Simulation with ARENA - SM Paints

50 %
50 %
Information about Simulation with ARENA - SM Paints
Technology

Published on March 9, 2014

Author: hrishik26

Source: slideshare.net

Description

Simulation of SM Paints production facility using ARENA simulation software. Making improvements using OptQuest software, and data analysis of current state simulation, to suggest recommendations for achieving desired level of productivity.

The following document describes the solution of the case of study SM Paints, presented in the Tenth Annual Contest of IIE/RA FEBRUARY 2013 SM Paint Production: Modeling Project System Simulation HRISHIKESH KHAMKAR OSMER DUBIQUE MERCEDES MICHELLE SLIFKA

SM Paint Production: Modeling Project Contents Introduction .................................................................................................................................................. 2 Problem Definition ........................................................................................................................................ 2 Scope and objectives .................................................................................................................................... 3 Model ............................................................................................................................................................ 3 Verification .................................................................................................................................................... 5 Validation ...................................................................................................................................................... 6 Animation...................................................................................................................................................... 6 Statistical Output Analysis & Recommendations ......................................................................................... 7 Initial Problem ........................................................................................................................................... 7 SCENARIO 1 ................................................................................................................................................... 8 Conclusion of Scenario 1 ......................................................................................................................... 10 SCENARIO 2 ................................................................................................................................................. 10 SCENARIO 3 ................................................................................................................................................. 11 Recommendations ...................................................................................................................................... 12 Bibliography ................................................................................................................................................ 12 Appendix ..................................................................................................................................................... 12 Appendix 1 .............................................................................................................................................. 12 Appendix 2 .............................................................................................................................................. 12 Page 1

SM Paint Production: Modeling Project Introduction SM paints is a paint manufacturing company which produces a line of paints, which comprises different batch sizes. Also, these orders are packed together using appropriate combinations of different packaging types (quarts, gallons, buckets). The whole facility is designed to perform manufacturing and packaging of these paint orders. After packaging, these orders are sent to the various distribution centers from where they are further transported to the company’s retail shops. The company feels the need to improve its production process to attain three main objectives,  Reduce holding costs (by reducing time spent by completed orders in the distribution centers)  Complete 98% of incoming orders within three days of order receipt.  Make room for extra orders, to generate more revenue ( after getting current system under control) These three objectives can be achieved if the production system is fine-tuned to remove any bottlenecks, present currently, and then make maximum possible utilization of various resources available. By doing this, it can be ensured that the lead time is kept under control ( even when the facility is overloaded while still using the current system) and thus, the orders could be delivered to the distribution centers as quickly as possible, eliminating the need to keep large stocks of inventories in the distribution centers. This will reduce the holding cost of the finished goods. (Assumption:- consumption rate of the orders is less than the lead time of manufacturing the incoming orders). Problem Definition As per the current situation of the manufacturing facility, SM paints is performing inconsistently when it comes to fulfilling customer orders within 3 days of order receipts. This is basically because the line of production has an improper allocation of resources to various types of batch sizes of the paint orders. Furthermore, the policy of SM paints puts restrictions on which type of processing tank is suitable for a particular size of an incoming batch. Because of this, incoming orders with more percentage of a particular size, lead to over loading of particular machines in the production facility, while other machines remain underutilized. This inconsistency leads to increase in the time spent by the orders in the manufacturing facility. To eradicate this problem we need to make changes to the current system which will ensure that 98% of the times incoming orders leave the system, after completion, within a span of three days. Page 2

