Assembly Line Balancing Questions -II

[sigtan]

Can someone please guide me for a high variety and demand mix environment, where cycles times are varying up to 200% with different product variants, how should one go ahead about balancing the lines? Do you take peak cycle times at the stations or consider an average or try to take a mode value??

(The line involves just one machine – wave solder to solder boards in a big batch and rest of it is a manual assembly line where we are assembling components into a circuit board. The cycle times shoots up between a simple and a complex board and also based on the technology – surface mount or through-hole. I have total 5 station and due to the precedence it is difficult to distribute work load around.)

Thanks for your help.

 

[tygerdawg]

My background is mechanical also, done a lot of assembly ops. I learned the IE stuff mainly by association with good guys. You may want to consider two options:
1. Purchase one of the more reasonable discrete event simulation packages (e.g.,

http://www.simul8.com

has one that I was interested in at one time). This will allow you to build a virtual assembly sequence model with cycle times, queue lengths, resource scheduling, etc. Run the simulation and then do quick "what-if" analyses to optimize...up to a certain level. This is the dumb MechEngr's way of doing Classical IE work without the knowledge. For example, you may find that multiplying the long-cycle time stations and putting in sufficient queue lengths will smooth out your flow. I typically found that going through those variables will generate a "resonance" point of max throughput.
2. If you pursue a Lean one-piece flow philosophy, then it's necessary to do the Takt-time thing. Might not be practical in a board shop environment, but look at

http://www.lean.org

for resources.

TygerDawg

 

[PSE]

Do to your product mix, a single analysis will not likely work for you. If possible, try creating "families" of similar type assemblies (complexity, number of parts, mix of through-hole and surface mount etc) and run your studies on those. Setup/changeover time will also be key in your studies.

Good luck

 

[sigtan]

Thanks TygerDwag and PSE. Since I do not have product family groupings right now, my first step will be to form those. I was trying to work around it for long to aviod the PFA (production flow analysis) due to the huge variety of boards, but now is the time to immerse into it completely.

Thansk for your motivation.