Value stream mapping using simulation with ARENA
- 1. Hrishikesh Khamkar Simulation Research Project Using Arena simulation to assess the effectiveness of a future state map, in Value Stream Mapping, by quantifying the change in non-value added time, as compared to current state map. Abstract:Value Stream Mapping is a technique of lean manufacturing in which the whole value stream of a production system is plotted on paper to understand the flow of material and information, and to find out areas with room for improvement. These improvements are done one by one, starting with making changes to factors which are easiest to modify (Rother & Shook, 1999). However, a Value Stream Map is a snapshot of an actual shop floor, and it may or may not give a realistic feel of the actual working conditions. Using simulation to analyze various aspects of a working shop floor production system helps us to assess the behavior of our production system in uncontrolled situations, and even in circumstances with increased complexity. Also, the results obtained can be analyzed by Lean coordinators to gain a better perspective of the operating system (Gullander, 2009). In this paper we would be studying and documenting the effect of various changes, made to certain operating characteristics of our system, on certain useful output parameters. Numerical values of this study will be obtained by running modeled simulations, using ARENA simulation software. Introduction:In the current world scenario, it has become of utmost necessity to „Continuously‟ seek improvements in our manufacturing systems. Quick analysis of possible variations in our manufacturing systems facilitates increase in productivity, which is the primary goal of most manufacturing companies (Gullander, 2009). Lean manufacturing is one of the most sought after method to improve productivity of any company. However, even when applying Lean tools it is of utmost necessity to know where to apply them and what kind of improvement should we expect from it (to know we are doing it right). Along with this, it becomes necessary to account for the behavior of the new system, achieved by applying lean techniques, under the effect of system variations and uncertainty in the arrival of input variables (Amr, 2011). This calls for a need to use powerful tools, like Simulation, to give us the liberty to study the effects of proposed changes to the manufacturing systems. Simulation can be used to implement proposed changes to a Value Stream Map, and then collect statistics which can be further analyzed to derive conclusions about implementing those changes. Various applications have shown that we require Simulation to find and study factors which couldn‟t be seen by using Lean alone (Marvel, 2006). Lean Manufacturing, when used in Value Stream Mapping, acts as an assurance step while we are trying to bridge the gap in between the current state map and the future state map. This helps us to point out the crucial factors to look into, while we plan to improve our productivity (Amr, 2011). In this paper, we try to study certain factors in detail, elimination of which can help us reduce Non-value added time of the product. Also, we are able to analyze and observe the change in statistical output, if we make certain changes which may seem like an obvious improvement in the first
- 2. look, but may affect certain performance parameters negatively. This becomes a good foundation to make ourselves realize to tread cautiously when working with Value Stream Mapping, because it‟s not just about making changes, but also about making the right kind of changes (Harris, 2001). Literature Review:Since the paper is a study which consists a combination of Simulation and Lean Manufacturing, books like „Simulation with Arena‟(Kelton, 2004) and „Creating Continuous Flow‟ (Harris, 2001) were of immense use. Along with this, papers related to „Dynamic Value Stream Mapping‟ and doing „Value Stream Mapping using Simulation‟ from Winter Simulation Conferences provided with literature related to the scope of the idea about applying simulation to Value Stream Mapping. Also, research papers of students pursuing studies related to justifying the applicability of simulation to lean manufacturing, were of use. Procedure:As mentioned by Gullander and Solding (2009), there are some weaknesses related to using Simulation or Value Stream Mapping, alone, to analyze and improve the performance of any manufacturing system. Hence, it is necessary to embed these two important tools into each other, such that the combination of these two is able to replicate the real life situation of our manufacturing system as well as leave sufficient room to introduce variability and randomness into various performance parameters of the whole system. A proper gel of these two techniques helps us to keep the system in dynamic motion, while we study the effects of randomly occurring events on the overall system. Here, we are using different models which have been designed to work with parameters, suitable to replicate most common problem-causing elements in a manufacturing system. These include effects on the output by an unbalanced line, effects of changing layout of a line to reduce accumulation of inventory before stations, types and methodologies of using same transportations but in different ways to reduce or completely eradicate travel times (non-value added time) etc. The data for all the models has been assumed after referring to examples in the book, “Creating Continuous flow” (Harris, 2001). The simulation model is designed to replicate a large element of randomness in most common aspects of a manufacturing system like, travel times, operator scheduling, breakdowns of a machine, inspection times etc. According to Harris & Rother (2001), for a production line to meet the demands of the customers the rate of production should be equal to the takt time (consumption rate) of the customers. To get this done, the production line must be designed to have minimum or zero amount of Nonvalue adding activities. This include waste due to motion, transportation etc. (7 wastes of lean). Hence, waiting in a queue for a long time indicates that the entity is spending time in the system which is not contributing to its value addition, from a customer‟s point of view. Therefore, we try to compare the effect of certain changes on the time spent by all the entities in the queue.
