Model 2 - MoreSteam.com
Contents
1. S Foe Time Series Station ra wi Debug I Clear fa Delete Delete B 7 far Geman Export eport Report Report Fie t Mode re F Up Links Sew al nb V motn SS J Block Name Queuet Add Process Parameters FIFO Max Items 9999 Inal On Hand Probability FIFO 0 0 1 0 Balk At This Time never Kitting Mode Optional Ki Item Name K Quanity Al Add Optional Parameters Cost Distribu on CQ Parameter 1 Distribution c0 Parameter 2 Distribution C 0 Figure 3 Queue Block of the Basic Model If Auto Links is turned on then Process Playground automatically connects the queue block to the existing demand block and the queue block becomes the selected block when it is dragged into the model workspace Notice that the flashing red line is now under the queue block Also notice that the Model Data area has been updated to reflect the data related to the new queue block Queue Block Data The queue block data consists of several different pieces of data which define how the queue will behave Some of the options will be covered in later sections but here are the basics e Block Name Just as with the demand block you can change the default name to another name that is more representative of your process e FIFO Max Items If your real queue has an upper limit to the number of items it can hold you can enter that limit here All queues operate as FIFO queues unless y2. When you are done you sieckwame Demando PROCESS PLAYGROUND should have something Add Demand Parameters Start demand at ime 0 End demand at ime end that looks like the picture to the right Demand Interarrival Time Demand Time Distribuson N 1 1 Notice that the demand block has a red flashing roe bar under it That bar is isi critical because it a indicates which block in the model is selected Item Name default When a block is Figure 2 Demand Block of the Basic Model selected the Model Data section changes to display the data associated with the selected block In this case you can see that the data displayed is related to the demand block Demand Block Data The demand block data consists of four groups of data e The block name At the top of the Model Data section you will see the default block name that is assigned when you first drag the demand block into your model In this case the default name is DemandO You can click or double click on the block name and change it to a name that makes more sense for your model Note that your changes to the block name are reflected in the text under the demand icon e The distribution Process Playground supports six different mathematical distributions for use in your model You can select your distribution type by entering the app
3. Process Playground User s Manual Revised 12 26 2011 Copyright 2011 MoreSteam com LLC p Contents Table of Figures EI DD o m M 4 avt goTo Ulett tela PP IEEE EE EET TETTE CDI DIS UU 5 User InteriaGe OVErVvieWw p 6 Model T The Basic Model etre aaaea ignc ed reb Ohr ad Less ec e EHE Cerda ba ee ane Boer rage 8 Demand Block D ta RM Et 8 QUEUE BIOCK DOO ssepe hese aa a Ea E EREE ERRE EEE E REEE E Aea AE T ERa OE E EEE KERERE ERER 10 Activity BIOCK Doro ep E 12 TheExit QUEUE oo treten E psc sec ies to eStore un fosse e creme 13 RUNNING TNE MOE T E B L M 13 The Results of the Simulation Run isses eene thnn seen nnns sse n tasa ss se tasa si sr nass seen nada sentada sess tasa ss sna dis 15 Model 2 The Basic Model with Variation sessseesseseseeeeeeeenen enne nnne nennen nnne en nene en nennen inneren nennen 17 Random Distributions in Process Playground cesses eene eene enne en nenne na nisse snnt nein nasa nnn 17 e CiTheconstant distriDUtlOn iren esee Ere eoe o Fei eer na E ERR Dee eo pae n ai cepe poa een xag epa ako 17 e rThe uniform distriDUtiODi x iiss ciel rente onte e ennt t eet reete rea besa cewd chases TASE Ri a P EAS ER RR IRE 17 K The Erlang dee Vut 17 e E The Exponential distribution scoeden aaeeeiai nenn nnn snis en nena nass s sss aiaa
4. LLC Model 7 Complex Demand Patterns ccccssssccessssseceessneeeeeseaeceeseaaeeesseaaeceeseaaeeseeeaaececseaaeeeeseaaeeeeseaaeeeseeaaes 32 Model 8 Batch Processing 2 rere cre perito see ne epa n eae ra Ce egy d tee rae gea pu Exe nene eere puru d se eee repu ve cna eU ince 32 Model 9 The Core Process Pull BOCK sariini ai a a 32 Model 10 The Resource Block sessseeseseseseseeeeeee ireen aa aiaga teen tenente snnt entren enne 32 Model 11 The Decision Block seseseseeesseseeieseeeeeee eene nennen trennen aaa aa nnns st nn nter nnne 33 Model 12 The Note BlOCK rire aia a rta quer da ace aeree Chapa ee Een ROTER DER eade 33 Copyright 2011 MoreSteam com LLC 3 Table of Figures Figure 1 Process Playground tette esto ad odd esent eet essaie une s ete ov sauce er eda es rae e Ree NU cR s 6 Figure 2 Demand Block of the Basic Modlel sess enne nnne ne nennnn nnns nnne tini sa nsns enne ranas 8 Figure 3 Queue Block of the Basic Model sssscccccceeessessaeceeecessesesaeseceeecusseeaaeaeeeescesseseaaeseceessessesaaeeeeeeeseeegs 10 Figure 4 Activity Block of the Basic Model esses nnne enne nnne nenne tnra asses enne ne asas asse nean 12 Figure 5 The Exit Queue of the Basic Model ccccccccessccecssssececsssseceesesececessececsessececsssececseseeeeseesecseseeseeeesesaeees 13 Figure 6 The First Run
5. Playground Because Process Playground is web based you never need to worry about installation updates or compatibility The program runs in any web browser that supports Adobe Flash Player version 10 1 or higher Your model files are stored locally on your computer so you have complete control over them The model files are also very small typically less than 100 kilobytes so they are quick and easy to share as well This manual will walk you through all of the features and functions of Process Playground as well as giving you tips on good modeling techniques There are also training videos that are associated with each of the chapters in this manual as well as additional features of Process Playground not covered in this manual The training videos and sample models are also available from the Process Playground launch page Copyright 2011 MoreSteam com LLC 5 User Interface Overview The Process Playground screen is split into 4 main sections Each section groups together similar functions or actions needed for the program e Model Controls The top left section includes the model controls This section itself is split into three sub sections which provide the controls from left to right for managing model files building models and finally running models in the buttons at the right of the section e Building Blocks The top right section includes the 7 building blocks available to model your processes These 7 building bl
6. oo Sees Now you will have to edit all of the item names in the model The demand block and Queuet will have the item name pot The bottom queue Queue3 will have the item name coffee The KIT QUEUE will have two items pot and coffee and the activity and exit queue will have the item name brewed Set the run length of the simulation to 10 time periods at the top of the screen and set the Initial On Hand field in the bottom queue to 10 as well Block Name KIT QUEUE Add Process Parameters FIFO Max Items 9999 Ini al On Hand Probabiity FIFO 0 0 1 op Balk At This Time Kitting Mode Optional Add Optional Parameters Cost Distribuson Cio Parameter 1 DistribuSon c0 Parameter 2 Distribuson CH Now you will need to setup the kitting function in the KIT QUEUE Select the KIT QUEUE and scroll through the two items you have built in this queue coffee and pot Depending on the order you entered the two items one of them will display the field Kit Item Name and the other one will not notice that the Probability FIFO and Balk At Time fields are also set only once for the queue not for each item In the Kit Item Name field enter the name brewed This will be the name of the new item that is created from the joining of 1 pot with 1 unit of coffee For each Add Process Parameters FIFO Max Items Probability FIFO 0 0 1 0 Balk At This