SM Paint Production: Modeling Project Scope and objectives The main shortcoming of the current system is the interconnection and dependency of various processing stations on each other. For example, the Mix & Grind (M&G) machine stays occupied even after completing the process. This is because, the content of the M&G tank should wait for a thinning tank of appropriate capacity to be available (since each batch size of the incoming order has restriction on the range of thinning tanks to which it can go, depending upon the size of the thinning tanks). Similarly, the thinning tank should wait for a holding tank to be available before it makes the transfer. As a result of which if there is a queue at the holding tank of a particular capacity, then the thinning tanks of that capacity will also remain occupied, until the holding tanks are freed up. As a chain effect, if by chance, all M&G machines are seized by the batches of the same size which are causing queues in holding and thinning, then all the incoming orders (irrespective of their batch sizes) will remain in queue until the M&G machines are freed up. While at the same time the resources allocated to perform processes of other batch sizes (apart from the one causing queue at M&G station) will remain unused or underutilized. We can solve this problem by improving the network flow of various batches, by allotting tanks of higher capacities to incoming orders of batch sizes smaller than the tank capacities. This can be done for tanks at Thinning station and Holding station as well. This will result in improved utilization of available resources and will shorten the queue length at all stations considerably. Along with this, we can also increase the number of resources at various bottleneck stations, thereby keeping every stage of the production line in a flow. The main objective, as far as improving the production system is concerned, would be to decrease the time spent by the incoming orders in the system (by reducing their waiting times in queues and utilizing all resources). SM paints wants 98% of the incoming orders to be completed within 3 days after its receipt. Guaranteeing this performance level would ensure quick delivery of reinforcements (orders) to the distribution centers, thus serving the lean initiative promoted by SM paints. Model Fig.1. Schematic of the process The given situation of SM paints has multiple levels of categorization of entities. To ensure that this real life situation is accounted for in the simulation model we use attributes for every level of categorization. As the orders arrive to the system, they are classified according to the batch size: small, medium, or large. Each batch size has more than one category of sizes (unit = gallons) and each size Page 3

SM Paint Production: Modeling Project has a particular % in the total order of that particular batch size. Using a decide module the incoming orders are divided into their batch size. The paint manufacturing process begins with the Mix and Grind process. At this step in the system, there are seven mix and grind machines working. However, all the Mix & Grind machines have just 16 hour shifts. The mix and grind machine perform the process, however they are not freed up unless and until the batch inside the M&G machine is transferred to the thinning tank of appropriate batch size. There are nine thin tanks varying in different capacities. After the delay of the mix and grind machines, there are decide modules in order to separate the batches according to the size. This is done to facilitate the transfer of appropriate batch size to the allotted size of thinning tank. (Reference: Simulation with Arena) The first decide module transfers the liquid to the thin tanks with a capacity of 6,000-10,000. If the batch size is larger than 10,000 gallons the liquid is transferred to thin tanks with a capacity of 10,000-14,000. This process is repeated for thin tanks with capacities of 14,000- 20,000 and 20,000 gallons. In the transfer of liquids to the thin tanks, the thin tank resources overlap in their capacities. For example, 1000-6000 and 8000-10,000 size batches can both use the 10,000 capacity tanks. At this step of the process we created sets of thin tank resources in order to allocate the batch sizes to the various thin thanks. Since the seized M&G machine won’t be free until the appropriate thinning tank is available, we have to use an overlap of Seize, Delay and Release modules in the following steps, 1. 2. 3. 4. * Seize the M&G machine * Delay for M&G process * Seize the Thinning tank * Transfer the batch to the thinning tank(Delay) * Release the M&G machine * Delay for the Thinning process Once the thin tanks are seized the transfer is complete. After this transfer the mix and grind machines can then be released. It should be noted that the queues are at the seizing of the thin tanks and not at the actual thinning delay process. This is also due to the transfer process and having to have the thin tanks of the appropriate size being seized first. After the thinning process, the batch is to be transferred to the holding station. First there are two decide modules in order to separate the batch sizes before they are sent to the ten various holding tanks. The batches are held in the thinning tanks till the holding tank of appropriate size is available. The process for the transfer of the batch from thinning to holding tank is quite similar to the transfer from M&G to the thinning tank. Similar to the thinning tanks, the batches are held in the hold tanks until the required fill lines become available. Again we used sets of hold tanks to differentiate between the capacities. The sets are used because the hold tank resources overlap. For example, the batch sizes of 1000 to 8000 and 8000 to 12000 can both use holding tanks of capacity 12000. At the seizing of the hold tanks there is a queue because of the transfer process. A delay is then used for the transfer rate of the gallons Page 4