- 3. Application and analysis:We will apply simulation techniques to models developed to imitate real life situations on the shop floor. Effect of Downtime on the waiting time in a queue:To understand the effect of downtime on waiting time of a system, we simulate a production line which has a layout as shown in Appendix 1.We have constructed two models, one with failure as a part of the machines operations and another without the failure. Usually, failure or breakdown of a resource is a very common problem in manufacturing systems, and it has following disadvantages, It is completely unpredictable Time to restart the machine is dependent on the nature of the breakdown During peak hours of production, it can easily lead to bottleneck formations in the production line. To overcome this problem most companies schedule regular checks and maintenance cycles of the resource machines. Even though this causes a temporary stoppage in the production system, it is very beneficial in the long run. Doing regular machine checks helps to keep up the performance levels and prevents unprecedented delays due to machine breakdown. (Assumptions have been made about the process times and arrival or orders to the shop floor. Also, the failure module is designed based upon assumed values for time of occurrence of failure and time taken to get rid of the failure). In the simulation model of Appendix 1, the resource at process assemble is given a failure schedule, because of which it breaks down occasionally, leading to the formation of long queues in the process. To overcome this problem, the shop floor supervisor can use Arena to simulate the regular checks of the operating machine (in this case assembly). To imitate a real life situation the user is given the liberty to choose the timings of machine checks. This is basically done using the Arrival blocks and Assign elements to use keyboard keys for deciding when to schedule a check-up of the machine ( “D” for decrease resource and “I” for increase resource)( this model is partially based upon sample model from SMARTS folder of Arena simulation, controlling resource capacity). Based upon a shop floor manager‟s perception combined with the performance expectations and past usage of the machine, the user can decide when to do machine checks. The simulation is carried on for a duration of 3 days and 3 replications. Statistics gathered are primarily about average time spent by an entity, in the queue of assemble process, throughout the simulation. Observations:Case 1:- Assembly process with uncertain occurrences of breakdown Case 2:- Assembly process with occasional checks of machine to avoid breakdown Time to complete Assembly process per entity Time to complete Assembly process per entity Average time = 2.921895 hours Average time = 2.843519 hours After completion of the simulation we realize that the average time taken by the entity in the assembly process is greater when there are breakdowns due to failure as compared to the simulation when there are frequent machine checks (3-4 every two hours). (Please refer Appendix 3 for detailed data collection)
- 4. Effect of reduction in process times and changes in resource allocation on waiting time in queue:Queue Name Waiting time Process A1.Queue Number waiting in queue 0.01767777 Process A1.Queue 0.4835 Process A2.Queue 0.01703996 Process A2.Queue 0.4887 Process A3.Queue 0.03090375 Process A3.Queue 0.8950 Process A4.Queue 0.04233805 Process A4.Queue 1.2646 Queue Name Number waiting in queue Queue Name Waiting time Process C1.Queue Process C2.Queue Process C3.Queue Process C4.Queue 0.03639 Process C1.Queue Process C2.Queue Process C3.Queue Process C4.Queue 1.0168 Queue Name Table 1 0 0.0475 0 0 1.4416 0 Table 2 Queue Name Number waiting in queue Queue Name Waiting time Process B1.Queue 0.1803 Process B1.Queue 4.7925 Process B2.Queue 0.1978 Process B2.Queue 5.9619 Process B3.Queue 0.1585 Process B3.Queue 4.6935 Process B4.Queue 0.1799 Process B4.Queue 5.4417 Table 3