7. 17 e N The Normal Distribution 3 1 n toten recti eene gue sabe eec hd eden ra iet ee Page de Faber detec 18 e T The Triangular Distribution cesses eene enne eu EEEE DEAE STEERER eTEN A sensi ansa sanas 18 Experiments with the Basic Model esses esee nennen nnne esent insana stes entitas a assa ss sss aaa E 18 Model 3 Adding Items and Controlling Flow by Item Name cccccssssccccecesseseneceeeeecesseseeaeeeseessessnseeaeeeeeeseseees 22 Multiple Items in the Demand Block esses nennen nennen nnne neni nasse en nenas a sans arse nean 23 Multiple Items in the Queue BIOCK ivii neaei ieden a aea aiad aaa eieaa aii aaas 23 Multiple Items in the Activity BlOCKk usseseseeeeseeeseeee eene nnne nemen nennen nnns enne nh anna nass enne an 24 Experiment Changing Demands and Cross Training esses eene nnne enne nennen nnne 26 Model 4 The Balking Queue and the Non FIFO Queue eese eene nnne neni nnns sre 27 Non FIFO QUU B haVi f e V 28 Balking and Exiting the Queue ierit centem accepere co eee ase eee a a A 28 Model 5 The Kitting QUGUue o ot aree tee rn verb set e eee vU ue Dae ue E Eee UR NUES ieee 30 Additional Training Videos tette eate deuda eater teen saco dete ede santet tek de aeaa iee aaaeaii iai Model 6 Replenishment Pull Systems Copyright 2011 MoreSteam com
8. Max Items 9999 In amp al On Hand The buttons for managing the items in a block are located Probability FIFO 0 0 1 0 below the default name There are buttons for scrolling Balk At This Time never through the items in the block adding a new item or deleting and item If you click on either of the scroll buttons Kt tem Name now nothing will happen since there is only one item in the duda block Add Optional Parameters Cost DistribuSon Parameter 1 DistribuSon Multiple Items in the Demand Block Parameter 2 Disribulon Go ahead and click on the New Item button now When you do you will see the Item Name field change from default to NewPart If you click on the scroll buttons now you ll see you now have two items in the demand wit block Go ahead and click on the text NewPart and icum die rename the item Part B Then scroll to the default item ail and rename it Part A Your demand block in the model Figure 13 The Default Item should now show two parts in its parts list Part A and Part B Now scroll to Part B in the Model Data section The default distribution type should be C for constant Go ahead and change that to E to match Part A Now both Part A and Part B will be arriving with an exponential distribution By adding the additional item to the demand block we have created the same effect as adding a second demand block to the model Go ahead and ru
9. Time Kitting Mode Optional Kit tem Name C K Quantty Add Optional Parameters Cost Distribuson Parameter 1 Distribution cQ Parameter 2 Distribution item Name jefe item you will see the field Kit Quantity This field is Figure 26 Queue 3 used to tell the kitting queue how many of each item are required to make a kit For our model we will leave both of these values at one however we could easily change the coffee item to 2 if we required two scoops of coffee to brew a got Figure 25 Kitting Queue Copyright 2011 MoreSteam com LLC 31 The figure to the left shows the kitting queue Model Data When you run the model with animation turned on you will see the KIT QUEUE pulling in one unit of coffee and then waiting for the demand block to generate a demand event called pot When the pot item joins the coffee item in the KIT QUEUE the two items are combined into a brewed item which is sent to the brew activity For this lesson and the following lessons you may want to refer to the training videos that are also provided on the Process Playground launch page These videos will take you through all of the functions of Process Playground while you watch the action on the screen Additional Training Videos Model 6 Replenishment Pull Systems Replenishment pull systems are a common Lean tool used in many projects Process Playground has been designed to make it easy to creat
10. feature in later sections The next element of your first model will be a queue block The demand block should always have a queue block following it or you may risk losing the demand in the model If a demand block is connected directly to an activity block and that activity block is busy when the demand arrives the demand will be ignored and lost in the model Therefore it is always good practice to have a demand block flow directly into a queue block which can accumulate demand as needed Process Playground has a feature called Auto Links which helps speed model building The on off button for Auto Links is located in the center section of the Model Controls area of the screen With Auto Links turned on Process Playground will automatically connect the next block you drag into the workspace with the block that is currently selected Remember you can see which block is selected by looking for the flashing red line below the icon in the workspace Since the Demand0O block is the only block in the workspace it should still be selected and highlighted by the flashing red line Now drag and drop a queue block somewhere to the right of the demand block You should have a picture that looks something like the picture below Copyright 2011 MoreSteam com LLC 9 Model a aiani Co i a Complete PROCESS PLAYGROUND Ope E Save Print Zoom Auto Manual A Rep Pull Run ws 7 M Modei Ji Model A Links Li ad Link Je ase d e N
11. process will perform and react to different types and ranges of inputs All processes have variation Variation in demand comes from our customers and variation in processing or service time comes from our process With Process Playground you have the ability to choose from several random distributions to help model the true variation in your process The next section describes the random distributions available Once you have reviewed the distributions we will run some experiments in the basic model Random Distributions in Process Playground The different distributions supported include e C The constant distribution Events following this distribution will happen without variation with an inter arrival time as given in the top data field Interarrival Time mean So with a value of 1 for the mean a new demand event would occur every 1 time period Note that inter arrival time refers to the time between arrivals of a unit of demand U Theuniform distribution When the uniform distribution is selected the minimum and maximum fields will become available allowing you to enter the minimum value and maximum value for the uniform distribution This is not an integer uniform distribution so all real numbers between the min and max will be equally likely to be generated from the random distribution K The Erlang distribution When you enter K as your distribution type you will be using an Erlang distribution with the mean you ente
12. total number of demand events generated by the block e The Queue Block o Current indicates the current number of items in the queue o Mean indicates the average number of items in the queue o Min and Max indicate the minimum and maximum number of items in the queue Note that the exit queue will typically show 9999 for the minimum value as it never has a minimum and only accumulates exits Copyright 2011 MoreSteam com LLC 15 Count shows the number of items that have entered the queue TCT mean is the calculation of the average total cycle time lead time to get to the queue Note that the first queue shows TCT mean 0 00 since the demand events flow directly to the queue with no delay The exit queue shows TCT mean 1 0 which reflects the 1 time period of processing in the activity block Since there is no variation in our base model there was no queue time e The Activity Block o Util is the percentage of available time spent working on items It stands for utilization Idle is the percentage of available time spent waiting for work Setup is the percentage of available time spent setting up work Setup time is considered non value add and does not count as good utilization of the activity e Total Cycle Time Histogram o The graph to the left at the bottom of the screen shows the distribution of Total Cycle Times for each individual item The Total Cycle Time is the time the item takes to arr
13. Available Minimum Batch Size Maximum Batch Size Setup Time 9 Processing Time Dist Cit Add Optional Parameters Cost Distribu on c0 Parameter 1 Distribuson 0 Parameter 2 Distribution co Item Name Figure 17 Model 3 with Two Items can only process Part B items The two activities share a common queue in the model but the queue actually maintains the first in first out FIFO order of all the items in the queue and will send the appropriate Part A or Part B as required when either of the two activity blocks becomes available Go ahead and run the model now Notice how items are being worked on by both activities now If you would like to see how many of each item are completed you can look at the Count value in the text below each activity block or you can create Part A and Part B in the exit queue and be able to see the individual counts for both items there Copyright 2011 MoreSteam com LLC 25 Experiment Changing Demands and Cross Training Model 3 now represents a typical process where we have two types of work arriving Part A and Part B which are then worked on by separate resources the two activity blocks But what happens if demand for Part A increases while demand for Part B decreases This is a fairly common occurrence in real processes and one that we can model quite easily with Process Playground Click on the demand block in your model and then scroll to Part A in