SM Paint Production: Modeling Project into the hold tanks. After the delay, decide modules are used again in order to release the hold tanks. The hold tanks cannot be released until the required fill lines become available. The hold tanks need not wait for all the required filling lines (quarts, gallons, buckets) to be available. If it requires quarts, it can send that part of the batch (quantity of liquid) which requires to be filled by quarts to the quarts filling line. Similarly, the orders for gallons and buckets can be made available when they are free. To assign proportions of the order batch size to combination of quarts, gallons and buckets, we use decide modules (to assign “condition true” percentage) and then an assign module to assign a value of quarts/gallons/buckets according to the given values of % by volume. At the filling station, we have used a variable “XX” which is a counter and its value increments every time a batch (entity) passes through the Assign module. This value “XX” is assigned to an attribute “No Order” which indicates the order number. This order number is used to group together quarts/buckets/gallons of the same order together after they are completely filled up. Another attribute “Batch No” is assigned to each incoming entity and it indicates the number of filling components required by that particular order. For example an order requiring all three filling types would have a Batch No = 3, an order requiring just Quarts and Gallons or Buckets and Gallons would have Batch No = 2 and an order requiring just Gallons would have a Batch No = 1. (Note:According to given data no order will require just Quarts and Buckets, because the summation of even the maximum values of Buckets and Quarts possible is equal to 60% of the entire order, thus indicating that Gallons will always be a part of all incoming orders). The batches are first diverted according to the Batch No. and then each batch number is separated into Quarts/Gallons/Buckets, and sent to the appropriate filling line, thereby making it possible to fill orders partially depending upon the availability of the filling lines. When all portions of an order (Quarts/Gallons/Buckets) are filled and ready, then they are batched together and sent to the Dispose module to end the system. (Reference Simulation with Arena) Verification To verify that the logic used in creating the model represents the actual situations given in the data, we ran a couple of trials.  We created single entity for each combination of batch size and each capacity of the three types of batch sizes. This was done by assigning specific values to the incoming order and then tracking its path through the system by running the animation at a slow speed. This ensured us that the logic defined in the system is correct and the entities follow the path they are supposed to, depending upon the attribute values assigned to them.  Also, we created errors in the system which will ensure us that in case of multiple entity arrival (like a real-life situation) the entities will behave the way they are supposed to be. This was primarily done by using one “Create” module to create one specific type of entity (let’s call it Entity 1) for a considerable interval of time. This will result in that specific type of entity seizing Page 5

SM Paint Production: Modeling Project all M&G tanks thus leading to formation of a “Queue”. Using another create module, we created those entities (let’s call it Entity 2) which would use resources at thinning and holding stations, which overlap the resources used by the previously created entities. This was followed by use of another “Create” module, which would create another Entity which would use resources overlapping with resources of Entity 2. This helped us to simulate a situation where queues are formed at various “Seize” stations thus leading to help us understand how some resources are underutilized while others have queues. Validation The model designed has accounted for various real life situations arising in SM Paints by using Modules from the Arena software.  The model accounts for arrival of orders from a random distribution, thus giving it an element of randomness.  All the resources are given operating conditions which represent the actual situation in the production line (like, cleaning times, schedules of operators etc.)  Movements of the resources, depending upon the restrictions of the capacities of various tanks (at holding and thinning stations) which they can use are accounted for.  Along with this the filling lines account for the time taken by transfer of the liquids from the holding tanks to the filling lines and thus to the packaging containers (like gallons, buckets or quarts) The model takes into account all the possibilities that can exist within the system. As a result of which, even though the simulated values may not be same as (equal) the values we get in the actual facility, they would be much closer to the actual operating plant. Hence, we prefer to generate Confidence Intervals on certain chosen parameters of the simulated model, so that our suggestions with regards to those parameters would be more or accurate. Animation The animation presents the given data in a pictorial format, as a result of which we can understand the behavior of the system, overall. Animation of various processes helps us to understand when the resource is idle and when it is busy. Also, we can see when the system breaks down do to failure or when it is scheduled for cleaning shifts or breaks. Watching the animation of the overall system helps us to understand the behavior of various input parameters on the output of the system or behavior of entities in various queues. Also, if we want to check the flow of every single entity through the system (for trial checks) we can use animation to track the flow of single entities (this can ensure the validity of our models logic as well). Page 6

