Project selection is tight to organizational policy
You don’t make improvements all at once. You have to select
projects which are bound to the organizational policy. They are most of the
time focused on customer requirements and profit.
Any project needs to be tight to financial saving/profits,
whatever the project you do.
DMAIC is for improving existing processes.
Define – example – the latency should be reduced
Measure – current latency is measured as 3 minutes.
Analyse – analyse the problem
Improve – latency is reduced from 3 minutes to 1 minute
C – control increase of latency, and keep in 1 minute or
less
There is an evaluation after each phase of Six sigma
MSA = Measurement System Analysis – Helps you understand how
accurate and reliable your data is
Correlation and regression – if we change one factor, how
will it change other factors of an item
One philosophy of SS is reduce variance. The concept of Lean
focuses on reducing waste. This is the difference between SS and Lean.
Lean also focuses on make process faster. It also helps us
have a safe work environment, and to improve employee moral. Everyone will then
work towards improvements.
Lean:
·
Identify value – identify processes which add
value and which don’t. if no value is added then that should be a waste.
Reducing waste makes profit.
·
Map the value stream -
·
Create flow -
·
Pull – You product only what is required. In
push you make 100 products everyday. In push, only based on the customer
requirement, the products are made. So no waste is available.
·
Seek perfection – Strive for perfection by
continuous improvement
·
Functional – example – in a manufacturing
company, the manufacturing users will be the functional team
·
Cross-functional – example – Finance department
users
·
Virtual – example – teams from different
companies connecting online. There is no face-to-face connection
·
Self-managed – made of its own. People from
different areas come up with solutions to a common problem
·
Quality circles – people at working level solve
problems together
·
Forming – form the team
·
Storming – team members might fight with each
other/they might not believe in each other/lot of arguments and mistrust
·
Norming – you make rules to avoid arguments,
define and agree on norms
·
Performing – real important work is getting
done, actual tasks are performed
·
Adjourning – breaking up of the team once the
task is completed
Brainstorming – you don’t make judgements. You look at the
problem and find solutions. Each member of the team will have their own idea
regarding the problem. All those ideas are recorded. The quantity of the ideas
is important rather than quality. We are interested in more and more ideas. Get
as many ideas as possible. You have to avoid criticizing the ideas because
criticism will hold new ideas. Unusual ideas could provide great solutions.
After collecting ideas you can combine ideas to propose a better solution.
Nominal group technique: people who are dominative will keep
on introducing ideas. Some may not provide ideas since they are not good in
communicating them. To avoid such situation, nominal group technique is used.
Multi-voting: in brainstorming you considered about the
quantity. The ideas collected should be reduced by way of multi-voting. The
team has to decide filtering ideas, and not the group leader. List down all
ideas first and then ask from each team member 3 preferences out of the list of
ideas. First preference will be given 3 marks, second will be given 2 marks and
the third preference will be given 1 mark.
Any meeting needs to have an agenda. Publishing the agenda
before-hand would make it easy for the participants to utilize their time
effectively.
Each team member has a specific task to perform, before
attending the meeting. Many times, the project status report will be reviewed
in the meeting.
Meeting minutes are important because it defines who is
assigned what.
history of Quality tools :
Seven basic Quality tools:
1.
Check sheet – a data collection form used to
manually record an event. Example, each time a participant joins a meeting, it
is recorded. Check list is a reminder where you mark that you confirm the
availability of something. Check sheet and check list are two different things.
This tool helps you prioritize the problems since it gives an overall picture
of what areas have more issues.
2.
Cause-and effect diagram – Fish
bone/Ishikawa/Cause-and effect diagram are all same. The tool shows the causes
and effects of a problem. Number of causes and effects are drawn. The head of
the fish is the problem. Causes are displayed as bones.
3.
Histogram – a bar chart which displays the
distribution of individual measurement. Example - The trend of filling the
water bottles can be determined by measuring, recording and presenting the data
in a chart. The cause of the problem can be identified by way of analyzing
off-points of the diagram (by the shape of the diagram).
4.
Pareto chart – 80-20 principle. It tells that
80% of the problems are caused by 20% of the causes. The problems which happen
frequently should be focused. This helps you prioritizing the problems. The
vital few from trivial many are solved in this method. Pareto chart is also a
bar chart. X of a histogram is a number. Example – number of bottles. In Pareto
chart, X axis defines categories. The Pareto chart arranges the problems in
descending order. biggest bar is on the left. smaller ones are to the right.