14. C 14 Model Controls Fun tong Oo qe PROCESS PLAYGROUND oe Link Debug Delete Delete Mode Block Up Links Building Blocks e ow Manual A Link Block Name Queue3 Add Process Parameters FIFO Max Items 9999 Inal On Hand 0 Probability FIFO 0 0 1 0 1 Balk At This Time never Kitting Mode Optional Kit Item Name K Quanity eN Add Optional Parameters Cost Distribuson cQ Parameter 1 Distribuson c Parameter 2 Distribuson cu Item Name default y Scroll Item Scroll item K lt ae are Queue Lenath or Servers Utilized Time Series Plo 96 0 48 0 1 m ad Model Workspace Now we will take a closer look at all of the items on the screen First of all notice the large pile of black squares above the exit queue and the black plus sign This shows that all of the WIP made it through the process and is in the exit queue We did not have any WIP accumulation in our first queue because we did not have any variation in this model Demand arrived every 1 time unit and it took exactly 1 time unit to process each item NewModel MSPPG Version 1 7 2 Model Data Figure 6 The First Run of the Basic Model The Results of the Simulation Run Now let s take a look at some of the information available on the screen after a run e The Demand Block o Count shows the
15. FIFO behavior with a simple input Click on the first queue again and now take note of the field called Probability FIFO The default value for this field is 1 which corresponds with 10096 pure FIFO operation If you enter 0 5 then 5096 of the time the queue will take the next in line following strict FIFO rules and the other 5096 of the time randomly determined the queue will randomly select an item from somewhere other than the first position in the line If you enter 0 0 then the queue will never operate as a FIFO queue and will always randomly select an item from somewhere in the queue To see the effects of this go ahead and click on the first queue in the model and then enter 0 5 for the Probability FIFO value Run the model and Total Cycle Time Histogram then click on the exit queue Remember you can turn off animation using the animation button to make the model run much more quickly Your results will vary but you should have a Total Cycle Time Histogram chart which shows a lot of variation in the TCT times for each item This reflects the 5096 non FIFO Figure 21 TCT with Non FIFO Queue random behavior of the queue If you rerun the model with the Probability FIFO value set to zero you should see even more variation in the TCT times Balking and Exiting the Queue Our model is now at a point with Probability FIFO value set to zero where the items in the line are experiencing very large variation in the amount of time t
16. and process modeling at the same time this guide will follow the tried and true method of learning by doing As you progress through this tutorial you will build your own library of reference models The reference models wi Copyright 2011 MoreSteam com LLC 6 cover every feature and option of Process Playground allowing you to become completely familiar with Process Playground because you will have recreated every model in this manual Please create a directory on your computer to store your models as you go If you name the models to match the names used in this manual you will be able to build a reference library to help with future modeling projects Ready to go Let s begin with the basics Copyright 2011 MoreSteam com LLC 7 Model 1 The Basic Model Your first model will be a simple model that consists of a demand block two queues and an activity With this basic model you will learn most of what you need to know to model simple business processes Building models with Process Playground is done by dragging the building block icons from the Building Block section to the workspace Start by clicking on the demand icon and dragging into the workspace You can drag it anywhere you would like to but it makes the most sense to put it at the left side of the workspace since the Model Controls rinto demand block will be the secs ABB nac cn la E o o Je ST a N beginning of the process Faesje a ee 6 JI fases
17. at have a Qonty NENNEN balked Your Add Optional Parameters modelshoud SSES look like the Parameter 2 Distributon c9 figure below iiom Kone Note the Scroll Item lt lt lt highlighted Item items to make Model4 MSPPG sure you have Version 172 them right Model Data Once you have Figure 22 Balking Queue made all the required changes go ahead and run the new model You should see a graph similar to the one shown in the figure when you click on the first queue Notice how the queue length started out at 10 but then after the 4 5 time periods had passed the queue balked nine items from the queue and sent them to the bottom activity Once the initial on hand inventory had balked from the queue the process stabilized and eventually returned to just one piece in queue Due to the randomness of the non FIFO operation your chart may look a little different but it should be close to the one in the figure Copyright 2011 MoreSteam com LLC 29 Model 5 The Kitting Queue A very common function in any process is the collection of several items together into a single assembled package This could be the individual pieces of paperwork required for a loan application assembled into a single folder or individual pieces of a machine assembled into the final product the machine In Process Playground we accomplish this collection of items by building a kit of parts The kit of
18. average demand at 1 time period just as inthe Block Name Demando basic model Since the processing time of our activity is still 1 time period this makes our target average utilization 100 Add Demand Parameters What do you think will happen to our process when Start demand at ine C Mec a demand variation is added Do you think the process will Eia donid adsis continue to perform well with no queue forming in front of the Demand Interarrival Time activity block To add variation to the demand block first click on the demand _ Demand Time Distribuson block so that it has focus the flashing red line below it The Model Data area at the left side of the screen will display the data for the demand block Click on the Demand Time Distribution input field A popup box will appear Click the radio button next to the Uniform distribution Notice that the minimum and maximum fields are now available for data entry Set the minimum value to 0 5 and the maximum to 1 5 This will give us a range of inter arrival times between 0 5 and 1 5 with a mean of 1 0 After you have entered the minimum and maximum values click the green Save Distribution button to store the data Now run the model for the default 100 time periods item Name M What happened Does the model appear to be stable based on the time series plot of the queue in front of the activity Click on that queue and look at the time series plot Turn off the animation in th
19. ce to the activity block click on the activity block to give it focus then change the number of servers from 1 to 2 Does your time series plot look more in control now Does it look more like the picture below Copyright 2011 MoreSteam com LLC 20 Figure 10 Queue Length with Two Servers There is still some variation but the queue length is definitely much better controlled than the base case with just 1 server However the utilization of the two servers has probably fallen to around 5096 The queue problem is under control but the cost of providing the service has doubled What about making some improvements in the activity Try changing the number of servers back to 1 and reducing the mean service time from 1 0 to 0 75 a 25 reduction in service time Now what does the queue plot look like Something like the picture below vers Utilized Time Series Plot Figure 11 Queue Length with Process Improvements The queue length plot is looking fairly similar to the plot when we put an extra server into the activity But now we have accomplished the gain by improving the process and removing 25 of the service time rather than by adding the extra cost of a second server Utilization should be very close to 75 now This exercise is an example of how you can use simulation to evaluate different options for improving the performance of a process without having to implement all of the options in reality You can now virtually p