SM Paint Production: Modeling Project By watching the whole animation we can point out the bottlenecks in the system after running the simulation for certain duration. This helps us to understand which regions of our model need to be worked upon to improve the output. Using graphical counters in the model even a layman can understand the mathematical behavior of the simulated model. Statistical Output Analysis & Recommendations Initial Problem After running the simulation for the given operating conditions, we ran the model to find the warm-up period for the average time in the system for an entity. However, we realized that the graph goes on continuously increasing. This indicates that the given current system never stabilizes. This was basically happening because of the scheduling of the Mix& Grind station. Since, it had a schedule of only 16 hours/day, hence for the rest 8 hours of the day incoming orders would accumulate and just stay in queue at the M&G process. However, the rest of the system would be up and running, but it would remain underutilized due to no input of new orders. After running trials on the given system, we realized that the utilization of resources at M&G is 95% while the rest of the resources at other stations had utilizations in the range of 20-35%. This problem occurs because of the absence of resources for 8 hours at the M&G station. As a result of which when the resources are available again, they have to first clear out the queues remaining due to their 8 hour break. Hence, the overall performance of the system is very poor, and about 60% of incoming orders fail to leave the system as completed orders (Refer Appendix A, Initial sheet). To make the system non-terminating we included an 8 hour shift at the M&G station. As a result the system became continuous and evened out the utilization of resources across most of the stations. The % of orders not completed within 3 days of receipt of order decreased down to 26.09 (Refer Appendix A, just adding one shift sheet). This indicates that the warm up period needs to be found out for that model which has a possibility of getting a stable graph, rather than a continuously increasing graph for the original system. The figure 1 shows the warm up period after adding one shift to the existing system. It comes out to be approximately 150 hours. Page 7

SM Paint Production: Modeling Project Fig.2. Warm-up period of the current system with 1 additional shift Trials to run the system for satisfying 98% of incoming orders with in three days of order receipt:We ran the trials for various ranges of the available resources, to find the minimum cost and minimum number of resources. The main aim of the trials was to find the proper combination of resources to complete 98% of incoming orders within 3 days of receipt of the orders. The statistics were found by collecting the time spent in the system by various entities using Read Write Module. Some of the trials and their results are listed below. SCENARIO 1 Actual system (with scheduling):- In this case the team chooses to add just 1 shift to the Min and Grind Machine. The result in this case for the performance of filling level is, 73.91% of all orders. The replication length of this run is 72 days and the number of replicates is 30 Cost of the policy Sources of Cost per Number cost day days $ Extra Shift 72 800.00 Total Cost of Fixed Cost 0 Total $ 57,600.00 $ 57,600.00 Table 1. Cost of the Actual system with scheduling Page 8

