time studies to measure NOT 1

[IE777]

In my department we have 36 machines running over 4 crews. I want to estimate the non operational time, a % value, in the department. You can assume that the machines are similar. In order to do a baseline NOT study I will pick 4 machines for each crew and observe them for 6 hours each (4 machines * 4 crews * 6 hrs each = 96 hours). This is very time and labor-intensive. Once, I have the baseline I want to do NOT time studies every six months--but I do not want to spend so much time. One idea could be to randomly select machines for 30 minute time intervals and note down the NOT there....do this for 10-12 machines (I will use statistics to calculate my sample size), for each crew. Are there other suggestions? My goal is to accurately estimate the NOT in the department due to mechanical problems, shortage of materials from supply, and other small delays.

 

[tomwalz]

I think we’ve had similar discussions here.

Basically I don’t think the direct observation approach works. Look up the GE Hawthorne study.

We take total parts shipped and divide it by total hours and track that. If we put a stop watch on people they tend to work at an artificially high rate while observed. If there is a problem we have an argument about whether it should count or not and it tends to create a much more expensive reaction than it would normally. Finally we tend to see the direct supervisor(s) hovering and expediting more than they would normally.

Nothing evil or sinister. Just people being measured and wanting to make sure they do a good job.

I would use the total parts / total time approach and then spend enough time on the plat floor, break room etc. to become invisible and then you can identify the bottlenecks. I like Goldratt’s The Goal. It works for me and it is a good novel as well. He has other works that are much more technical.

Part of what costs us money are things that interfere with production which are really hard to see because they are accepted as standard practice. Rework is at least 5 times as expensive as doing it right. Any accident costs a couple hours and a few hundred dollars. Grainger doesn’t always get the spare part to you overnight so you might want to keep some on the shelf. And so on.

Tom


Thomas J. Walz
Carbide Processors, Inc.

http://www.carbideprocessors.com

 

[v777rk]

The best way for you to AVOID redoing the Time Studies every say 6 months will be use PMTS systems like MTM, MODAPTS etc.

The best way we have found is to combine the 2. Do a 1 time baseline floor observation and then use PMTS. In this way you know to add compensations if needed for difficult postures etc which people say is NOT well defined in PMTS systems.

Regards
V777rj

 

[gerhardl]

What is the target: the study itself, or smoothing out and improving the production?

For the study itself it could be that the 'human factor' is contributing more to downtime than ineffective methodes, as aswers above indicates.

On the other hand older/newer machines, other differences in equipment, material and routines will also of course contribute.

Why not use the 'keep it simple' (Scandinavian?) methode:

a) Establish a simple way of regularily checking and recording productivity pr unit (machine or team).

b) Publish the result regularily.(Measurement 100% of the time) Praise the good/top result (xx team manning machines 1 to 8 on top this month). Focus on the total result for the factory / department as such. Do NOT crate 'loosers'. Do not downgrade, mention or press teams with not so good results,or making lists with 'bottom ranking' (Could be a bad machine, an unplanned breakdown illness or new emplyees on learning curve)

c) Have a suggestion box or scheme for opening up for improvement suggestions, or better ask each person the two or three most likely things in daily work to improve the working situation for the individual.

????