20. e flow of items in the model based on which activities can work on the different types of items To demonstrate this we will now add a separate path in the process for the Part B items to flow through Copyright 2011 MoreSteam com LLC 24 this first click on the queue in front of the activity so that it has focus the flashing red line beneath it Then drag and drop a new activity block below the current activity block in the model It should automatically connect to the queue block as long as you have the AutoLinks button turned on Once you have added the new activity block to the model then use the Manual Link button to add a link between the new activity and the exit queue To do this click the Manual Link button first and notice that it will turn green While it is green click on the new activity station Once you have clicked on the activity station the button will turn red indicating that you need to click on the destination for the link Now click on the exit queue and the Manual Link button will return to the normal state When you have completed the manual link between the new activity and the exit queue the last thing to do is to edit the Item Name in the new activity and change it from default to Part B You should end up with a model that looks like the figure below Note that the first activity block can only process Part A items and the second activity block Block Name Add Process Parameters Servers
21. e model and rerun the model several times How much does the time series plot for the queue change each time Chances are high that you will have something that looks fairly similar to the plot below Figure 7 Demand Variation h or Serv wu n ML M Figure 8 Queue Length with Uniform Distribution So it appears that the uniform distribution causes a short queue to form periodically but the process appears to stay stable and in control for the most part The uniform distribution is a fairly well behaved low variation distribution What if we change the distribution to an exponential distribution Click on the U 0 5 1 5 shown in the delay time distribution field and change it to an Exponential distribution with mean 1 Run it several times to see how much the time series plot varies Did you get a time series plot that looks more like this Copyright 2011 MoreSteam com LLC 19 Figure 9 Queue Length with Exponential Distribution Obviously the exponential distribution has a much higher amount variation when compared with the uniform distribution The result is a process that is much less stable This leads to a very important insight about the relationship of variation and utilization in a process as the variation present in a process increases the long term target utilization of the process must decrease in order to maintain a stable process In the figure above the queue length reached a maximum length of 13 items Thi
22. e replenishment pull models This video will take you through the entire process of building testing and analyzing the performance of a replenishment pull system This video will also introduce the use of the Demand block as a pull demand generator rather than the push demand generator we have seen so far Model 7 Complex Demand Patterns Process Playground has the ability to create complex demand patterns using the start and stop time settings for individual items in the demand block Using this feature almost any pattern of demand variation can be recreated in your models Model 8 Batch Processing Process Playground supports batching and un batching of items using the activity block Setup time can be applied to the batch while individual processing time is calculated for each item in the block This video will show you how to create and analyze batch processing models Model 9 The Core Process Pull Block Core Process pull is a fundamental Lean concept that is critical to almost every concept The Core Process pull block allows you to quickly build models which incorporate this type of pull system Model 10 The Resource Block The Resource block allows you to create a pool of resources to be shared amongst a group of activities The resource pool might represent people in a transactional model or tooling in a Copyright 2011 MoreSteam com LLC 32 manufacturing model The resource block also supports the start and sto
23. ese to values greater than one the activity block will attempt to pull the largest batch possible when it becomes available for work In order to operate in batch mode the activity block must have a queue block in front Copyright 2011 MoreSteam com LLC 12 to accumulate items The accumulated items will then be batched together and sent to the activity for work The activity block must also have a queue attached behind it downstream in the process to accept the batch when it is completed The downstream queue will break the batch apart and hold the items as individual pieces for the next activity The next activity may have its own rules for batch sizes e tem Name As with the demand and the queue blocks the activity block supports multiple items The default item named default tells the activity block that it can process any item passed to it If you change default to anything else the activity block will be limited to processing only those types of parts For example if you change default to Part A then the activity will only be able to process items named Part A More on this topic in a later section For now make sure to leave the item named default The Exit Queue All models will need a final queue attached to the last step of the model This queue will act as an exit to accumulate all of the finished items that have made it through the process Without an exit queue the last activity would have nowhere t
24. eset If you need to stop a simulation run just click this button to stop the run and reset the model e Debug Mode and Single Step These two buttons allow you to step your way through a model and see each event before it happens This is a fairly advanced topic that is covered later in this manual The default run length is 100 time periods and that will be fine for your first run Go ahead and click Run Model to start the simulation The default mode of operation is with animation turned on The animation button is just to the left of the debug button If it is white then animation is turned on If it is grey then animation is turned off With animation turned on you should see small black squares moving from block to block in the model Each black square represents one piece of work in process In this first basic model the WIP will flow right through the model and pile up in the exit queue When more than 50 pieces of WIP are present in a station no more black squares will be added and a black plus sign will appear If you turn off animation you will see that the model runs much more quickly Animation can be helpful for testing models and presentations but it takes a lot of computer resources and runs much more slowly After you have run the model your screen should look something like the picture below Make sure you have clicked on the exit queue to see the same graphs as shown in the picture Copyright 2011 MoreSteam com LL
25. hat will force the queue to operate as a FIFO queue 50 of the time and the other 50 of the time the queue will randomly pull an item from somewhere other than the front of the queue You can enter values between zero and one in this field If you enter zero the queue will never allow the first in line to leave the queue unless that item is the only item in the queue e tem Name Just as with the demand block the queue block allows multiple items to be tracked in the queue For now leave the item name as default More options for using the Item Name field will be covered in later sections The next element of your first model will be an activity block This is the block that actually represents work or a task being performed in your process Make sure that Auto Links is still turned on and that the queue block is selected the flashing red line should be below the queue block Now drag and drop the rectangular blue activity block into the model Place it somewhere to the right of your queue block Once you have created the activity block in your model you should see the Model Data section change to reflect the data options for the Activity block Your model should look like the next picture Copyright 2011 MoreSteam com LLC 11 PROCESS PLAYGROUND Model Controls rts SimTime Complete EUST LU Oper Print m Auto M Per Pull E EBE Ly Fj TA Modal Links Jaz a Coa f Time Series Station Dabiig Clear T Debug f