SM Paint Production: Modeling Project Trial 1 :- In this case the team choose to add 1 shift to the Min and Grind Machine, 1 line for the Quarts and 2 Lines for the Buckets. The result in this case for the performance of filling level is, 99.97% of all orders. The replication length of this run is 72 days and the number of replicates is 30. This configuration has less error that the configuration suggested by the simulation team, but the cost of the policy is higher than the one chosen by the simulation team. (Refer Appendix 1, 1 shift 1 QL 2 BL) Cost of the policy Sources of Cost per Number cost day Quantity days $ Extra Shift 800.00 1 72 $ Quarter Line 1 72 $ Bucket Kine 2 72 of Fixed Cost $ $ 1,200,000.00 $ 400,000.00 Total Cost Total $ 57,600.00 $ 1,200,000.00 $ 800,000.00 $ 2,057,600.00 Table2. Cost of the Trial 1 Trial 2:- In this case the team choose to add 1 shift to the Min and Grind Machine, 1 line for the Quarts and 1 Lines for the Buckets. The result in this case for the performance of filling level is, 99.58% of all orders. The replication length of this run is 72 days and the number of replicates is 30. This one has less error than the previous policy because it takes fewer entities to the system. (Refer appendix1, 1 Shift, 1QL, 1 BL) Cost of the policy Sources of Cost per Number cost day Quantity days $ Extra Shift 800.00 1 72 $ Quarter Line 1 72 $ Bucket Line 1 72 Total Cost of Fixed Cost $ $ 1,200,000.00 $ 400,000.00 Total $ 57,600.00 $ 1,200,000.00 $ 400,000.00 $ 1,657,600.00 Table 3. Cost of the Trial 2 Trial 3:- In this case we are adding 1 shift to the Min and Grind Machine, and 1 Line to the Quarts Line. But in this case the performance of filling level is 99.90% of all orders. The replication length of this run is 72 days and the number of replicates is 30. Since this politic is the cheapest one that ensure the 98% of filling ratio the simulation team decide to choose this configuration as the suggestion for the constrain of 3 working days. (Refer Appendix 1, 1 Shift, 1 QL) Page 9

SM Paint Production: Modeling Project Cost of the policy Sources of Cost per Number cost day days $ Extra Shift 800.00 72 $ Quarter Line 72 Total Cost of Fixed Cost Total $ $ - 57,600.00 $ $ 1,200,000.00 1,200,000.00 $ 1,257,600.00 Table 4. Cost of the Trial 3 Conclusion of Scenario 1 By observing the above data of various trials, we can conclude that the best way to reach 98% order completion rate (within three days of arrival of the orders) is by adding one more shift to the M&G machines, thus keeping its capacity 7 throughout the day (thereby removing bottlenecks) , and adding one more Quart line. In this case we are completing 99.9 % of the incoming orders within 3 days of receipt, and the cost of adding new resources is also the least possible at $1,257,600. The values for different resources were computed by using OptQuest software and trying various ranges for different resources. The objective function was to reduce the total cost (including the variable cost. SCENARIO 2 In this scenario, we are asked to accommodate more proportions of the large size batch orders. However, while trying to accommodate these we need to suggest appropriate changes to the resources at various stations, while keeping in mind that only $50,000/day is the profit for accommodating the new batch orders. In this case the simulation team used the best configuration found to meet the 98 % filling rate in three working days. This configuration was adding 1 shift to the Mix and Grind Machine, and 1 Line to the Quarts. For that configuration, the performance of order completion rate within 3 days of order receipt was 99.90% of all orders. Using that configuration for the private level contract the percentage of completed orders within 3days of receipt fell down to 10.48%, taking into consideration the total income for the percentage of orders delivered in the 72 days and the cost of the best configurations, no profit will be achieved ( Total expenses = $880320). After obtain this results the simulation team suggests to the company reject the proposal of the private label. (The replication length of this run is 72 days and the number of replicates is 30). (Refer Appendix 1, adding 1 shift, 1 QL for PL) Page 10

SM Paint Production: Modeling Project Profit of the policy Sources of Cost per cost day $ Extra Shift 800.00 $ Quarter Line $ Label Income 50,000.00 Number days 72 72 72 Total Income of Fixed Cost $ $ 1,200,000.00 Total $ 57,600.00 $ 1,200,000.00 $ 377,280.0000 $ (880,320.00) Table 5. Cost of the Scenario 2 SCENARIO 3 In this scenario we tried to find alternatives to the current system to try and improve the productivity without adding any more resources to the system. To get this done we tried a couple of options based upon, *ways to reduce production time, *ways to reduce waiting time in the queue, *improve the processes and reduce non-value added times in the production line We thought of primarily two ways to get this done, 1. Removing the constraints imposed on the batch sizes by the capacities of various tanks: This technique can be used to improve the usage of various resources available in the system and reduce their idle times. The facility has a network of connections according to which we can connect each process tank with every other process tank. Thus if we can connect the low size batches with tanks of higher processing capacities then we increase the uniform distribution of orders to various resources, as well as reduce the burden on certain resources. This uniform distribution can be done moving the orders towards new resources after checking for condition of “minimum utilization of that resource.” For example, the 6000 size orders can be processed in tanks of sizes ranging from 6000 to 20000. (Refer appendix 2, it shows improvement in the actual model after removing constraints imposed on the batch sizes). 2. Changing the packaging of the orders (i.e. using just one type of packaging, Gallons) : This technique will help the packaging station to work faster and complete the orders quicker than earlier, because now we have just one type of gallon to be filled, and hence it will be easier to directly transfer the order into the gallons rather than waiting for the rest of the part to be filled by quarts or buckets, like in the original case. This will free up the holding tank much quicker and hence the thinning tank too. Page 11