5.
Scatter diagram – a plot between two variables.
One variable is dependent and the other is independent. An example of an ice
cream seller, X axis has the temperature, Y axis has the number of ice creams
sold. In the X axis we keep the independent variable which is the input
variable. Y axis has the dependent variable which is the output variable. With
the rise of the temperature if the number of ice creams sold will come down,
will be able to be identified here.
6.
Control charts – line charts depending on time.
Take the reading every day. We need control chart to identify when things go
wrong in the process. If things are running within the limit, then you are
good. If changes in the process are due to common causes, that is not
considered since placing too many controls would make your process complex. If
major deviations are there then causes will be identified and controls will be
placed.
7.
Stratification – breaking down the data into
categories so you can make sense of it.
Example of a check sheet
Ideas identified in the brainstorming are put in to the
cause-and effect diagram.
Six sigma matrices
In order to identify in a SS project whether you are in the
correct track, you can use Six sigma matrices.
·
Defects per unit (DPU): A nonconforming unit is a defective unit.
Defect is nonconformance on one of many possible quality characteristics of a
unit that causes customer dissatisfaction. One defective item can have multiple
defects.
·
DPMO (defects per million opportunities): defect
opportunities are circumstances in which CTQ can fail to meet. Example – a unit
/a bottle has 5 parts. Each part has 3 opportunities of defects. Hence the
bottle has 15 opportunities to go wrong. This is called the defect
opportunities. If we have 10 bottles, and if there were 2 defects then 2/(15 *
10) is the defect per opportunity. DPMO is DPO * 1 million = 2/(15 * 10) * 1
million = 13333
·
Rolled through yield (RTY) – yield (Y) = (units
entering a process – Defects ) / units entering a process = (P – D) / P
Let’s say you have 3 processes P1, P2 and
P3. In each process, a certain level of yield is there. RTY = Y1 * Y2 * Y3
where Y1, Y2 and Y3 are yields of P1, P2 and P3 respectively
·
Cycle time = when you want to make things fast,
you use the metric “Cycle time”. Example – you have three processes P1, P2 and
P3 which take different times to complete. Cycle time is T1 + T2 + T3. if one
process failed to product the number of items expected, the next processes
would also produce less.
·
Cost of quality – two categories – visible and
invisible. Visible means rejection, rework, repair cost and the cost of
inspection which are visible for inspection. Invisible means the lost sales,
excess inventory, additional controls and procedures, complaint investigation,
fines, legal fee etc. that are not available for inspection.
·
Prevention – cost spent for prevention of issues
·
Appraisal – cost spent for appraisals
·
Failure – cost spent for failures
Failures can be internal and external.
Prevention cost –
There is no common accounting standard to calculate and
measure the cost of quality. The management understands the language of money.
Management would not understand the details relating to quality. Management is
interested in identifying how to make money. You have to persuade to the
management that avoiding defects would increase the profit.
VOC is difficult to measure. CTQ can be easily measured.
External sources - Look at the market trends to get an idea
of where in this company you need to improve. If your competitors are doing
good than you, you need to look into their approaches also.
Internal sources – focuses more on the non-performing
processes and the employees. Also, the concerns from your employees would also
be helpful in identifying areas of improvements and better ideas to perform the
same.
List down all the internal and external sources and analyse.
This list will be a huge list, hence you need to filter the sources to identify
the most prioritized sources to be focused on. Follow the below basic
qualifications in order to filter the list.
·
There is a gap between current and the desired
performance
·
The cause of problem is not clearly understood
·
The solution is not pre-determined nor is the
optimal solution
Process consists of inputs, processing and outputs
In addition to that, supplier and customer are also
available in a process.
Communication plan
Work breakdown structure and Gant chart are important
planning concepts
WBS – used to split project work into multiple components.
Gantt chart:
Project management tools:
Tree diagram – one idea is branched into many ideas
continuously.
Quality of video can be subdivided as quality of the audio
and quality of the visuals.
SIPOC = Supplier, Inputs, Processes, Output, Customer
Measure phase
Descriptive statistics = you describe the statistics. You
describe on the mean value of the height, the height of the students.