26. hey wait in queue If the items were people it would be reasonable to expect that after some amount of time waiting the people may choose to leave the line This behavior is called balking and Process Playground allows you to model balking at a specific time threshold you enter Another common use o Copyright 2011 MoreSteam com LLC 28 balking in queues is when the items in the queue are perishable If they pass a certain length of time in the queue they perish and are disposed of Click on the first queue in your model and note the field named Balk at Time directly below the Probability FIFO field If you enter a non zero time in this field the queue will balk any items which spend more than that amount of time in the queue For our experiment enter 4 5 as the balking time Items that balk from the queue are renamed from the original name and take on the name balked Click on the first queue so it has focus then drag a new activity into the model below the current activity Then drag a new queue to connect to the new activity Finally rename the default item default in the new activity to balked This will tell the first queue that the new activity will only accept items Block name Queuef nam ed Add Process Parameters balked thus FIFO Max Items 3999 Infial On Hand providing a new path in the Kitting Mode Optional process for any Kit ltem Name p items th
27. ilot any number of improvement concepts for your process fine tuning the results before you run your first live pilot For example how did we come up with the 25 reduction in service time in the example above In a simple process like this we could use some complicated mathematics or our years of experience Or we could build a table in Excel and run some experiments with Process Playground and quickly gain the insight we need to know how much we need to improve the process Copyright 2011 MoreSteam com LLC 21 Model 3 Adding Items and Controlling Flow by Item Name The basic model you built had only one type of item the default item in the model In the text under each block in the model you can see a list of parts or items For the basic model we left the single item for each block as the default named item Process Playground has a powerful feature which allows you to easily build complex models with many items flowing through the process in the model without needing to add any additional blocks to the model This is accomplished by adding additional items to the various blocks in the model Depending on the block the addition of new items will cause different things to happen To start we will make a few simple changes to the basic model and see what happens Open up the original basic model Model 2 from the previous section This is the basic model with variation To make sure we start in the same place make sure you
28. ive at the station e Queue Length or Servers Utilized Time Series Plot o This graph shows the length of the queue over the entire run of the simulation For the exit queue this graph will just show a line representing the accumulation of exits in the queue For a real queue with variation in the model this graph will look quite a bit more interesting This is the first place to spot whether or not your process is stable e System WIP metrics o Tothe right of the Time Series Plot you will see some text that is titled System WIP Metrics This data shows the minimum maximum and average WIP across the entire model except the exit queue as well as a count of total WIP generated by demand blocks the first number in the Total WIP line and the total WIP in the model including items created by kitting queues more on that topic later e Run Report o The Run Report button near the top left of the model will generate a report of the metrics for the queue and activity blocks in the model The text in this report can be selected and then copied and pasted into Excel or any other application for further analysis Click the Run Report button again or move your mouse over your model blocks to close the report window Copyright 2011 MoreSteam com LLC 16 Model 2 The Basic Model with Variation The next model will be a simple extension of the basic model with the addition of random variation The goal of simulation modeling is to see how a
29. n 2 minutes yields the same long term results as a Poisson process with 5 arrivals per 10 minutes In the Poisson process the random number of arrivals would all appear at the 10 minute mark while the exponential process will allow the arrivals to appear on average every two minutes e N The Normal Distribution The Normal distribution allows you to enter the mean and standard deviation of the traditional Normal distribution In a simulation you can never go back in time however the Normal distribution does have the potential for creating a random value which is negative If this happens the negative tail of the Normal distribution will be truncated to the value zero T The Triangular Distribution The triangular distribution is a popular random distribution when modeling business service processes The distribution closely matches real life performance of these processes based on several characteristics of the distribution In business processes there is typically a minimum value added time representing the shortest amount of time required to complete a task The task can never be completed faster than this minimum time so the minimum of the triangular distribution matches nicely with this true minimum In business processed there is often a maximum service time representing the point where an order has become so late that management intervenes and changes the process to get the order completed This management intervention point is nicel
30. n the model now You should see a very large out of control queue forming in front of the activity block We doubled the demand rate by adding the second item in the demand block but did nothing to increase the capacity of the activity block Multiple Items in the Queue Block Now that we have added two items to the demand block and doubled the demand in our model let s turn our attention to the next block in the Figure 14 Items in the Queue Block model the first queue Once you Copyright 2011 MoreSteam com LLC 23 have run the model you should be able to see in the text below the block icon that the current inventory in the model is quite high due to the model have more demand than capacity However when you ran the model you did not see any animation of the inventory levels above the queue block Why is this happening If you look at the Item Name field you will see that the default item is still present in the queue block And there really are zero default items in the queue because the demand block is only generating Part A and Part B items The text under the icon in the Model Workspace shows the entire inventory in a queue regardless of its type In order to see the animation of each type of item you have to add each of those items to the queue Go ahead and double click on default in the Item Name field and rename the item Part A Then click the New Item button and rename the NewPart item to Part B N
31. ng to the top activity as well or we need some sort of smart queue management which will not always use FIFO rules to determine how to distribute work Process Playground does not have the capability to model complex queue management rules like that Figure 19 New Utilization Numbers so let s go ahead and add Part B to the top activity so that activity will be cross trained as well As we did in the bottom activity let s have Part B in the top activity have a mean service time of 1 25 and Part A in the top activity have a mean service time of 0 75 The utilizations between the stations should now be fairly equal as both stations are able to work on both items Process Playground automatically manages the flow of items between the stations based on the queue of items waiting and when the activities become available Model 4 The Balking Queue and the Non FIFO Queue We will use Model 1 as the basis for Model 4 learning several new options in the queue block as we transform the model To start open up your copy of Model 1 Recall that this model should be a simple process which consists of a demand block flowing into a queue flowing into an activity and then finally ending in an exit queue Both the demand and activity blocks should have C constant distributions and mean arrival and service times of 1 Model 1 has no variation and is a perfectly balanced model If you run the model you will see in the exit queue that the TCT mean
