How to solve problems effectively?

Problem solving can be challenging if the root cause is unknown and the problem needs to be solved right away. Everyone involved with the problem looks at it in different ways.

Listed below are some key perspectives as viewed by different people.
- The Problem Solver's view: Get to the root cause as quickly as possible and implement corrective actions to fix the problem.
- Management's view: The business is losing money or reputation because of the problem. Why is it taking so long to fix it?
- Process owner's view: We don't want to supply bad or suspect product or service to the customer, so we continue to be extra careful. Additional inspection levels, 100% inspection, quality audits are some of the ways to prevent poor quality product or service leaving the facility.

Because of the difference of opinion about the problem, most problems don't get fixed permanently. Usually the first symptom cause is considered the root cause and quickly fixed. That is why the problem re-appears after a while.

So how do we solve problems effectively?
1. Use structured problem solving method like the DMAIC approach
2. Solve problems that make business sense i.e. that will benefit the business
3. Use data to prove root cause and validate solutions!

Lean Six Sigma uses the DMAIC approach and tools for process/data analysis. DMAIC or Define-Measure-Analyze-Improve-Control are the phases of a Six Sigma project. Statistical tools at each phase of the project allow problem solvers to use process data to prove root causes and validate the solutions will fix the problem.

LSS projects focus on problems that make a business case and bring bottom line improvements.
The figure below shows key elements of an LSS project.

The clearly defined structure of LSS allows practitioners to solve problems effectively.

p, % and ppm for discrete data

Discrete or attribute data deals with counts. There are two types of attributes - defectives and defects.
Defectives are number of parts that are non conforming to a standard. Defective counts are parts that fail to meet a specific criteria set by the customer.
Defects on the other hand are non conformities or problems/issues within parts. A single part can have multiple defects i.e. the defect count can be greater than the total number of parts inspected.

It is important to understand the fundamental difference between defects and defectives because the statistics, tools are different for each. Here we will look at the stats/charts for defective data.

Defective data can be presented in several ways. There are 3 ways to calculate and chart defective data.
1. p or proportion
2. % or percent
3. ppm or parts per million

Most continuous improvement practitioners get confused between the 3 ways and this causes misinterpretation of the information presented. To clearly understand the difference, we will use sample data and perform the calculations.
For a set of 250 samples, quality inspection found 12 bad parts i.e. 12 defectives.

Cp, Cpk vs Pp, Ppk

In the last post we looked at the different process capability indices like Cp, Cpk, Cpm etc. But there is also Pp and Ppk indices that measure process performance. There is a fundamental difference between the two types of indices.
We will now try to understand the difference between process capability and process performance.
And when to use what capability index?

Process capability measures the ability of the process to consistently produce output within specification limits. Process capability indices are calculated assuming the process is statistically stable i.e. has only common cause variation.

Process Performance is an estimate of the process capability during initial setup. The indices are calculated assuming the process is unstable. Such a process has both common and special cause variation. A comparison between process capability and process performance helps understand the difference between the two.
 

 
Have you used Cp, Cpk or Pp, Ppk for measuring your process capability? We would love to hear from you at info@sybeq.com.

Histogram & capability analysis

In the last post we looked at tools for analyzing distribution of variable data. Histogram is the most popular distribution analysis tool. Using the histogram we can also predict the capability of the process.

Histogram shows the frequency distribution of the data. The curve shows shape of the distribution. Skew measures asymmetry of the distribution. A positive skew indicates a histogram with a longer right side tail whereas a negative skew has a longer left side tail.

So where is the tie in between the histogram and capability analysis?
Process capability analysis compares process performance against specifications. Histogram shows the distribution of process data. If the histogram graphically showed the specification limits then it is showing if the process is performing within specs or not. So visually the histogram would show if the process is producing good parts or not.

Numerically capability of a process can be analyzed by calculating the capability indices. There are several capability indices each showing an aspect of process capability. Most popular capability indices are Cp, Cpk, Pp, Ppk.

Read our White Paper on different capability indices and their uses.

What capability indices do you to determine process capability? How do you use the histogram for distribution analysis?

Distribution Analysis for variables

Distribution analysis for variable data is one of the critical analysis that help us understand how much
variation exists in the data. This in turn helps us understand how wide the process is and whether it falls within the specification limits.

There are several tools for studying distribution of variable data. The tools listed below are some of the popular ones:
1. Histogram
2. Stem and Leaf plot
3. Dot plot
4. Box plot

Histogram being the most popular to understand the nature of distribution of variable data. Dot and Stem & Leaf plot are more detailed versions for analyzing data distribution.
Box plots (also known as Box and Whisker plot) are used to compare distributions. Example for using a box plot would be to compare machine performance.

The figure below shows a comparison of the 4 tools.

We would love to hear which of the above tools have you used for analyzing distribution of variable data. Write to us at info@sybeq.com

Analytical tools for variable, attribute data

We have looked in the data analysis piece of Lean Six Sigma projects.
By now we know that data analysis is key in a Lean Six Sigma project. At the same time it is equally important to avoid analysis paralysis. A Green or Black Belt must know that there are different analytical tools for the type of data.

There are 2 data types
1. Variable or Continuous : Data that is measured and readings can have decimals. Eg. Weight, length, diameter. 3.56lbs, 4.5mm
2. Attribute or Discrete: Data that is counted and readings are integers. Eg. number of rejects, number or cracks

In attributes, there are 2 categories specific to the type of count.
Attribute types:
1. Defectives - The number of parts that are rejected are defectives. Also called non conforming units.
2. Defects - Number of specific problems with parts eg. burred, cracked, scratched parts. Defects could be fixed by repairing them. These parts are called non conformities.

For every data type there is a set of analytical tools. The figure below shows the commonly used tools by the type of data.

Types of analysis in LSS projects

LSS projects focus on data driven process improvements. Lean Six Sigma is all about data. Every phase of the DMAIC approach in LSS needs to be backed with data. Most of the data is collected in the Measure phase. This data gets analyzed, verified and validated using LSS tools.

Along with data analysis it is important to know about the process. So process analysis plays an equally important role in LSS projects. Knowledge about the current state of the process helps understand what the process steps of the current process. Here it is necessary to use the Go See Lean principle where the LSS team walks the process as it is currently running. Some Green and Black belts make the mistake on relying work instructions or hear say for the current state. However this may not represent what is actually being done on regular basis.

The figure below shows the types of analysis and some useful tools:


Analysis of critical parameters is the key to a successful Lean Six Sigma project. The challenge lies in identifying the critical parameters, gathering data and choosing the right analytical tools. Over and above knowledge about the process plays an important role too.

Have you experienced challenges related to analysis in your LSS project?