Groundhog day versus butterfly effect revisited in discrete material flow management.

Cotet, Costel Emil ; Dragoi, George ; Abaza, Bogdan 等

1. INTRODUCTION

In our increasing productivity research studies based on manufacturing systems architecture optimization we identified as one of the main difficulties in the area of synchronizing process and material flow simulation.

The main goal of our algorithm is to increase productivity by improving the discrete material flow management using the process simulation as a preliminary data. We analyze the results of the material flow simulation and we identify the flow concentrator for a preliminary manufacturing architecture based on process simulation results (Cotet & al., 2007). We propose a solution for eliminating flow concentrators. We perform a second simulation to validate the optimized manufacturing architecture by obtaining an increased productivity. Some of the necessary data for the material flow simulation like cycle times for the work points defined in our models are provided by process simulations describing each work point manufacturing cycles. The material flow simulator is integrating the process simulation results at the level at each work point in order to provide a complete model of the manufacturing system.

Analyzing the different behaviors in modeling for the process simulation and for discrete material flow simulation, we had defined two approaches starting from the plot of two films illustrating the models different characteristics.

The Groundhog Day film is about a man who finds himself living the same day over and over and over again. He is the only person in his world who knows this is happening, and even if he is free to change what he says and does from one Feb. 2 to the next, it will always be Feb. 2 for everyone else in the world, and he will always start from the same place. All the others will repeat themselves unless he changes the script for one day, but tomorrow they will have forgotten their new lines and be back to the first draft of Feb. 2. He is therefore trapped in a seemingly endless "time loop" to repeat the same day in the same small town (Ebert, 2005). We define the Groundhog Day effect the process simulation algorithm that reproduces every manufacturing cycle identically with the previous. According with this effect, by simulating one work point activity for multiple manufacturing cycles the same simulation will be reproduced multiple times with the same parameters. If we intend to change the script for one manufacturing cycle we can perform a new simulation and introduce this new model between the identical ones, but the previous and the next episodes in the manufacturing chain cycles will remain the same. It is impossible for us to have a script where a manufacturing cycle became different from the previous.

The Butterfly Effect film is based on the chaos theory teaching us that small events can have important consequences illustrated by butterfly flapping its wings in Asia could result in a hurricane halfway around the world. In the film, every time the main character changes his past, he goes to the exact moments when he blacked out. In the first timeline he simply blacks out traumatic moments; later, he is able to revisit these blackout moments by re-reading journal entries about them, which suggests that the blackouts could have been caused by his ability to revisit the past and he "blacks out" when his future self is revisiting his past self--a causality loop (Ebert, 2004).

We use the Butterfly Effect in material flow simulation because allowed us to define the script of a manufacturing cycles chain in which some of the manufacturing cycles are different by introducing stochastic distribution laws and not fixed values for MTBF (the mean time between failure) or MTTR (mean time to repair). In this case some of the manufacturing cycles will be different due to repair times who will personalize the multiple cycle chain.

As we already emphasize at the beginning of this chapter, one of the main difficulties in building an integrated model is that due to those two special effects characteristics process and material flow behavior are very different. In order to synchronize the material flow and process models of manufacturing cycles we had to modify the process simulation program in order to personalize the process simulation changing the Groundhog Day with the Butterfly effect approach.

[FIGURE 1 OMITTED]

2. SIMULATING THE INTEGRATED MODEL

The first step of our algorithm is to realize the preliminary parametric model of the manufacturing system and to simulate the material flow in the system in order to identify the flow concentrator (fig.1). Different algorithms are used for the diffused and concentrate manufacturing systems, for terminating or non terminating simulations according with the specific constraints characteristic for each case. For a designed manufacturing architecture it is always useful to simulate the material flow conduct before applying our design into practice in order to avoid potential flow concentrators generating low productivity or even blockage. The leading actor able to manage this area will be the flow simulator.

We define diffused manufacturing systems as architectures with more than two work points connected by transport & transfer systems and using deposits at local or system level (Cotet & Dragoi, 2003). We agree here with the thesis that within the class of stochastic simulation models, one further distinction is necessary: simulations can be either terminating (sometimes called finite) or nonterminating in nature, with specific algorithms for each category (Sanchez, 2001). We consider a material flow and process synchronous simulation the simulation of a model where at the level at the work point the process simulation is concomitant with the material flow simulation. We consider that due to the complexity of the mean time between failure (MTBF) and mean time to repair (MTTR) modeling for various machines and manufacturing systems one can provide the best solutions for such productivity improvement based on stochastic distribution laws and not on fixed values, even if fixed values for these parameters would be used for some particular finite short simulations.