32. o send completed items and the activity would quickly become blocked stopping the i Model Controls ann a PROCESS PLAYGROUND flow of items to the W a activity m ES Go ahead and add an Block Name Queue3 exit queue to your Add Process Parameters model Be sure that pm the activity block is eee selected with the Kitting Mode Optional flashing red line below pum it before you drag in Add Optional Parameters the exit queue If you NES Parameter 1 Distribuson o1 make a mistake and Parameter 2 Distribution c the exit queue is Item Name default attached to another Figure 5 The Exit Queue of the Basic Model block in the model jump to the next section to learn all about managing links in your model Then come back here to learn how to run the model Your model should look like the picture to the right Running the Model The time has come to run your first model in Process Playground Running the model is simple and there are only a few things you need to consider before the first run The first thing you need to consider is how long to run the model Are you modeling a restaurant If so Copyright 2011 MoreSteam com LLC 13 may need to run the model for the length of the work day Or are you just modeling a peak demand period for a call center If so then define your
33. oce ode acne vehe tona dee etn bon dun acid aaea oe ae Roo dd 25 Figure 18 Adding Part A e ote ede e deuote tacere nace eramus EES 26 Figure 19 New Utilization Numbers oit ohio heec ch un eec eoa rea n eoe eae Eod 27 Figure 20 FET with Queue OF TO nre teet e e eet ees geste edt rae eue Pre be eee dO eects 28 Fig re 21 ICT with Non FIEO QUeUe eere erect reete e OE bere et vete eee dee uU E aea nuce eeu 28 Fig re 22 Balking Queue otro ette eri terme eese e tea veter tecti ret re end 29 Figure 23 AITASUIIIME HH 30 Figure LEA ASUIIMEMPR 31 Fig r 25 QUEUE 3 rtainn rann ae de sales asdvansedes Tadvecesvasescds sale scesssbadeds A EEEE AEE EE lace Ss 31 Figure 26 Coffee Brewing sc cree cha cove cis mte t hem tentat ibunt E E 31 Copyright 2011 MoreSteam com LLC 4 Introduction Welcome to MoreSteam com s Process Playground Process Playground PP is a web based discrete event simulation program that allows you to create virtual models of your processes Once you have built a model of a process there are many avenues of exploration available e What if the demand level increases or decreases can my process handle the change e What if variation in demand or service time changes what will happen e need to improve the lead time of the process what should change first These and man
34. ocks provide all of the modeling functions available in Process Playground e Model Data The left side of the screen holds data related to the currently selected block You can tell which block is selected because it will have a red flashing bar below the icon Queue 1 in the figure Each building block will have different types of data displayed in the Model Data section e Model Workspace The last section is the main workspace for Process Playground Building Blocks Model ipsc Run Lang 77108 PROCESS PLAYGROUND on save g m m Model z 4 Je Li Model J 7 A z 1 SEN 5 Do Delete Delete stopa Aout u Demand lt p pert pa epo inks bown Links Reset Pace see Rey Block Name Queue A Add Process Parameters FIFO Max Items 9999 Hi Inal On Hand 0 Probability FIFO 0 0 1 0 1 Balk At This Time never Kitting Mode Optional Kt ltem Name Kit Quan amp y 1 Add Optional Parameters Cost Distribuson c0 Parameter 1 Distribuson c Parameter 2 DistribuSon cQ Item Name default Version 1 7 2 Model Data Model Workspace Figure 1 Process Playground This is where you will drag and drop building blocks to create your process model In order to help you learn about Process Playground
35. of the Basic Model esses nennen enne nnn nenne entra sanas sessi setis dass asser saa 15 Figure 8 Queue Length with Uniform Distribution eeessessseseees eene eene enne nnne 19 Figure 7 Demarnd VariatiOD etie tere rese e peser E tud verses rete er voe eek e Ue ende re er erede 19 Figure 9 Queue Length with Exponential Distribution eeeeseeseseeee nennen ennemi nnn nennen 20 Figure 10 Queue Length with Two Servers cccssssccecccessssesesaeceseceseesesaeseceeecesscesaeseeeescesseseaaeseeeeesesseaeaeeeeeessesaes 21 Figure 11 Queue Length with Process Improvements ccccccccccssssssssceceeecesseseneeeceeeceeseseeaeseeeesseeseeeaeeeeeeseeeees 21 Figure 12 Model 2 Starting Point cater tee ced ees cevb bode ee Ce Ere o eaa ge ee eee do ra c de Te ERE Fe sauceaceceeivedsaate 22 Figure 13 The Default Item cene eec ert eec ere e tear ed exe o uec e ed Ere ee avons Pee ae weeded eae 23 Figure 14 Items in the Queue Block ssessssseesseeeeseeee nennen enne nnn nnne neni nnns ases ense ti asas nasse enne an 23 Figure 15 Queue Block Current ON Hand cccccssccccccecessessesececccesceseeaeseceeeceseeseaaeaeeeeecesecseaaeeeceesssssesaaeeeeeeseseees 24 Figure 16 Part B Stuck In QUO UEC cccsscccccceeessesssaeceeececeesesauaeceeecuseeseaaeseceeecusseseaaeaeeeescesseseaueseeeesseeseaaaeeeeeeeesesees 24 Figure 17 Model 3 with TWO Items uere n
36. ou change another parameter discussed later in this list e Initial On Hand You can easily start your process with existing Work in Process WIP by placing initial on hand inventory into the queues in your process Copyright 2011 MoreSteam com LLC 10 e The next three data items all starting with Repl Pull are used when the queue is converted to a replenishment pull system buffer This topic will be covered in a later section e Current On Hand You cannot edit this field but during and after a simulation run you can review the current on hand value of each unique item in the queue by checking this number e Current On Order This field is also related to the replenishment pull system operation and will allow you to watch the amount of inventory on order change during a simulation run e Kit Quantity This field is used when the queue is transformed into a kitting queue As a kitting queue the queue will pull a mixture of items into the queue and then output the items as a single assembly or new item This will be covered in more detail in later sections e Kit Item Name A text entry in this field transforms the queue to a kitting queue This topic will be covered in its own section later e Probability FIFO This field allows you to disrupt the normal First In First Out operation of the queue The default value is 1 or 100 which indicates that the queue will operate as a FIFO queue 100 of the time If you enter a 0 5 t
37. ow rerun the model with the animation turned on the animation button is white not grey and watch the animation of the two items You should now see unique colors and animated stacks s of parts in the queue for each item Once an item Figure 15 Queue Block Current On Hand has been created with a unique ltem Name that name will remain with it throughout the simulation Multiple Items in the Activity Block The next block in our model is the activity block So far the activity block has processed all of the items coming to it regardless of the name given to the item Whenever the activity block has an item with the name default the activity block will be able to process any item that comes to it However if you change the default item to something else you can adjust the processing times and capabilities based on the item name To see how this works click on the activity block and in the Model Data section double click on the Item Name default and change it to Part A Now rerun the model What happened If you look in the first queue you should see a large stack of Part B items the green animation items in the picture to the right By changing the default item in the activity to Part A we told the activity that it could only work on Part A items The Part B items had nowhere to go so they just stopped and remained in the queue Figure 16 Part B Stuck In Queue Process Playground automatically manages th
38. p time features of the demand block allowing you to create complicated models of employee shifts Model 11 The Decision Block The Decision block is a simple block that offers a random redirection of items in the model based on the probabilities you enter The decision block is most often used to create random rework and scrap loops in models Model 12 The Note Block The final building block available in Process Playground is the Note block This block is not functional but allows you to add notations directly in the model and connect those notations to the functional blocks of the model Copyright 2011 MoreSteam com LLC 33
