If we remember our previous post, we made 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 Improve stage.
“Improvement is not possible if we do not measure what is happening and we do not know what is causing it”
The goals of the improvement stage must be two:
- To determine what all the important factors are (Vital X’s) that condition the variability of the process (parameters, materials, machines, people, suppliers, work flow, etc.)
- Develop the mathematical or circumstantial relationship between the inputs and outputs of the process (what is the relationship between cause and effect)
One of the best tools to use at this stage, particularly when we are talking about improving production processes where there are raw materials, machines, people, adjustment parameters, etc., is Experiment Design.
We won’t go into details about this tool because it is not the purpose of this post, but we will briefly explain what it consists of.
Experiment design is a methodology that allows you to establish a plan where you can test a variety of factors efficiently, and which allows you to extract conclusions, delimiting the experimental errors.
First of all you must identify the vital variables that you will analyse and then identify their values. For example, if you want to analyse the quality of welding in a reflow welding process, the variables could be: the alloy used, the amount of alloy used, the preheating time, the reflow temperature, etc. There could be more, but it is important to limit them to the vital ones.
Experiment design consists of establishing values for the variables and establishing controlled variations (they must always be within specifications) on these values to see and check at the process output the relationship the variation has with the final result obtained. In this way we can better focus the result of the process.
During the experiment design you can combine the variation of one or more variables, but you must maintain an order and above all be familiar with the level of interaction of each variable and its variation.
It is also necessary to replicate the test to make sure that the variation repeats the result time and again and thus, be able to certify that the experiment carried out has a solid foundation for its implementation.
A very important part is the practical implementation of the tests and the collection of data. It is advisable to prepare a small procedure that describes everything that must be prepared to carry out the test. The staff participating, the measurement equipment used, prototypes and samples to be tested, machines where the test is carried out, etc.
It is very important that you make sure to take notes of all the incidents that are observed during the experiment and the tests: anomalous behaviours, equipment the measurements are made with, people making the measurements, etc.
Everything that is suspected to have an effect on the result must be taken down, taken into account and also confirmed by replicating the test.
After the experiment design the results are analysed and the conclusions are established.
A very good practice consists of, before definitively implementing the actions derived from the conclusions, carrying out a controlled manufacture to verify the results.
At IKOR we have used this methodology to address processes as important as electronic circuit soldering processes, achieving results and improvements that ensure more robust and stable processes.
In future posts we will explain the next stage: Control.