A running diffused manufacturing model for terminating simulations is presented in fig. 2. Each work point parameters were established based on process simulation results as described before.

Using the Witness reports one can identify the flow concentrators and according with manufacturing planning constraints can propose technological solutions or modifying manufacturing architecture based solutions to eliminate those concentrators. In order to validate the increased productivity obtained on the optimized manufacturing system architecture, a new simulation is necessary. If one weighs the productivity for the preliminary and optimized architecture the increasing productivity due to the optimization can be quantified. A very important issue of this algorithm is to evaluate if the optimized architecture supplementary costs are or not covered by the productivity gain. This financial analysis is based on a NPV algorithm and validates the financial profitability of the optimization. At the first level main parameters for the work points modeling in material flow simulation are provided form CAM simulations describing the manufacturing process (Lee, 1999). This way the material flow simulator is integrating the process simulation results at the level at the work point in order to provide a complete model of the manufacturing system.

In order to realize the synchronized simulation the material flow terminating simulation algorithm is applied for each manufacturing system of the enterprise. In this model the work points are the basic simulation model units describing the manufacturing system material flows. In the end at the level of the entire virtual manufacturing system the process simulation for each work point is synchronic with the material flow simulation describing the entire system activity. This local synchronizing process of the material flow simulation allowed us to change some of the manufacturing cycles introducing stochastic distribution laws values for MTBF, MTTR or failed parts adapted to each work point characteristics. If we use the integrated model versus the asynchrony classic one the accuracy of the simulation results for the presented case study increase with 10 %.

[FIGURE 2 OMITTED]

3. CONCLUSION

Starting from the script of two films (The Butterfly effect & Groundhog Day) we describe here the theoretical models we defined in order to generate a manufacturing architecture optimization algorithm. The material flow and process simulation models based on specific software solutions are the main actors of this simulation project undertaken with the goals of demonstrating and confirming production rates of a manufacturing process based on a proposed design layout and operational data and of identifying ways of improving the design of the system in order to increase those production rates. According with this algorithm one can analyze the results of the material flow simulation and identify the flow concentrator for the manufacturing system. If an architecture modification is proposed as a solution for this problem a second simulation to validate the optimized architecture and the obtained increase of productivity is necessary.

4. REFERENCES

Cotet, C. E., Dragoi, G. & Carutasu, G. (2007). Material Flow & Process Synchronous Simulation In Concentrate Manufacturing Systems, Annals of DAAAM for 2007 & Proceedings of The 18th International DAAAM SYMPOSIUM, "Intelligent Manufacturing & Automation: Focus on Creativity, Responsibility and Ethics of Engineers", Katalinic, B. (Ed.), pp. 180-181, ISSN 17269679, ISBN 3-901509-58-5, Zadar, Croatia, October 2007, Publisher DAAAM International Vienna 2007.

Cotet, C.E., Dragoi, G.S. (2003). Material Flow Management in Validating Concentrate and Diffused FMS Architectures. In: International Journal of Simulation Modelling IJSIMM, no. 4, December 2003, pp.109-120, ISSN 1726-4529, Vienna.

Ebert, R. (2005). Groundhog Day, Available from: http://rogerebert.suntimes.com/apps/pbcs.dll/article?AID=/ 20050130/REVIEWS08/501300301/1023 Accessed: 200806-22

Ebert, R. (2004). The Butterfly Effect, Available from: http://rogerebert.suntimes.com/apps/pbcs.dll/article?AID=/ 20040123/REVIEWS/401230301/1023 Accessed: 2008-0622

Lee, K. (1999). Principles of CAD/CAM/CAE Systems, Addison Wesley Longman, Inc., ISBN 0-201-38036-6, USA.

Sanchez, S. M. (2001). ABC's of output analysis, Proceedings of the 2001 Winter Simulation 2001, Peters, B.A., Smith, J.S., Medeiros D.J. (Ed.), CD-ROM, Presses Association for Computing Machinery (ACM), New York.