39. parts then takes on a new name as a new item in the model To see how this works we will build a simple process of brewing coffee Start with a clean sheet in Process Playground and drag a demand block to the top left corner of the workspace then drag a queue into the workspace It should automatically connect to the demand block if you have AutoLinks turned on Next while the new queue still has focus drag another queue into the model to connect to the first queue Rename the second queue KIT QUEUE You should have a model that looks like this Figure 23 Kitting Queue The next step is to turn off the AutoLinks function by clicking the AutoLinks button at the top of Process Playground It will turn grey to indicate that the function is off Now drag a new queue into the workspace and place it below the first queue Queue1 Now you will need to use the Manual Links button to create manual link between the new queue you just dragged into the model and the KIT QUEUE Once you have completed these steps turn AutoLinks back on and then click the KIT QUEUE block so that it has focus Then drag an activity block into the model It should automatically link to the KIT QUEUE block Finally drag another queue block into the model to the right of the activity block this will be the exit queue When you are finished your model should look like this Copyright 2011 MoreSteam com LLC 30 Figure 24 Coffee Brewing TCT mean NaN dJorzzc
40. r and the parameter K 4 The Erlang distribution with K 4 is created by summing four random variables generated from an exponential distribution with the mean equal to 1 4 of the mean you have entered The benefit of this distribution is that it closely mimics a log normal distribution resulting in a distribution of random variables which are positive with a single mode slightly less than the mean and a right skewed tail E The Exponential distribution The exponential distribution requires only the mean value to be entered as the variation of the exponential distribution is equal to the mean The exponential distribution is highly variable and often used in modeling the arrival and service times of processes with very high transaction traffic such as telephone systems o Poisson arrivals to a process are a very common modeling method and the exponential distribution can be used to generate Poisson arrivals to a process For a Poisson process demand is defined as a random number of arrivals during a fixed period of time The distribution used to generate the inter arrival time of those arrivals during the fixed period of time is the exponential distribution If Copyright 2011 MoreSteam com LLC 17 Poisson process has an average of 5 arrivals during a 10 minute time period then the mean inter arrival time between each of the arrivals is simply 1 5 of the 10 minute time period or 2 minutes So an exponential distribution with mea
41. r copy of Model 2 has the demand distribution type as E for exponential mean inter arrival time of 1 and the service time in the activity station set as C for constant with mean service time of 0 75 You model should look like the figure below Model Controls teen 8 mre corpi PROCESS PLAYGROUND I E S inks Link Model J f elete Sto m Lini Rese Building Blocks Aee eow Block Name Queue Add Process Parameters FIFO Max Items 9999 Infjal On Hand 0 Probability FIFO 0 0 1 0 1 Baik At This Time never Kitting Mode Optional Kt Item Name Kit Quantity 1 Add Optional Parameters Cost Distribuson Ci Parameter 1 Distribuson c0 Parameter 2 Distribuson Cio Item Name default sew tom Ram j NewModel MSPPG Version 1 7 2 Model Data Figure 12 Model 2 Starting Point Model Workspace Copyright 2011 MoreSteam com LLC 22 Note that each block in the model has a list of parts associated with the block The first block we will work with is the demand block Start by selecting the demand block Now take a look at the lower portion of the Model Data section Do you see the Item Name field and the name default for the item Block Name Queue7 name Your screen should look like the figure at the right 44 Process Parameters FIFO
42. ropriate character for each of the distribution types in the space provided in the data section As you make your selection the appropriate numerical parameter fields will become available for you to enter the data needed by the distribution The default type is C Copyright 2011 MoreSteam com LLC 8 representing a constant More information on the available distributions can be found in the later section Random Distributions e Start and End time With the demand block you can define when the block will begin to produce demand and when it will stop producing demand The default values of zero and 1 indicate that the demand will generated for the entire simulation run If you change the end time of 1 to any positive value that value will define the time when the demand block will stop generating demand during the simulation run The start time defines when the demand block will begin generating demand If you set the start time to a value higher than the end time the block will generate 1 demand event at the start time you entered and then stop producing any demand e tem Name The default item name is default A powerful feature of Process Playground is the ability to create many items within each block These items can have the same name or different names and Process Playground will automatically route the items in the model based on their names For now just leave the Item Name as default You will learn more about this powerful
43. run length to match the length of the period you wish to model The length of the run is determined by the number you enter in the Run Length field at the top of the Model Controls section There is no specification of the time period in Process Playground That is left for you to interpret for yourself Just make sure you are consistent If you decide that each time period should represent 1 minute then make sure all of your demand and service times in each demand and activity block are entered in minutes You can define the time period to be anything you want a second a minute an hour a day whatever you want Just be consistent and use that time period throughout your entire model The buttons at the right side of the Model Controls are related to running the model The functions are pretty straight forward e Run Model This button starts the simulation run This button also acts as the pause button during the run The color of the button will tell you the state of the model o White the simulation is not running and ready to be started o Green the simulation is running and you should see the simulation time value in the Sim Time box just above the buttons growing as the model runs o Red the simulation is paused Once you start the simulation you may pause it by click the start button again The button will change from green to red Click it again to restart the run and the button will change back to green e Stop amp R
44. s Delete S Export Report Report Model t Mode lt I be Links own Links Block Name Add Process Parameters Denman Count 10200 Parts deaur Add Optional Parameters Cost Distribuson Parameter 1 Distribuson Parameter 2 Distribuson Item Name default Figure 4 Activity Block of the Basic Model Activity Block Data The activity block data consists of several different pieces of data which define how the queue will behave Some of the options will be covered in later sections but here are the basics e Block Name As with the other blocks you can rename the activity block as desired e Distribution Parameters The selection of distribution type and the distribution parameters are identical to the selections in the demand block Once you select a distribution type the required parameter fields will become available for you to enter your data e Servers Available Each activity can operate as a single server person or machine or as multiple servers For instance if you have 3 ATM machines at a bank you can model them as a single activity with 3 servers e Setup Time Each individual item entering the activity will incur a service delay time based on the distribution selected However each batch including a batch of 1 will also incur a setup time if you enter a value here e Minimum and Maximum Batch Size the default values for the minimum and maximum batch size are one If you change th