SM Paint Production: Modeling Project Recommendations Based upon the trials conducted by our team and the results of the simulation model constructed, we would like to suggest the following: 1. The current facility can reach performance levels where it can guarantee that 98% of the orders are completed within 3 days of its receipt, subject to the condition that the current resources are realigned to connect in a better way with each other. In other words, if we can lift the restrictions on various processing tanks with regards to the batch sizes they can process, then only we can ensure better utilization of resources. 2. The order completion rate can be even more than 98% if we can buy 1 quarter line, 1 bucket line and increase 1 shift at the M&G process. 3. To accommodate for orders from outside resources, to make profit, we need to ensure that the incoming orders do not influence the current system by forming long queue times and causing bottlenecks. This can be ensured by easing in the orders into the system rather than rushing them in quickly. 4. Increase one shift at M&G station to make the system continuous, so that once it starts it stabilizes after a certain number of replications, and does not go on increasing continuously due to formation of queues. Bibliography Law, A. M. (2001). Simulation with Arena. Boston: MacGraw-Hill. Law, A. M. (2006). Simulation Modeling Analysis. Chicago: MacGraw-Hill. Navigator, S. (2006, 08 14). System Navigator. Retrieved 02 14, 2013, from System Navigator: http://www.systemsnavigator.com/sn_sales/SimulationManufacturing2.jpg Appendix Appendix 1 Excel file uploaded to MyCourses. Contains data and results of various trials performed in the simulation model Appendix 2 Excel file uploaded to MyCourses. Contains data and results of the trial for the new system after removing the restrictions on flow of orders to different sizes of processing tanks. Page 12

Add a comment

Related presentations

Related pages

Arena Simulation

Arena discrete event simulation software provides supply chain simulation, ... Arena Simulation Software helps you get it right the first time.
Read more

Academic | Arena Simulation - Discrete Event Simulation ...

Arena is the #1 Choice of Academic Institutions Worldwide. Hundreds of universities and colleges worldwide use Arena Simulation Software. In fact, it is ...
Read more

Simulation with Arena | Contest Problems - Innovative ...

Simulation with Arena, 5/e: Student Center: Arena Software Dow... Contest Problems: Feedback Help Center: ... Contest Problems 13-15 (718.0K)
Read more

Simulation in the Automobile Industry - Productivity ...

550 SIMULATION IN THE AUTOMOBILE INDUSTRY Paint Shop Body Shop ... This study involved simulation modeling and analysis of a paint shop and an adjoining
Read more

Example Simulation Models - University of Houston

V. Example Simulation Models . A. Introduction to Arena Simulation ...
Read more

Simulation With Arena Solution Manual | Chegg.com

Simulation with Arena ... Why is Chegg Study better than downloaded PDF Simulation with Arena Solution Manuals? Chegg isn't the same as a solution ...
Read more

Simulation Software | Simulation Analysis & Tools | Autodesk

Deliver better designs more quickly and make great products with Autodesk Simulation ... Advanced building simulation and structural analysis help you ...
Read more

Simulation for Windows - Free downloads and reviews - CNET ...

The most downloaded Simulation software, including The Sims 3, Euro Truck Simulator 2, and Jurassic Park: Operation Genesis. CNET; REVIEWS; NEWS; DOWNLOAD ...
Read more