumer Risk.

Alpha Risk and Beta Risk is used in Power & Sample Size estimation

Saturday, September 3, 2011

Kanban Card In Inventory Control

Days after the presentation, I try to meet up with VJ to understand his problem and the requirement of inventory control. He tell me he need to keep 5 rolls tape in the cabinet, and the supervisor is required to trigger stock replenishment when the found 2 rolls left.

I explain to him about the Kanban Card control and his problem can be easily solve with Kanban Card. Here the solution I propose to him

1. He has to prepare 6 Kanban Card as shown

2. He has to prepare a display board to store the Kanban Card every time operator withdraw the material from the cabinet.

With the 2 system in placed, now the supervisor’s job become very easy. What they need to do is just look at the display board and trigger stock replenishment on those envelope that fill up with Kanban Card.

I think this is a very simple visual control system that can be apply without incur much cost. The production control executive can just manage the inventory by controlling the Kanban Card qty. Supervisor can just perform their daily function by looking at the Display Board.

Is this a better and practical idea ?

Sunday, August 28, 2011

A Common Problem In Inventory Management

A production control executive presented the method he setup to manage the inventory of a mounting tape. Actually, the method is simple and conventional, he prepare a stock card to record all the event of material issuing, the production supervisor is requested to check the stock card every shitf to ensure material replenishment order being triggered when the stock level is reaching minimum stock level.

There are few problem i encountered in such system;
1. The operators drawing the material is required to note down the event on the stock card. This is an additional task for the operators and it is a manual recording, there is potential for human error.

2. Production supervisor is required to verify the physical stock verses the stock card and trigger stock replenishment when ever necessary. Is this an additional task for the supervisors ?

3. The production control executive also highlighted the issue of double ordering due to communication breakdown of supervisor between shift.

These are the few problems the production control executive is facing, but i believe similar problem happen in others company, don't you think so ?

Ok, i will post my solution in the next post.

Friday, August 26, 2011

Residual Plot

Residual (e) is the delta of observed value of dependent variable (y) and the estimated value (Y).

e = Y - y

Theoretically, sum of ( e ) should be close to 0 if the resudual of a regression model is normally distributed.

An appropriate regression model should have a random and normally distributed residual. We can check this condition with Residual Plot.

Example of Residual Plot

Thursday, June 9, 2011

Minitab - Cross Tabulation and Chi-Square

GP is sending a set of discrete data and ask for help to test for significant differences. Looking at the data, I know Chi Square or Contingency Table should be the right analysis tool to test the data. But… OMG, I can’t remember how to run it in Minitab.
Honestly, it takes me few days to do the treasure hunt in the Google. Finally, I got the answer.
Hmmm… I think I should document it for my future reference.

Here the Minitab step by step that I copy down ( Minitab Ref ID 210 )

Step 1 : Tabulate the data as below
                                Pass       Fail
B.Improve                3788       52
A.Improve                3835       5

Step 2 : Stack the row ( Data > Stack > Row )
Store stacked data in, C6.
Store row subscripts in,  C4.
Store column subscripts in, C5.
Click OK.

Step 3 : Stat > Tables > Cross Tabulation and Chi-Square
For rows, enter C4.
For columns, enter C5.
In Frequencies are in, enter C6.

Step 4 : Display. Check all the 4
Row percents
Column percents
Total percents

Step 5 : Other Stats.
Check Fisher's exact test for 2x2 tables. Click OK in each dialog box.

Step 6 : Chi-Square
Check Chi-Square analysis

Finally, this is the answer I get.

Tabulated statistics: C4, C5

Using frequencies in C6

Rows: C4   Columns: C5

         Fail    Pass      All

1          52    3788     3840
         1.35   98.65   100.00
        91.23   49.69    50.00
        0.677  49.323   50.000

2           5    3835     3840
         0.13   99.87   100.00
         8.77   50.31    50.00
        0.065  49.935   50.000

All        57    7623     7680
         0.74   99.26   100.00
       100.00  100.00   100.00
        0.742  99.258  100.000

Cell Contents:      Count
                    % of Row
                    % of Column
                    % of Total

Pearson Chi-Square = 39.044, DF = 1, P-Value = 0.000
Likelihood Ratio Chi-Square = 45.424, DF = 1, P-Value = 0.000

Fisher's exact test: P-Value =  0.0000000

Sunday, March 20, 2011

Debate About JMP Minitab or Cornerstone

"Cornerstone is not user friendly; we need better data analysis software like JMP or Minitab for our data analysis."

Somebody brought up the topic in the last management review, and we spend sometime to debate the disadvantage of existing data analysis software and how it make our life inefficient.

I try not to involve in the discussion, because I m new to the software and basically not qualify to comment.

I use Minitab most of the time for data analysis, I got no experience with JMP and recently I try to play around with Cornerstone due to the need for Green Belt training. Honestly, I do face some problem at the beginning, mainly due to the arrangement of pull menu (the way Cornerstone group the data analysis function) and the reporting format. Technically, I don’t face much problem after hours of trial.

I didn’t spend much time to study the full function of Cornerstone, but at least, I get what I want. This is most important.

Personally, type of data analysis software to use is not the top priority, what is most important is the knowledge of data analysis, knowledge of digesting information from a given data set. Without the basic understanding, even given with the best statistical software, so what….

Data analysis software is just a big calculator for you to perform more complicated mathematic calculation other than the common Microsoft Excel. Whether it is Cornerstone, Minitab or JMP, bottom line, it must return the same result.

Saturday, March 5, 2011

Kanban Size Estimation

Is there a formula for Kanban size calculation?
Answer is : “Yes”. But the formula I ever learnt, I feel most of them are very academic and not really practical to the real production environment.

