Cancel
Showing results for
Did you mean:

# the MUs block on the buffers with infinite capacity

Experimenter

what's happening is a strange thing. If I set a infinite capacity on  the  buffers the MUs block on the first two buffers, and only after they are filled the MUs goes to the machines and the simulation procede.

but if I set a finite capacity it doesn't happen, and MUs goes normally in the simulation.

WHY?? (look at the attachment)

8 REPLIES 8

# Re: the MUs block on the buffers with infinite capacity

Phenom

I think it happens because the Source produces parts with the setting 'Number Adjustable' and the distribution represents a range in time instead of an interval.

Thus, the creation of all parts is scheduled before any other event. And an infinite buffer right next to the source allows it to do that.

If you have a finite buffer, the source is blocked and cannot produce any more parts.

See help topic on 'Number Adjustable'.. That might be the reason for this behaviour.

# Re: the MUs block on the buffers with infinite capacity

Experimenter

but I continue not to understand why It happens, what do i have to change?

# Re: the MUs block on the buffers with infinite capacity

Phenom

That depends on how you want to use the Poisson's distribution for creating the MUs.

If your intent was to have the Interval between generation of MUs to be based on the distribution, you should change the Time of Creation in the Source object to 'Interval Adjustable' in place of the 'Number adjustable'..

# Re: the MUs block on the buffers with infinite capacity

Experimenter

But this don't solve my problem. Look the attachment.

# Re: the MUs block on the buffers with infinite capacity

Phenom

From what I understand now, is that the problem is that the rate of creating the parts is very high compared to the processing time? Please correct me if I have a wrong understanding..

The lambda parameter is set to a very low value here.

The lambda value does not represent the number of occurences per second.

It represents the time interval between occurences in seconds.

# Re: the MUs block on the buffers with infinite capacity

Phenom

@antonitrt

You are creating 4 different parts ( A,B,C,D)

with a randomly set frequency.

Ready to leave the source, each part is directed

through the flow control ( FCSource)

either to Buffer1 or Buffer2 .

With infinite buffer capacities there is always enough buffer capacity to

With finite ( in this case 2X8 ) buffer capacities the source will block

as soon as one buffer is filled with 8 parts and the last part to leave

is directed to the full buffer.

The source will produce the next part as soon as the filled buffer releases on part

This is according to the set tablefiles between 20 and 36 min.

With infinite capacities it will go on (endlessly) producing parts.

I have checked for 1 min simulation time

the source produces:

16 parts ( 2 x 8 places)

1334 parts  (infinite)

What you are seeing in the infinite version is just the

continuous production of (thousands of) MUs whilst of the stations are finishing

there cycle ( after 20min  -36.0min) .Nothing else !

Creating these amounts of MUs takes (real) time for the system but has no effect

on the simulated data !

Therefore the suggestion I can give in this case is  not to use infinite buffers

in connection with Sources with a very low cycle time as this constellation is just consuming

computer resources for not required MUs in the infinite Buffers.

Set the buffer capacity to aligned size which does not affect your Simulation.

Highlighted

# Re: the MUs block on the buffers with infinite capacity

Experimenter

ok. Now I've understood.

last questions I want to ask you is:

1) If i want the average processing time, have I to look in the statistic report-->working time--->mean value? Or is there an other way?

2) for the average time between two arrival, have I to llok in the statistic report og the resource---> waiting time---> mean value?

Phenom