Inferential = you take samples and based on the sample you
make a guess. Example – based on the sample, you can tell that who will win in
an election.
Measurement of central tendency – something which can be
taken as an average. Exexample – average of the heights of the students in a
class.
How much of variation you have in the heights / what is the
range/ what is the spectrum of the heights is the measurement of dispersion.
Mean = commonly known as the average.
Median = when information is arranged in ascending or
descending order, the middle value is called
as median
Mode = most occurring item in a set of data
The problem with the mean is that if you have an outliar
then the mean value conveys an inaccurate value. Any unusual value would
reflect an incorrect result.
Median = middle value
To take the median value, the set should be arranged in an
ascending or descending order.
Median avoids the issue in the mean value method.
However, if there are even numbers in a set, there is no
middle value. In that case you have to take the median of the two middle values
Mode =
Measurement of dispersion:
Why do we need this method? Example – you are offered a job
with 1100$ per year. That is the average salary. What might be the variation
here? 1100 $ could have extreme variations such as 1 + 1099. This is why you
need to understand the variations. If you were placed in a salary of 1$ which
still is included in the average salary of 1100$, there will be an issue.
Hence, understanding the variation is important.
Range is the highest and the lowest value and their
difference.
Calculating the standaard deviation :
Calculate the mean value (average). Calculate and take the average
of the difference between each number and the mean. Example : (M – N1 + M – N2
+ M – N3) / n. this value will always be zero. Hence, taking the mean in the
above way is not helping. Hence, take the square values to avoid the result
being zero. Standard deviation = √{(M –
N1)2 + (M
– N2)2 + …. + (M
– Nn)2}/n
Data collection plan – in the Measure phase, the purpose is
to collect data to understand how the project is performing now. You have to
have a plan to collect data. in your data collection plan, below factors need
to be considered:
Example:
Data could be qualitative or quantitative.
Another method of classifying data is continuous and
discrete, explained as below:
Example – 2 km is a continuous data since it can be
represented in multiple manners such as 1 Km 1000 m. Discrete data are data in
steps. Example, how many students are there in a class? If it is 40, 40 cannot
be 40.1.
Check sheets – you mark the availability in groups of 5
-> ||||
There are two types of values, true value and the reference
value. True value is unknown most of the time. Exact measurements are
impossible. Hence, the value which we measure is taken as the reference value.
A measurement system consists of the Operator, equipment,
procedure.
Due to changes in the operator, equipment or in the
procedure, the measurements could vary. 
Accuracy is something
relating to measurement of central tendency. Precision is relating to measurement
of dispersion (measurement of variation).
Bias – the difference between
the observed avwrage of measurements and the reference value.
Linearity – measures the bias
across the operating range of a tool or instrument. The bias changes over the
range of measurements.
The linearity represents the
increase or decrease in the Bias over the reference value and the average
value.
Stability – measures the bias
over time. Also known as drift. If your instrument is stable, the bias will not
change over time.
Repeatability -
Even though the instrument or
the operator is not changed, if the same variance is given then it is called
the repeatability. Repeatability is because of Peter (operator). Same operator
is taking the measurement. Any variance is because of the gage.
Reproduceability – is because
of different operators taking the measurements.
Reproduceability = Appraiser
variation (AV) – named because the variation is due to the operator.
Gage R&R = combined
estimate of repeatability and reproduceability = checks whether consistent
variations can be reproduced or not.
Lean tool – used to identify
and elimitate waste. 5S and value analysis are such tools.
Value analysis - Step by step
activities should be listed. Brainstorming should be done. Analysing should be
done and then should identify which activities are value addaded and which are
not.
Customer expects that you do
the work right in the first time.
Repairing – not value adding,
waste – not value adding.
Assembling – value-adding,
serving cofee – value-adding, overproduction – non value-adding. Because you
can’t give the produced quantity free unless the customer asks for more goods.
Till that time, money is tied up with these overproduced items.
Overprocessing –
value-adding. Doing things above and beyond is overprocessing.
FMEA – Failure Modes and
Effects Analysis
Example – what all ways your
car can fail? Less pressure in the tire, problem with the engine etc. list down
all What could go wrongs. This is a pro-active tool. Failure modes are the What
could go wrongs. Effects are the implications that can arise due to the What
could go wrongs.