In my previous experience in box build assembly line design, I always consider a Kanban size = 3 into my production system. You may ask why 3, not other lucky number such as 8.

The secret behind is very simple, because my boss say : Kanban size = 3
He demand for 3 is based on his many experience in this field, 3 is the number where he think reasonable and achievable.

Anyway, you can’t just copy the number blindly and apply it to your production system. I’m sure it will not work. You have to go back to your production line, understand how the people run the line then only you can tailor a Kanban size the suite to your production system.

Hahha… talk so much, sound like all nonsense isn’t it.

Ok, lets discuss the case below.

Any Kanban size determination require basic process info as below
  • Takt Time and Raw Process Time (RPT) of your downstream process
  • RPT and Setup Time of your upstream process

There are two WIP Level that need to take care in a Kanban
  • WIP Level in Normal production run
  • WIP Level for “Pull” signal triggering

WIP Level for “Pull” signal triggering is the minimum WIP that enough to keep the downstream process running while waiting for 1st Lot replenishment from the Upstream process. WIP Level above this would be the WIP Level in Normal production run, where we can decide the level based on several considerations, such as

Optimum production time of Upstream process for every conversion
Production storage capacity
Resource availability to handle too many conversion frequencies

For this case, I would determine the “Pull” triggering point as below

Lot # = [(Upstream RPT + Upstream Setup Time) / Downstream RPT] + 1
So, answer is ~3 + 1 = 4

This mean, when the WIP level drop to 4 Lot, a signal will send to Upstream process for Kanban replenishment. But due to long set up time, Upstream process require 130min set up + 40 min production time to deliver the 1Lot to the Kanban tray. Per the RPT given, Downstream process require minimum 3Lot of material in order to keep the line running. I add additional lot in this calculation, because with the consideration Upstream process is not necessary handle one product, as conversion take place. Therefore, additional waiting time may be needed for last Lot clearance before machine conversion work can be start.

For the next level of WIP that supporting normal production run, you need to consider the most economy way to run the process on each conversion. There are company actually set the policy to limit conversion frequency ( such as 2 per day, 4x per day ) with the consideration of conversion cost, productivity and resource availability.

Just my Six Sense

Wednesday, March 2, 2011

Common Question In Six Sigma Training

In a Green Belt training, there are participant ask such question;

  1. This is a Lean topic; can it be a Six Sigma project?
  2. Can I do a Six Sigma project without using statistical tool?
To answer the first question, per my opinion, in fact, there is no such thing of Six Sigma project or non Six Sigma project. Six Sigma; by itself is just a systematic problem solving methodology that following the DMAIC flow. As long as you are able to define your problem statement, you are able to define a metric to measure your performance, high chance; you are able to solve your problem through the DMAIC flow.

For the next question, can we run Six Sigma project without using statistical tool?

Before giving a “Yes” or “No” answer, I think we have to make ourselves clear the statistical role in the process of problem solving. In any case, we use number for performance measure, to tell the difference of before and after a chance being introduced. Besides using number, I don’t see any method that can be precisely describe a chance.

Now, let me answer the question.

Answer to you is “Yes”, you can avoid using statistic in your Six Sigma project, if you are able to describe a change precisely or effectively to your boss and your boss accept your description, then, I don’t see any problem to apply your method in your Six Sigma project. This is the basic requirement you have fulfill in your Six Sigma project.

There are many tools in the Six Sigma program; there is no one solution that fit all. You have to apply your expertise, your experience and yours intelligent to pick up the right tool to solve problem throughout the DMAIC phase.

Hope I answer your question;

All the best to you

Saturday, February 26, 2011

Run Chart

Run Chart is a basic line chart that may be use to monitor a process stability over time

Characteristic of Run Chart
1. Horizontal line plot across the data point are Median of the data set
2. It do not consist of UCL and LCL as shown in other Control Chart.

How to detect a Run
1. A "Run" could be a Single or Multiple data point above or below the Median
2. When a data point move from one side to another side of Median, a "Run" start and the "Run" ended when it cross the Median line again.

Type of Run could be observe from a Run Chart
1. Number of Run about Median
2. Number of Run Up or Down

Minitab software report Observed and Estimated Run in the Run Chart
Minitab test for two non-random behavior detect trends, oscillation, mixtures and clustering of the data. Such patterns suggest that the variation observed is due to "special cause"

Test for Randomness
Number of runs about the Median
More run observed than expected

Mixed data from two populations
Fewer run observed than expected

Clustering of data
Number of runs up or down
More runs observed than expected

Oscillation – data varies up and down rapidly
Fewer runs observed than expected

Trending of data

Minitab test the approximated p-value for Mixture, Clustering, Oscillation and Trend
Mixture Pattern - a mixture characteristic observed when there is absence of data point near to the Median line. Mixture often happen when combination of data from different population, or data set represent different process condition

Cluster pattern - Clusters may indicate variation due to special cause, such as, incident of broken tool, significant process parameter change.

Trend pattern - A trend is a sustained drift in data, either moving upward or downward. It may be an indication of process about toward out of control.

Oscillation pattern -  occurs when the data fluctuates continuously up and down rapidly, it indicate the process instability.

Data analysis ( based on above Run Chart )
Run across Median
- Fewer Run observed than expected
- approximated p-value for mixture = 0.96. ( no significant evident to proof the indication of data from different population )
- approximated p-value for clustering = 0.03 ( indicate strong evident of data clustering )

Run up or down
- Fewer run up or down than expected
- approximated p-value for trends = 0.13
- approximated p-value for oscillation = 0.86
- Data indicated significant of trends and oscillation, however, oscillation is more significant than trends in this case.

Reference - Minitab Release 14