introtoarena.3 (1)

8/16/2019 IntroToArena.3 (1)

http://slidepdf.com/reader/full/introtoarena3-1 1/13

Intro to Arena

A Third Simulation



Model 3

We add more features to Model 2, also from Ch. 5 of Simulation

with Arena.The justification for adding features is quite simple the firstmodel was a !first protot"pe#, and we ha$e just gone through thefirst $erification%$alidation c"cle. At the end, we concluded thatthe s"stem actuall" operates on two shifts a da", and that the

second shifts has two operators assigned to the &ewor'operation, thus easing the (ottlenec' pro(lem we encountered.

Another pro(lem we ha$e to resol$e is the fact that the Sealer(rea's down, rarel" enough, and for short enough times that ourfirst appro)imation could ignore it, (e we now must incorporate it

in the model. *rom e)tensi$e records, we could determine thatthe mean uptime (etween failures was +2 minutes,e)ponentiall" distri(uted, and that the repair time had a mean of- minutes, also e)ponentiall" distri(uted.

ow do we add all this to the model/



Model 3

Adding Shifts.

0ur run length of 2 minutes is equi$alent to a little more than- shifts 134 -4 minutes%shift6. We now need to model theshifts e)plicitl", since the num(er of operators changes from oneshift to the ne)t. This will (e done (" adding a !&esourceSchedule# to the &ewor' resource.

The sealer failures are shift independent and will (e modeled usinga &esource *ailure or 7owntime.

8et9s start with the !:nspector# for &ewor' ; dou(le;clic' to openthe dialog. <otice that !Capacit"# has the $alue + this means

that there is onl" 0<= cop" of the resource a$aila(le, so that onl"one part can (e wor'ed at one time. Since we will ha$e twooperators during the second shift, we must change that.



Model 3

When "ou choose !Schedule# from theCapacit" T"pe list,"ou ha$e two newfields appearing. Thesecond pro$ides "ouwith the scheduling

options :gnore,>reempt and Wait.:f a capacit"decrease isscheduled and the

resource is idle, allthree options causethe resource to

(ecome inacti$e immediatel". :f the resource is in use, theresponse differs. :f !:gnore# is chosen, the resource capacit" is

decreased immediatel". When the resource is released (" the



Model 3

entit" to which it is allocated, it is placed in an inacti$e state. :tC0?87 happen that a request to increase capacit" arri$es (eforethe resource is released ; in this case the increase occursimmediatel". The effect is that the capacit" change ma" ne$er (e !seen# in its effects.

The !Wait# option will wait until the entit" releases the resource

(efore starting the capacit" decrease.The !>reempt# option preempts the resource (" ta'ing it awa" fromthe entit" controlling it, starting the capacit" reduction, and holdingthe entit" in an internal 1to Arena6 storage area until the resource(ecomes a$aila(le. At that point the entit" will (e reallocated the

resource and will continue processing.

When do we use each option/ There are no a(solute rules (ut asuggestion would ha$e us use the !:gnore# option if the duration ofthe scheduled capacit" decrease is long when compared to theprocessing time. We would use the !Wait# option if the time

(etween capacit" decreases is large compared to the duration of



Model 3

the decrease. The use of the !>reempt# option should (e clear.

:n the Ser$er 7ata area, in the Schedule window t"pe

&ewor' Schedule

<ow clic' on the !Schedule# (utton at the (ottom of the area. Anew !Schedule# dialog will appear, with @=nd of list on it.

Clic' on Add, to come up with a further dialog.

Ma'e sure "ou ha$e Capacit" +, 7uration -4

After clic'ing on Add, repeat the Add, (ut ma'e sure "ou now ha$eCapacit" 2, 7uration -4. Bou should ha$e the configuration in thene)t slide



Model 3

Ma'e sure that !7uration# hasa $alue. :f "oulea$e the fieldempt", thes"stem willinterpret the

$alue as infinit"and it will ne$erchange after(eing set.

As long as "ouha$e positi$eentries, thesimulation willrepeat through

the $alues "ou ha$e chosen.



Model 3We now turn tomodeling the

Sealer failures.Clic' on theSealer, clic' onthe !&esource#(utton to get

the dialog atthe right. :nthe !*ailures#area clic' onAdd. *ill in asindicated onthe slide. The?ptime and

7owntime follow the distri(utions we claim to ha$e gleaned fromhistorical data. <ote we chose Time;(ased rather than Count;(asedfailure. Count;(ased would depend on the num(er ser$iced<otice that we choose !Wait# consistent with the schedule choice.



Model 3

The ne)t thing

we will do issa$e morestatistical data; the defaultsare useful, (utnot all that wema" need.

The Commonpanel has aStatistics

module, whichallows us todefineadditional

statistics to (e collected. <otice the fi$e areas, denoting fi$e t"pes

of statistics. 8et9s assume we are interested in num(er of parts in$arious queues 1>rep D, Sealer, &ewor' areas6, and c"cle times.



Model 3

8et9s (egin (" setting up collection of

Time;>ersistent 1or 7iscrete;Change6statistics. Clic' on the Time;>ersistent Add (utton.

Chec' the Eueue (utton in the 7ata0(ject area. The dialog will change

to the one on the right. *ill in

7ata 0(ject

Eueue Select

Eueue <ame >art D >repF&FE

Sa$e 0(ser$ations to *ile

Chec'

0S *ile <ame in 7ou(le Euotes 1appears after chec' is done6

!>rep DFE.7AT#



Model 3

Add two more queue statistics

7ata 0(ject

Eueue Select

Eueue <ame SealerF&FE

Sa$e 0(ser$ations to *ile Chec'

0S *ile <ame in 7ou(le Euotes !SealerFE.7AT#

;;;;;;;;;;

7ata 0(ject

Eueue Select

Eueue <ame &ewor'F&FE

Sa$e 0(ser$ations to *ile Chec'

0S *ile <ame in 7ou(le Euotes !&ewor'FE.7AT#



Model 3

Add a Tall"

Tall" <ame

ShippingFTa

Sa$e 0(ser$ations to a *ile

Chec'

0S *ile in 7ou(le Euotes

!Ship.7AT#

And "ou end up with adialog that loo's li'e theone on the right.



Model 3=)ercises.

+6 Go to the help files and learn how to use the 0utput Anal"Her.

26 Go to the help files and learn how to use the :nput Anal"Her.

What can "ou sa" a(out the !fit# of the file part(prp.dst to the<ormal 7istri(ution/ What is the mean;square error/

I6 Modif" =)ercise + of Model 2 (" adding agent (rea's. The +hours are di$ided into 4;hour shifts. Agent (rea's are staggered,starting at J minutes into each shift. Agent lunch (rea's 1Iminutes6 are also staggered, starting I +%2 hours into each shift.Compare the results of this model to the result withotu agent(rea's.

-6 ?sing the model from =)ercise 2 of Model 2, set the inter;arri$al;time distri(ution to e)ponential and the process;time distri(ution foreach >rocess to <ormal, with a mean of J minutes. Setting thestandard de$iation of the normal distri(ution to $alues of +, 2 and I,

ma'e three different runs of +, minutes each and compare theresults.

introtoarena.3 (1)

Documents