- 5. In Value Stream Mapping we prefer to take the current state and based upon the output requirements of the production line we plan the production line (Rother, 2009). Based upon this requirement the appendix 4 model is planned to show the improvements happening in a system as we make step by step improvements. The first step indicates repositioning of the inspection station and increase in resources. Referring table 3 we realize that there isn‟t much of a change. While in the next step we decrease the process time for process 2, which results in decrease in the waiting time of the queue, thereby indicating that if we want to improve the process and decrease the overall waiting time then we should focus more on doing kaizen activities to reduce the process times. Optimizing the transportation time in a production system:By reducing the transportation time we can facilitate fast delivery of raw material, subassemblies etc. In this case however, when you have a shortage of resources and transporters, then we should try to increase the performance of the transportation system by either improving the speeds of its movements or by increasing optimization of transporter motors (i.e. making them cover as much distance as possible in as less time as possible) or increase the number of transporters. (Refer Appendix 5) Conclusion and future work:Applying simulation to VSM has a its own advantages, as mentioned by Gullander (2009) like, 1. Getting a real time view of the shop floor 2. Viewing the bottlenecks in the system 3. Less cost incurred in trying new layouts on shop floor 4. Improving worker utilization (after ensuring improvements in simulation) Keeping this in mind it seems advantageous to use simulation for assessing the effect of randomness on the working of system. This helps us to know how a particular layout or resource will behave if it is overloaded or provided with new resources (Donatelli. N.D). As mentioned by Gullander (2009), it will always be useful to make proper prototypes of actual simulation layouts rather than trying them on fictional plants. Along with this it is of utmost necessity to get accurate data (as much as possible) to plot the input parameters and distributions for operator times. This will make the output more realistic and even more useful when making suggestions to a production facility regarding what changes will be needed to get what they want. In future work, it will be good to accommodate production control plans with the actual simulation and then make improvements to the process plans.
- 6. References:Donatelli, A & Harris, G. (n.d.).Combining Value Stream Mapping and Discrete Event Simulation. Retrieved from http://www.scs.org/confernc/hsc/hsc02/hsc/papers/hsc003.pdf Gullander, P & Solding, P. (2009). Concepts for simulation based value stream mapping, Proceedings of the 2009 Winter Simulation Conference M. D. Rossetti, R. R. Hill, B. Johansson, A. Dunkin and R. G. Ingalls, eds. Retrieved from http://www.informs-sim.org/wsc09papers/215.pdf Kelton, D & Sadowski, R & Sturrock, D. (2004). Simulation with Arena Rother, M & Shook, J. (1999). Learning to See, The Lean Enterprise Institute, Brookline, MA. Amr, A& Amr, M &Crowe, J. (2011). Integrating Current State and Future State Value Stream Mapping with Discrete Event Simulation: A Lean Distribution Case Study. Proceedings of SIMUL 2011: The Third International Conference on Advances in System Simulation. Marvel, J & Standridge, C. (2006).WHY LEAN NEEDS SIMULATION. Proceedings of the 2006 Winter Simulation Conference Harris, R & Rother, M. (2001). Creating Continuous Flow: An Action Guide for Managers, Engineers and Production Associates Appendix:Appendix 1:- model to indicate model with failures Appendix 2:- model to indicate model with machine checks Appendix 3:- excel sheet comparing statistics Appendix 4:- model to show various process and improvements made in them Appendix 5:- Model to show transportation model Appendix 6:- model to show improvement in transportation model (Uploaded as separate files on RIT MyCourses portal)