If you remember in the previous post we gave a brief introduction to the DMAIC cycle as a tool to improve processes. D. Define, M. Measure, A. Analyse, I. Improve and C. Control.
In this post we will talk about the “MEASURE” stage.
“The rigorous and objective measurement of processes is the starting point for any analysis that we want to perform later on: without measurement, there is no improvement.”
This is the stage in which we have to “take the temperature” of the process; the lack of data, and above all data concerning the results of a process, prevents its control.
The steps in this stage are:
- Define the variables to be measured (there should be no ambiguity)
- Define and validate the system of measurement
- Establish the limits of the values of the variables
- Calculate the process capability
Since many variables may exist, we must choose those that are most representative of each process, that provide most information, i.e. those that indicate whether the process is good or bad. The variables to be measured depend on the nature of the process. (Dimensions, Time, Quality, Money, Compliance etc.)
The method or system of measurement is very important, and sometimes to get objective and useful information involves excessive difficulty. It is very important to define how each variable is obtained and analyse where the data comes from and how it is calculated or what equipment is being used to obtain the measurement (dimensional), and the method of measurement being employed.
When we use measuring equipment it is important to carry out studies of repeatability and reproducibility (R&R Tool) to observe variations in the measuring equipment and the method of measurement used by the operator. (Many of the problems may be hidden here.)
Once the variables to be measured have been defined and the methods of measurement validated, we will establish the limits to know when the process produces good and bad data. This definition of limits should be carried out based on the specifications (e.g. tolerances in dimensional variables, items with faults in terms of quality variables, response time in time variables etc.). The aim is to define the maximum allowable limit beyond which each variable stops being considered good and starts being considered bad. In this way the process can be said to be under control when the variables are within the allowable limit, and action can be taken on the process when the variables exceed the limit.
Now it is just a question of starting the process and measuring, collecting the data for each variable during the process and establishing a system to define whether it is under control or not, and for this we turn to the calculation of process capability.
We know that a process has variations and we have established limits for the allowable values of the variables that we are measuring.
We must determine the relationship between the process variation and the limits that separate the good from the bad (specification limits). This ratio indicates the ability of a process to meet specifications and is what we call “process capability”
USL = Upper Specification Limit; LSL Lower Specification Limit
Margin = USL – LSL
Process capability: Cp = USL – LSL / 6 σ
If the values of the process variables are within the Six Sigma curve, the process will have high capability and there will be virtually no defects (low dispersion). This should be the goal.
When there is more dispersion (see four sigma curve, two sigma curve or the curve is simply not centred on the nominal value and there are values outside the limits), the process needs to be analysed and improved, and the process will produce defects.
At IKOR we have used this methodology to address processes as important as the construction and validation of systems for testing and measuring the capability of processes, achieving results and improvements that ensure more robust and stable processes.
In future articles we will explain the next stages, Improve and Control and how to address these in order to guarantee success in the improvement of a production process.