Parts Produced Vs Total Downtime

[BuickBoy]

Parts Produced Vs Total Downtime - Graphical Interpretation

Hi all!

I'm working on a project analyzing downtime for a manufacturing floor. The department makes between 6 and 8 millions parts per month with 10-18 thousand minutes of unplanned downtime collectively (changeovers, maintenance, jams, material changes, etc.).

I've attached a representative graph I am currently using. The goal here is to trend downtime and trend number of parts produced. I feel the graph does a good job at trending both and shows a decent comparison between the two.

Some ideas we had for improvement was instead of showing total downtime minutes above its point each month, displaying a scaled value for downtime minutes/parts produced (which is a fraction of a percent without scaling) as a value that can be displayed as percent change each month. That would allow us to see if we improved or not month to month. However, the value isn't very easy to interpret.

The downtime line should never rise above the number of parts produced. Ideally they would mimic each other or the gap would get larger indicating downtime is going down and parts are going up (more efficient changeovers, less maintenance, better operator training etc.).

I use other metrics such as OEE (Overall Equipment Effectiveness) to further analyze the area but right now I'm focusing on being able to make a comparison showing how well we improved or did not improve by this representative.

This data is viewed monthly by department managers, plant managers, and occasionally vice presidents.

My point is... how do you guys view this graph? What improvements would make this data easier to understand?

 

[pmsrw3]

I must be dense, but I don't get why the two plots should mimic each other. Shouldn't they go in opposite directions? More downtime means fewer parts? And why should the downtime line never rise above the number of parts produced? If, for instance, the floor was down all month, downtime minutes would be huge and parts produced 0.

Anyway, I would try plotting downtime against parts produced, connect the points in time order and label each with the month it represents. You want to see the curve head down and right. Umm, you might also want to divide by workdays in the month, so that February doesn't look different from January just because it's shorter.

 

[gsal]

Yeah, I would play around with the graph a bit...

First, I would clearly label the primary and secondary y-axis...it took me a while to figure which one is which...yeah, thanks to the label data points...I wish you did not need those data points, they get in the way.

I wish the plots were not just about on top of each other...this can be arrange by changing the scale on one of the axis...

...or you could somehow normalize the data to some meaning production-per-day number and minutes-down-per-day...then, you plot those per-day numbers for each month...that way, as suggested, one month will not look so different from another just because of too many holidays in it or some other very well justified plant closing, etc. You could also plot the horizontal line you used to normalize your production to come up with production-per-day and see how your production moves about such line...the same goes for downtime...