FMEA is a proactive tool
which is used before the problem happens, and not after the problem happens.
Example – You have a perfume
making company. Receiving the perfume materials is one process in which wrong
ingredients is one failure mode which will result in poor quality of the
product. Severity can also be given from number 1 to 10. Occurrence = chance of
the problem happening again. On a sclae of 1 to 10, a value is given to the
occurrence. Current controls are listed. RPN = Risk Priority Number which
indicates which item you want to be attended immediately. RPN = Severity *
Occurrence * Detection
RCA (Root Cause Analysis): in
the analysis phase of the project, we talked about tools such as Lean. RCA is
one such tool. It is actually a set of tools or an approach.
RCA is a structurerd process
to identify root causes of an event that resulted in an undesired outcome and
develop corrective actions.
5 whys method: ask why
multiple times to reach to the root cause of the problem. Example -
5 times you have to ask – 5
is not restrictive, you can keep on asking why till you reach a solution which
is within your control.
Force field analysis might
not help you identify the root cause. It looks at two things, what are favor of
the proposed change and what are the forces that are resisting the proposed
change.
In the Matrix chart, vertical
list shows the things expected by the customer. The chart helps you select the
best option.
Quality function deployment =
QFD. This will be discussed in the Green belt.
Data types – continuous and
discrete
Discrete data > Binomial distribution. Example –
flipping a coin. Head or tail.
Continuous data > Normal
distribution
The climax in the above graph
is the mean value, and the curves in the left and right display the variations.
Common cause – example – you
are late by 5 minutes everyday. Special cause – example – you are late by 30
minutes since you got late to wake up. Common causes cannot be /difficult to be
eliminated.
You need to focus only on the
special causes.
Correlation – the effect of
one variable on another variable.
Y = f(X) in this Y is the
output and it depends on X.
Score does not directly
depend on the number of study hours. There are other factors. However, the
number of hours in one fact so we can identify the correlation between study
hours and score.
In th above, the relationship
between the number of hours and score has a strong relationship. Because, the
scatter plot can be represented in a line, and the scatters are close to the
line.
The equation above represents
how strong the relationship between number of hours and score is.
Roughly, Y = X + 16 in the
above case.
How to derive the formulae
will be discussed in the Green belt.
If you are expecting a voulme
of 150cc in a perfume bottle, and if the outcome is 149cc then would you change
the ingredients?
Null hypothesis is that mean
is equal to 150 cc (which is the existing situation). That is the fact as of
now. Alternative hypothesis is something which you want to prove.
Example – my wife has a doubt
whether the roof is leaking or not. Until we find out that the roof is leaking,
our null hypothesis is false. if we found that the roof is leaking, the
alternative is true.
In the above example, you
either accept or reject the null hypothesis by way of identifying whether a
leak is there or not.
The result of the hypothesis
testing is finding out the probability. (P)
If the probability is less,
the null hypothesis is rejected. Else, the null hypothesis is accepted.
When you reject the null
hypothesis based on the result of an odd sample, you call them errors.
Type 1 error – innocent is declared
as guilty. Example – a false fire alarm leads to unnecessary inconvenience.
Type 2 error - Guilt is declared
as innocent
Kaizen is an improvement
technique. Kai + Zen = Change + Good. It means that change for good.
Kaizen even is you look at a
process, documenting, identifying waste, making change, document the new process
..
Another improvement technique
is Kaizen Blitz.
A Kaizen Blitz, or rapid
improvement is focused activity on a particulaar process or activity.
PDCA Cycle is also another
improvement technique. Plan + DO + Check + Act
PDCA is an iterative process.
Hence it is called as a cycle.
Cost benefit analysis should
be done when you are improving a process
Value = Benefit – Cost
Control plan – this is a
controlling technique
Control plan ensures whatever
the changes you made remain.
You keep on recording the
data to identify trends in the control chart.
In addition, the control
chart imposes upper and lower control limits, by which you can identify the
special causes which reflected data that went beyond the upper control limit
and data thaat went below the lower control limit. At what time you need to
ignore the fluctuation, and at what time you need to analyse and control
fluctuation can be seen in the control chart.
Document control – in DMAIC,
documenting is not specifically mentioned. However, the changes that you make
should be documented, which benefits the control phase and analysis phase.
