Cancel
Showing results for 
Search instead for 
Did you mean: 

Maximising the throughput by changing buffer size

Creator
Creator

Hello ,

 

I am a student and building a model for brewery industry with 2 quality check . First quality check is for the returned bottles(if they are some cranked bottles they are rectified and feeded back to the main line) and second for the final product. I have introduce the failure at the filling pils beer in bottle section(fills_pils) and those defected bottles are collected at the store 1.Can you tell me what to do with the defected bottles? Should I drain them or something else? Also, I want to optimise the throughput of my model by changing the buffer size.

But i don't know how to do that. Can you please help me in  this.

I have attch the model fro reference.

 

Thanks in advance.

13 REPLIES

Re: Maximising the throughput by changing buffer size

Gears Esteemed Contributor Gears Esteemed Contributor
Gears Esteemed Contributor
the defected bottles you should move to a drain for destroy them.
The size of the buffers you can change using the experiment manager. See help and the examples in the example collection.
Steffen Bangsow
freelance simulation specialist  
web: www.bangsow.eu
mail: steffen@bangsow.net

Re: Maximising the throughput by changing buffer size

Valued Contributor
Valued Contributor

Hello Jui,

 

As far as you're first question, to me this depends on what you are trying to model. I don't know if you are modeling some kind of real system or not. I am also not very familiar with this type of manufacturing industry, so I can't really say what they would do in real life. 

 

It seems to me the defective bottles would be recycled in some way. To do that, I would output from the store to a singleproc, or group of singleprocs which model the recycling process. Once they have been recycled you can reintroduce them somewhere near the beginning.

 

Or if you are modeling this plant as just throwing away the defective bottles then you can just drain them. If you want to recycle them and need more explanation on that you can ask me further questions on how to do that.

 

For your second question, optimizing is actually rather easy. Under the tools library there is an object called the experiment manager. Insert that into your model. 

Once you have the experiment manager open its dialog and we will change a few settings. For you to optimize throughput based on the buffer size you will need to give it an attribute to measure throughput with, which will go in the "Define Output Values" table. I am assuming that your good bottles go to store. As long as that is the case than we will use one of the Store's attributes to measure throughput. Fill out the "Define Output Values" Table as such:

ice_screenshot_20160624-095447.jpeg

 

Next we need to fill out the "Define Input Variable" Table. Check the "Use Input Values" option, then open the table. The variable we are changing is the buffer's size so we will enter that in as the input value. See screenshot below:

 

Re: Maximising the throughput by changing buffer size

Valued Contributor
Valued Contributor

Apparently I was a little too long winded. Sorry about that. Here is the rest of it.

 

 

 

ice_screenshot_20160624-095722.jpeg

 

Once you have defined the input variables use the "Define Experiments" table to define the values you want to test. Such as a capacity of 5,10,15,20 so on. And Remember to set active to T for true.

 

I would change the number of observations per experiment to 10 or something along those lines, then hit start. It will run and then spit out a report with a graph comparing each capacity to the resulting throughput. Also, as a note. With the experiment manager you either need a fixed number of entites being produced or a finite simulation run. There has to be something ending the simulation at some point for it to work.

Re: Maximising the throughput by changing buffer size

Siemens Phenom Siemens Phenom
Siemens Phenom

Hi Experimenter,


a failure is mostly not suitable for the modeling of rejects. The exit control of fill_pils conatins a good idea: If uniform random number (generated by z_uniform(1,0,1)) is less the defection_rate then the movable unit is moved to a Drain object. At the end of simulation you can use the Typestatistics to evaluate the number of rejects. I also recommend replacing your Store objects by Drains. You can investigate the results of the Quality_Control.


The most important buffer seems to be Buffer1, but the use of this buffer depends on the processing time of the subsequent station. Maybe it is not the right value or the right statistical distribution.


I unchanged saved your model with Plant Simulation 12.0, such that all members can open it.

It is an interesting project.


Regards,
Peter

 

Re: Maximising the throughput by changing buffer size

Siemens Phenom Siemens Phenom
Siemens Phenom

Hi Experimenter,


a failure is mostly not suitable for the modeling of rejects. The exit control of fill_pils conatins a good idea: If uniform random number (generated by z_uniform(1,0,1)) is less the defection_rate then the movable unit is moved to a Drain object. At the end of simulation you can use the Typestatistics to evaluate the number of rejects. I also recommend replacing your Store objects by Drains. You can investigate the results of the Quality_Control.


The most important buffer seems to be Buffer1, but the use of this buffer depends on the processing time of the subsequent station. Maybe it is not the right value or the right statistical distribution.


I unchanged saved your model with Plant Simulation 12.0, such that all members can open it.

It is an interesting project.


Regards,
Peter

 

Re: Maximising the throughput by changing buffer size

Creator
Creator

Hi,

 

I tried what you said(same description for input and ouput values) but my throughput is same for all buffer capacities.

 

Capture.PNG

How can I optimise the model?

Re: Maximising the throughput by changing buffer size

Valued Contributor
Valued Contributor

Ok actually I just realized an error in my answer that may have caused the problem, did you set the source to only produce a certain number of parts, or did you set the simulation to only run for a certain amount of time?

 

If you when with a limited number of parts then that type of experiment (Throughput) won't work. Since the simulation is limited only by number of parts, you will get the same result every time.

 

If you limited it by time, then I'm not really sure what the problem is. My only suggestion there is try more extreme values.

 

Try 1,2,3,4,5 as your lower instead of 15, and 100, 125 as your upper limit.

Re: Maximising the throughput by changing buffer size

Creator
Creator

Hi,

 

I am not limiting the source to produce a limited number of parts.Also, I tried with smaller values aswell but still having the same throughput.

Can you suggest any other technique to optimise the model?(regarding cost or energy)

Re: Maximising the throughput by changing buffer size

Valued Contributor
Valued Contributor
I find it odd that you cannot get any different values in that case. I feel as though there is something going on that I am not aware of there.

You can optimize energy, but you would have to enter energy information into the energy tab of every station in order to begin doing that.

I'm not sure about cost, I think you would have to introduce your own attributes into the model for that.