Six Sigma DMAIC is a structured engineering method for improving processes by reducing defects, variation, and waste. This cheat sheet helps students remember what to do in each phase: Define, Measure, Analyze, Improve, and Control. It is useful for design projects, manufacturing studies, quality labs, and engineering problem solving.
Students need it because DMAIC connects data, customer needs, and practical improvement decisions.
The most important ideas are defining a clear problem, measuring the current process, finding root causes, testing improvements, and keeping the process stable. Key tools include CTQ requirements, SIPOC diagrams, process maps, Pareto charts, fishbone diagrams, 5 Whys, control charts, and capability metrics. Important formulas include defect rate = defects / opportunities, DPMO = defects / opportunities x 1,000,000, yield = good units / total units, and Cp = (USL - LSL) / (6 x sigma).
A strong DMAIC project uses evidence instead of guesses and verifies that the solution works over time.
Key Facts
- DMAIC stands for Define, Measure, Analyze, Improve, and Control, and each phase should produce evidence before moving to the next phase.
- A good problem statement includes what is wrong, where it happens, how often it happens, and why it matters to the customer or system.
- A CTQ requirement translates the voice of the customer into a measurable target, such as maximum cycle time = 30 seconds.
- Defect rate = number of defects / number of opportunities, and it measures how often a process fails to meet a requirement.
- DPMO = defects / opportunities x 1,000,000, and it allows comparison between processes with different numbers of units or checks.
- Yield = good units / total units x 100%, and it shows the percent of output that meets requirements.
- Cp = (USL - LSL) / (6 x sigma), and a larger Cp means the process spread is smaller compared with the specification width.
- A control plan lists the key variable, target, measurement method, sampling frequency, reaction plan, and owner for maintaining the improvement.
Vocabulary
- DMAIC
- A five phase Six Sigma improvement method used to define a problem, measure performance, analyze causes, improve the process, and control results.
- CTQ
- Critical to Quality is a measurable feature that strongly affects whether the customer or user considers the output acceptable.
- SIPOC
- A high level process map that identifies Suppliers, Inputs, Process steps, Outputs, and Customers.
- DPMO
- Defects per million opportunities is a normalized defect measure calculated as defects / opportunities x 1,000,000.
- Root Cause
- The underlying reason a problem occurs, not just the visible symptom of the problem.
- Control Chart
- A time ordered graph with a center line and control limits used to judge whether process variation is stable or unusual.
Common Mistakes to Avoid
- Skipping the Define phase is wrong because an unclear problem statement can lead to measuring the wrong data or solving the wrong issue.
- Confusing defects with defective units is wrong because one unit can have more than one defect, so the defect count and defective unit count may differ.
- Using averages alone is wrong because Six Sigma focuses on variation as well as center, and two processes can have the same mean but very different spreads.
- Assuming correlation proves the root cause is wrong because two variables can move together without one directly causing the other.
- Stopping after an improvement test is wrong because the Control phase is needed to monitor the process and prevent the old problem from returning.
Practice Questions
- 1 A process has 18 defects across 600 units, and each unit has 4 defect opportunities. Calculate the DPMO.
- 2 A production line makes 950 good units out of 1,000 total units. Calculate the yield as a percent.
- 3 A part has USL = 10.20 mm, LSL = 9.80 mm, and sigma = 0.05 mm. Calculate Cp.
- 4 A team wants to change a machine setting immediately after hearing one operator’s opinion. Explain which DMAIC phase or tool should come first and why evidence matters.
Understanding Six Sigma DMAIC Reference
A DMAIC project works best when the team sets clear boundaries before collecting data. The process needs a start point and an end point. For example, a school cafeteria study might begin when an order is placed and end when the meal is received.
Words such as late, damaged, or accurate need operational definitions. A late meal might mean one received more than five minutes after a stated target. This prevents different people from recording the same event in different ways.
CTQ measures should be tied to a real user need, not just a number that is easy to collect. A customer may care about a device arriving intact, while an engineer measures package drop damage or cracked parts.
Measurement quality can limit the whole project. Data is not automatically trustworthy because it came from a sensor, spreadsheet, or inspection form. Students should check whether a measuring tool gives similar results when the same item is measured repeatedly.
They should check whether two inspectors make similar decisions. This is called measurement system analysis. Sampling matters too.
Data taken only during a quiet morning may not represent a process that runs all day. Recording time, shift, machine, material batch, and operator can reveal patterns hidden inside an overall average. A useful baseline includes both the typical result and the amount of variation around it.
Analysis is about separating likely causes from interesting guesses. A Pareto chart helps focus attention because a few defect categories often account for much of the loss. A fishbone diagram creates possible causes, but it does not prove them.
The team must test each important cause with observations or data. The Five Whys method is useful only when every answer is supported by evidence. Students should watch for the difference between correlation and causation.
Two events can occur together because of a third factor. For instance, defects may rise on a late shift because a certain raw material batch is used then. Control charts add another layer of evidence.
Random movement within expected limits usually comes from common causes built into the process. A point far outside the limits or a nonrandom pattern can signal a special cause worth investigating.
Improvements should be tested on a small scale before they become standard practice. A pilot can show whether a change reduces defects without creating a new problem such as longer cycle time, higher cost, or unsafe work. Changing several factors at once makes results hard to interpret.
When possible, change one factor deliberately and compare the results with the baseline. Capability measures describe whether the natural spread of results fits inside customer specifications. Cp considers process spread, while Cpk also reflects whether the process is centered between the limits.
A process can have low variation yet still fail if its average is too close to one limit. Control means building the successful method into daily work through clear instructions, regular checks, assigned responsibility, and a response when results begin to drift.