Classical versus numerical control machines in discrete material flow simulation based manufacturing architectures optimization.

Anghel, Florina ; Cazacu, Dragos Alexandru ; Bucur, Carmen Cristiana 等

Abstract: The research results presented in our paper are based on reports provided by different simulation models, using Witness software for a production cycle of a main shaft from a milling machine in order to determine the best manufacturing architecture based on improved productivity.

The main purpose is to find out the number of machines necessary in order to obtain good production rates. We will also compare the productivity that will result from manufacturing the shaft when using classical machines, such as debtor machine, lathe machine, milling machine, boring and grinding machine with the one obtained when manufacturing the shaft on numerical control machines.

Key words: simulation, production cycle, optimization, productivity, material flow concentrator.

1. INTRODUCTION

Before starting a manufacturing process it is advisable to run a simulation for the production cycle. In this way one can determine the productivity, the necessary number of machines, the personal involved in the production cycle in order to obtain a good production (Dragoi et al., 2004).

When starting a simulation we must have established the basic objectives (Rosu et al., 2005). In this case we considered that there was an important order of these shafts and we just need to honour it, as soon as possible.

Knowing all necessary times for all involved operations (they were calculated) we determined the machines that we needed and after running several simulations we optimize the number of each machine in order to obtain a good production, without material flow concentrators (Anghel & Cazacu, 2006).

We define diffused manufacturing systems as architectures based at least 2 work points surrounded & assisted by transport, transfer & deposit facilities (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).

2. DESCRIPTION AND OPTIMIZATION OF THE MODEL WHEN USING CLASSICAL MACHINES

The machines that are going to be used are: debtor machine (D1), lathe machine (S1), milling machine (F1), boring and grinding machine (MAR1), each of them having an operator supervising the production cycle. After running the simulation, for a finite period of time (one month), 22 working days, 16 working ours a day, we obtained a productivity of 138 pieces (preliminary architecture).

On the S1 entrance there is a parts blockage (material flow concentrator) because of the big times necessary for the manufacture on this machine. The lathe is the only machine that works continuously and all the other machines are waiting for it to finish the blanks.

[FIGURE 1 OMITTED]

Because of this problem we started an optimization process. We introduced a second lathe machine, obtaining in this way an increasing of the productivity from 138 to 171 pieces per month. For this second lathe machine we also have to introduce another operator, because with one the productivity cannot be raised.

Now the parts are passed on the milling machine quicker, but another material flow concentrator is found on the boring and grinding machine that can't keep up with be big number of parts send by the lathe machines. This was why we continued optimizing the model by introducing a second boring and grinding machine. In this new shape we obtained a productivity of 276 pieces, but the lathe machines are again a flow concentrator.

The best solution was obtained after we added another lathe machine; in this way the productivity was raised to 322 pieces per month, from only 138 in the preliminary analyses. For developing this solution in production we have to increase the investment with 3 labours, 2 lathe machines and another boring and grinding machine.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

3. DESCRIPTION AND OPTIMIZATION OF THE MODEL WHEN USING NUMERICAL CONTROL MACHINES

When building the model with numerical control machines we chose: a debtor machine (D1), a lathe machine (S1) and a boring and grinding machine (MAR1). We don't need a milling machine with numerical control because the manufacturing necessary times on this machine are too small and the investment would not be justified. So for milling operations we will use a classic milling machine with an operator, which will supervise the process. For all the other machines we will assign a monitor; the presence of an operator is not needed.

After the first simulation the productivity is of 208 pieces per month. By using numerical control machines we already obtained an increasing of the productivity if compared to the productivity given by classical machines in the preliminary architecture (138 pieces per month).

But the lathe numerical control machine is again overload and the conveyor C1 is always full. After introducing a second lathe numerical control machine we raised the productivity to 417 pieces per month. But we can still see a material flow concentrator on both lathe numerical control machines. So, for a better optimization we will use 3 lathe numerical control machines, which will lead to a higher productivity: 486. We also notice that this time, there wasn't a material flow concentrator on the boring and rectification numerical control machine.

So after we added 2 lathe machines we obtained a continuous material flow with no concentrators. We could go further, trying to clone also the debtor machine (D1), but this would lead us to the starting point.

[FIGURE 4 OMITTED]

4. CONCLUSION

In the simulations that we run the main purpose was the increasing of the productivity. For this we added machines in order to eliminate the material flow concentrators that were located on different machines.

This kind of simulation shows important aspect regarding the manufacturing process: eventual material flow concentrators, how the presence of the operator affects the production cycle (for example: if we have two lathe classical machines we will need two operators in order to increase the productivity). Another important aspect of this simulation is that we can determine the necessary number of machines for an imposed productivity.

After doing a number of optimizations, this consisted in adding machines in order to eliminate material flow concentrators we managed to obtain a productivity of 322 pieces per month when using classical machines.

On the other hand after we found the optimized model we found out a productivity of 486 pieces per month when using numerical control machines.

After establishing the desired productivity we will need to do a calculus in order to find out if the investment of buying 3 labours, 2 lathe machines and another boring and grinding machine in case we use classical machines production or 2 lathe machines in case we decide to use CNC machines is justified. If we find it impossible to get financially to the optimum solution we might use a cheaper solution found in the way of optimization.

Because we considered that the only target is to increase the productivity on month we did not make the investment profitableness. However as a first step in our further research we already started to develop an algorithm based on NPV analyses in order to establish if the investment needed for the material flow optimization is justified.

5. REFERENCES

Anghel, F. & Cazacu, D. A. (2006). Using value analysis in mechanical engineering design production costs reduction, in: Proceedings of The International Conference on Manufacturing Systems ICMaS, Ispas, C., Ghionea, A. & Constantin, G., (Ed.), pp. 499-502, ISSN 1842-3183, Bucharest, Romania, 26-27 October 2006, Editura Academiei Romane, Romania

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.

Dragoi, G.; Radulescu, B. & Tichkiewitch, S. (2004). A cooperative system for the design of the mechatronic products with the multidisciplinary optimization in Methods and Tools for Co-operative and Integrated Design, pp. 281-290, edited by Serge Tichkiewitch and Daniel Brissaud in Kluwer Academic Publisher, ISBN 1-4020-1889-4,. Netherlands.

Rosu, S.M.; Dr?goi, G.; Guran, M.; Cotet, C.E. & Rosu, L. (2005). Knowledge management support to product design process within CESICED platform, Proceedings of The Second International Conference of Integrated Engineering, pp. 189-190, with CD-ROM support for the full papers, ISBN 973-625-259-0, .16th-18th October 2005, Timisoara, Romania

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., CD-ROM, Presses Association for Computing Machinery (ACM), New York.