I am an industrial engineering supervisor. I work for an automotive sensors, solenoids, and switches supplier for the automotive market. We are 53rd out of the 150 top suppliers.
I have some questions for you regarding OEE implementation and software:
Q1: Data collection can be a problem (Garbage in Garbage out) and the best of software solutions can fail for lack of complete and accurate data. Production people see any form of data collection as a hassle. How does OEE address this?
Arno Koch • In stead of adding an extra hurdle, lets find a way to reduce the operators burden, by gathering ONLY the data that is at the bottom line necessary. Normally the operator collects a lot of data that is not leading in any way to improvement. By eliminating such registrations it should be a relieve to the operator.
OEE is a real Operator tool that should help him or hear to do the job with fewer problems. It goes a little to far here, but the book “OEE for the production team” tells all about it. Maybe you want to join one of the workshops in Vegas September 99, I will be telling you all about OEE.
Q2: Lean does not strive for maximum output of the machine but rather producing to customer demand.
Arno Koch • Sure! What does the customer want? The right quality for the best price at the moment he needs it. Right? Question is: How to do so?
Would it help if your machines would run when you need them?
Would it help if it had the output performance what it was designed for, so you can produce cheap and quickly?
Would it help if you would produce zero defects?
Right? OEE makes visible where it is NOT like that….
What means ‘Lean’? It means: Using Muda glasses: detecting all your losses and solve them. OEE detects the losses on the actual spot were it is all about.
Generating money by adding value… where is this done? At the machine… How to look for losses when actually adding value? OEE! Viola…
Q3: So, output of a cell (that can have a group of machines and people) is tracked hour to hour against takt time and downtime recorded and addressed as a part of the cells environment. Wouldn’t OEE in this case be counterproductive?
Arno Koch • When you need to use your machine, your machine would be ideal if:
It runs without stopping, at maximum speed with no quality loss.
That is the definition of 100% OEE. Everything else is a loss. Big question is: Where am I losing?? OEE makes this visible and clear. And you are right: Some equipment has over-capacity. So if you need to reduce it’s speed because there is no demand… that is a loss.
But think about it: Are you able to run your machine on maximum speed? Mostly not, and mostly this is not desired. But than, are you able to run the machine in a controlled way on a lower speed, precisely on the needed speed to fulfill demand? Most machine cannot do so either… That is the REAL problem that needs to be solved!
Q4: I do agree that what gets measured gets improved. Software does a good job collecting data and giving it back in reports. What is lacking is trend data. What should good OEE software report?
Arno Koch • The first OEE software I developed (the ‘OEE Toolkit’) was designed to report the right information that you need to get real focussed improvement. It reports always based on the three issues as told above (plus more if you do it clever, and good software should do so…). The quality of OEE software is determined by the quality of its analyses and its ability to influence the behavior of its users. Unfortunately most software is designed from an IT point of view, rather than from a ‘improvers’ point of view.
Q5: Hmm, but our main measure of productivity is pieces per person hour and labor dollars per piece.
Arno Koch • So you measure the result of your productivity. Let me ask you a question: Do you know how much potential productivity you lost? And can you pinpoint where precise you lost it? And what to solve first?
Q6: To this we attempt to not speed up production but rather to reduce labor. Unless the capacity is needed – then we recalculate takttime.
Arno Koch • If you look through Muda Glasses, there is only one ideal takttime! Not running so means having a loss….
Q7: To this end we kaizen cells, attack scrap and rework, and stabilize machinery. We currently track and trend all of these areas and have been since 1989. And yes we do believe that all this leads to driving down costs.
Profit = Selling Price – Costs
Arno Koch • I would modify your algorithm a little:
Profit = Volume x ( Selling Price – Costs)
How to raise volume? Deliver your customer:
• Highest quality
• Lowest cost
• Highest delivery…
Again: Can you be sure you attack the right loss, where is the major LOSS (=cost quality delivery) to fix?
Hey, I am glad to have linked up with you (it’s a small world these days).
Nice chatting to you Bob!
I hope I could tickle your mind to ‘learn to see’ (Jones…et al.)