INTRODUCTION:
SIX SIGMA process improvement set of tools, originaly from Motorola corporation. Proces is improved throuh problem root cause identification, resulting further in lower level of defect or delays in manufacturing and bussiness processes. Project is justified only if the root cause is not known, if it achive major change in customer perception or if it deliver major finantial benefit. Project should be carried out by the trained experts, that have symolic qualifications (BB, GB etc.), projects are manged by Black belts, trained team members are caled Green belts, but it is possible to have team members without formal training. Project should have phases, known as DMAIC (Define, Measure, Analyze, Improve, Control). Method is using impresive list of tools, some of them are scientific, so sometimes DMAIC is called “scientific” method. Last decade is bringing new rising trend, merging with Lean methodology (Lean manufacturing, TPS), also known as Lean Six Sigma, forming set of tools vital for structures ussually known as „Continiuos improvement management“, „Bussiness exellance management“, with main purpose to achive processes exellence and finantial benefit etc.
METHODOLOGY DETAILS:
There are two basic models, one is rarely used model, DFSS (DMADV), searching for product/process design improvements, and another model that is used much more often, aiming lower level of defects/delays in processes that use DMAIC:
Define:
We should start with problem statement, followed by project scope, calculations about potential benefits, using tools like CTQ, moving problems to finantial field, using tools like Min/Max (cost/benefit calculations, answering to question shall we procced with project or not). During this phase it is possible to use a list of speculative methods like QFD, KANO, or ABC model, ARMI model, or CAP model. We should form project team, Project Charter, milestones, process mapps,etc. it is of vital importance to choose projects in a proovable manner, that we can through lower level of defect or delays in manufacturing and bussiness processes, to achive major change in customer perception (quality), or a kind of benefit that would be much over project budget.
Measure: in this phase we are collecting data, measurements, to establish starting “baseline performance”, starting level of our project Y. We could use historic data, but we could also organise to collect propper data, so starting step should be to create DCP (data collection plan). Data could be continiuos or atribute, we are processing process variations, VOC, VOP, hystograms, distributions, central tendencies, data processing tools (MiniTAB, MS Exel, etc.), common and spetial causes of variations, Control charts, sample size calculations, all that we need to calculate process capability and stability, with parameters like Cp (proess capability), Cpk (process stability), Zscore, DPU, DPMO, ppm etc. Within this phase we also calculate MSA, showing us levele of measurement variations themselfs, that also show us what portion of measurement variation is up to operators, and what portion is up to equipment and methodology, resulting together in so cold GR&R, so, if that one is less than 10%, we could trust our data and proceed with procedures.
Analyze:
In this phase it is crutial to establish cause effect relations. We are dealing with causes (individual Xs, inputs, factors) that may have impact on our effect (our project Y, output, responce). We are using all kind of speculative priorititaions methods, searching for structure or priorities among our potental causes (C&E diagram/Ishikawa/Fishbone, 5 Why, combination of C&E diagram and 5 Why, C&E matrix, etc.). Whithin BB area we are using heavy artillery to proove cause effect rellations, known as Hypothesys testing, not possible without software like Minitab. There is a long list of hypothesys that we can use, some of them are rather simple (like distribution normality testing), but some of them are more complex, like ANOVA or (multiple) Correlations with reggressions, commonly used in thesys or scientific articles, capable to isolate correlations and calculate precise impact (up to the level of clear formula, like Y=f(X1, X2, X3,…), sometimes called Transfer function).
Improve:
Here we should first involve best available experts, who know process the best. Later we should use again speculative prioritetisation tools, but this time, together with some creativity tools (Brainstorming, 6Hats, …), AHP, Pugh matrix etc. As a very powerfull tool BB are using DOE (desigh of experiments). DOE is used in problem solving, but also for the purpose of product or process design. Point is to achive answer, through limited number of experimenats, how shall we set up our process inputs (factors), to result in desired level of process output (responce). Sotware (Minitab) is making statistical calculations, searching for correlations (mostly mixture of ANOVA I Reggression), and finally capable, under “Optimiser” part of application, to offer factors setting to achive desired responce (max, min, target). After this we should conduct pilot trials, we should report about improvements, comparing to the baseline performance etc.
Control:
Pilot and implemention should go together with Control plan. Control plans have detailed description of process but also all controls involved, inspections and measurement that should be done, records that should be used, who will do and what kind of data processing, statistical studies, responce plans etc. Monitoring should involve „scorecard“, Control charts (I-MR, XS, P, U, etc.), for sustainable process we should also use SOPs, etc. Project should be closed with final calculations, project closure, reports, celebration and recognision for the people who gave major contribution, and finally transferring process back to the process owner (Process Sponsor) etc.
.