45. s is based on a process that has a long term target utilization of 10096 For the run above the actual utilization was 93 How could that be lower utilization AND longer queue length at the same time The answer is in the chart Early in the run the queue length is often at zero During this time due to the randomness of demand arrivals there were times when the activity had nothing to work on As a result the utilization of the activity went to zero for a period of time However the long term average demand rate was still 1 0 so later in the run the demands began to arrive a little more quickly and the steady processing rate of 1 0 could not keep up The result is less utilization than targeted and longer queues two bad results wrapped into one simulation run So what is the solution for the long queues that formed In the long term gaining some control over the demand variation is the best solution Perhaps not all of that variation in demand rate is coming from the ultimate customer but rather from our own internal processes In the short term the only solution is to lower the target utilization rate This can be accomplished in two ways 1 Add more resources to the activity block this is like adding more people or overtime 2 Reduce the time required to perform the activity this is like removing some non value added work or streamlining the work being performed Experiment with your model a little To add another resour
46. the item list Change the mean inter arrival time for Part A from 1 0 to 0 75 time periods Now scroll to the Part B item and change the mean inter arrival time for Part B from 1 0 to 1 5 While demand has shifted from Part B to Part A the overall demand for both parts together has remained constant Go ahead and rerun the model now what do you see You should see some Part A items in queue waiting to be processed and no Part B items Also you should see that the top activity block the one which can only process Part A items is 100 utilized while the bottom activity which processes only Part B items is probably between 60 70 utilized Block Name The shift in demand from Part B to Part A has caused an imbalance Pea aa in our process something that we see happening in our real Servers Available processes every day A common solution is cross training so that more of our resources can perform more of the tasks needed However how much cross training and which resources Simulation Processing Time Dist modeling can help answer that question for us In the case of our Maximum Batch Size Setup Time Add Optional Parameters simple model here we don t have too many choices to make Let s add some cross training to the bottom activity Parameter 1 Distribuson c Parameter 2 Distribu on C 0 To do this click on the bottom activity and in the Model Data section click on the New Item button to add a new item
47. to this activity Rename the new item from NewPart to Part A This creates the ability for the bottom activity block to work on Part A in addition to the vane RR existing Part B With cross training it is very common for the cross z trained resources to be a little less productive on their alternative work as opposed to their main work We can reflect this in our model by changing the processing time for the new Part A item in the bottom activity Change the mean service time from 1 0 to 1 25 for Part A to reflect this lower productivity Figure 18 Adding Part A Now rerun the model with the changes to the lower activity in place What happened this time Did you get utilization numbers and queue lengths similar to those shown in the next figure Copyright 2011 MoreSteam com LLC 26 You probably experienced a bit of a flip in the utilization numbers with the top activity utilization changing from 100 to the 70 range while the bottom activity utilization went up Why did this happen The bottom activity is now taking both Part A and Part B based on the FIFO order of the queue This means that at some times the bottom activity began work on a Part A item leaving only Part B items in the queue The top activity then had nothing to work on since it was not cross trained to work on Part A items So cross training for the bottom activity helped but our experiment has shown that we either need to provide cross traini
48. value is 1 00 and looking at the Total Cycle Time Histogram you will also see that all items arrived at the exit in exactly 1 time period The first change we will make to the model is to add some initial on hand inventory into the first queue Since our process is perfectly balanced this inventory will simply add a permanent amount of work in process in the queue which will be worked on in a first in first out basis Click on the first queue and enter the value 10 in the Initial On Hand field Now rerun the Copyright 2011 MoreSteam com LLC 27 model When the run is complete click on the exit Total Cjcle Time Histogram queue so you can see the Total Cycle Time Histogram graph It should look like the figure to the right Also notice that the TCT mean value for the exit queue is now above 10 This reflects the 10 items in queue that will require 10 time periods to process lengthening the total cycle time through the process Now you know Figure 20 TCT with Queue of 10 how to introduce work in process into a simulation from the start using the Initial On Hand field Non FIFO Queue Behavior The next feature of the queue block that we will experiment with concerns the behavior of the queue as a FIFO queue In reality many queues do not maintain a strict first in first out behavior There are slips and skips that allow random items in the queue to jump the line and get processed first Process Playground allows you to model this non
49. y modeled by the maximum parameter of the triangular distribution Finally many business processes are best able to report the mode rather than the mean service time of the process The mode is directly entered into the triangular distribution or it can be easily calculated from the minimum maximum and mean Choosing the correct random distribution for your process model is very important There are many statistical software packages available which will help identify the appropriate mathematical distribution for data you have However remember that the data you have is a sample of the underlying population and spending large amounts of time to come up with complex array of mathematical equations and distributions to represent that data is rarely a valuable exercise for a business process Your process model is an abstraction of reality to begin with and probably does not demand super precise input data to generate the insight you need from the model Experiments with the Basic Model Now let s make a few changes to the basic model to see how the model reacts to different forms of variation Because we will be using randomly generated variation for the rest of this manual your simulation results will never exactly match the results in this manual But they should be close and our average values should be very close Copyright 2011 MoreSteam com LLC 18 For the first experiment we will add some demand variation ST We will keep the
50. y more questions can be investigated and answered with a simulation model of your process But what if you re starting from scratch and designing a brand new process Creating a simulation model of your new process or of several alternative new processes will allow you to perform an infinite variety of virtual pilot runs before you even begin committing resources to the new process Simulation allows you to test the new process against large ranges of demand and variation and experiment with various failure modes before they happen Imagine a Failure Mode and Effects Analysis that is 90 complete before the first live pilot With Process Playground we have developed an easy to learn easy to use and easy to install you don t have to install anything web based simulator Most traditional simulation programs take a boil the ocean approach They have added a lot of complexity that allows you to model almost every nuance of any process The result is often a long learning curve before you can build the models you need We have taken a simpler approach Process Playground We analyzed the tools needed to model almost all of the typical processes which might be the subject of a continuous improvement project We included those tools and nothing more into Process Playground There are no hidden menus or dialog boxes everything you can do is right there on the screen Our students usually need little more than a day to learn everything about Process
Download Pdf Manuals
Related Search