LOGIC CONTROL DESIGN AND DISCRETE EVENT SIMULATION MODEL

IMPLEMENTATION FOR A DE-MANUFACTURING PLANT

Andrea Cataldo (Researcher)

Istituto di Tecnologie Industriali e Automazione - Consiglio Nazionale delle Ricerche e-mail: andrea.cataldo@itia.cnr.it

(PhD executive) Dipartimento di elettronica, Informazione e Bioingegneria – Politecnico di Milano

e-mail: andrea.cataldo@mail.polimi.it

Riccardo Scattolini Dipartimento di elettronica, Informazione e Bioingegneria – Politecnico di Milano

e-mail: riccardo.scattolini@polimi.it

ABSTRACT In this paper the design of a multi-level control system for a de-manufacturing pilot plant is described. In particular the implementation of the Low Level Control System (LLCS) based on Control Sequences (Sn) has allowed to decouple the High Level Control System (HLCS) from the field, by making the advanced control algorithms development independent from any specific plant technological implementation. The LLCS integrated with the de-manufacturing process have been emulated and modelled into a dynamic discrete event simulator. This approach allows a better automation system architecture definition mainly for Advanced controls techniques. 1. INTRODUCTION In the context of the advanced control techniques, Model Predictive Control (MPC) is nowadays the most influential advanced control method in the industrial practice, mainly in the field of process control, see e.g. [1]-[4]. MPC is an on-line optimization-based control approach which allows to account for operational constraints and multi-variable systems and can be used to deal with continuous and discrete decision variables besides including logic relations [5]. According to these advances the application of hierarchical MPC in the manufacturing industry is very promising; in fact great advantages could be achieved in the design of the overall production system architecture in terms of scalability, adaptivity to changing environments, reconfigurability with respect to the addition, replacement or removal of subsystems. MPC can be mainly used according to two main control system architecture: the centralized approach and the multi-layer one [6]. The first structure consists of a central control unit which collects the data coming from all the sensors and determines the optimal control inputs for all the actuators of the system. For complex large-scale systems, a centralized structure results in high computational requirements, which makes it unsuitable for on-line, real-time control. A second possibility consists on resorting to multi-layer, hierarchical, control structures [7], [8]. At the lowest level of the automation system architecture, local Programmable Logic Controllers (PLC) manage the field plant sensors and actuators while at the higher level an MPC controller coordinates and assigns higher-level tasks, aiming to the manufacturing plant efficient management.

In this paper a de-manufacturing pilot plant [9], [10], see Figure 1, has been taken into account, which main task consists on repairing or recovering raw materials from electronic boards by means of moving pallet through a transport line. In order to design and implement a control system, which description represents the objective of this paper, the process equipment mechatronic structures have been analyzed, which lead to choose a multi-level control approach. The study of the lower level suggested to use pure logic control software functionalities, characterized by features like modularity and reusability in order to reduce the software complexity and to simplify its maintenance [11]. Then a set of Control Sequences has been defined, capable to perform each kind of requested pallet route. Moreover in order to make independent the higher level [12] from the lower implementation one then a pilot plant abstract representation based on direct graph has been obtained. This kind of mathematical formalism allows to develop any advanced control algorithms running at the higher level without any care for the LLCS development. The paper is organized as follows. In Section 2 the de-manufacturing pilot plant is described. In Section 3 the LLCS concept and implementation, together with the implications gained for the HLCS, are considered. Section 4 is devoted to describe the Discrete Event Simulation (DES) model of the plant integrated with the regarding LLCS emulation, while Section 5 draws some conclusions and hints for future works.

2. THE DE-MANUFACTURING PILOT PLANT

The de-manufacturing pilot plant, see Figure 1, has been designed by the Institute of Industrial Technology and Automation, National Research Council (ITIA-CNR), Italy, where it is also located, with the aim of testing and repairing of electronic boards.

Figure 1: The de-manufacturing pilot plant

It is mainly composed by the following elements: Fifteen Transport Modules TMn, n=1,…,15, connected together according to a

specific layout and composing a modular and flexible transport line; Cell M1, a load/unload robot cell, where the electronic board is loaded on a pallet,

which is then placed on the adjacent Transport Module of the transport line;

Cell M2, a testing machine, where the board is tested and the specific failure is identified;

Cell M3, a reworking machine, where the board is machined in order to be repaired; Cell M4, a discharge machine, where the non-repairable boards are discharged from

the pallet and destroyed. A sequence of operations that can be performed by the pilot plant on each board is the

following: The board is loaded on the pallet by M1; The transport line moves the pallet to M2 where the board is tested and its failure

mode is identified; The pallet with the board is moved to M3 where the failure is repaired (if possible); The pallet is moved back to M2 and the test is repeated. If the board is properly

working, it is sent back to M1 where it is unloaded from the pallet and stored in the warehouse, otherwise it is sent to M4 where it is discharged from the pallet and destroyed;

In any case, the pallet is ready to load a new board to be tested. The basic element used to move a pallet is the Transport Module. It allows the pallet

laying in specific zones of it or moving towards the possible directions which are imposed by the transport line topology. Each Transport Module has a specific configuration in terms of mechanical structure and automation system instrumentation. In order to deeply explain the Transport Module working function, the TM8, one of the most flexible Transport Module in the de-manufacturing transport line, has been taken into account as example. In order to move a pallet, a Transport Module can use up to four different kind of devices with regarding sensors and actuators, see Figure 2, in which all the possible pallet movements are also indicated by means of continuous arrows. They are:

a) The Main Track which is used to move Forward or Backward (M_Tr_F/B) the pallet along the main direction of the Transport Module;

b) The block pallet Piston (Ev_P) which is used to stop the pallet during its movement onto the Main Track;

c) The Stacker Crane which is used to laterally move the pallet by means of the Up and Down commands (Ev_Sc1/2_U/D) and the regarding Left and Right motors activation (M_Sc1/2_Tr_L/R);

d) The Setback which, by means of the Up and Down commands (Ev_SL/R_U/D), allows to stop the pallet in the right position during its lateral movement.

Figure 2: The Transport Module

The deeply analysis of the transport line mechanical structure, has allowed to deduce that a pallet can lay or move to / from different Transport Module areas; according to its specific configuration there can be up to three areas, called Buffer Zone. By generalizing, in the sequel BZi,j will denote the j-th (j=1,2,3) Buffer Zone of the i-th (i=1,…,15) Transport Module. Then the pallet placed in a Buffer Zone can be moved forward, in some cases backward, or to the lateral positions, so allowing for different possible paths along the transport line. In Figure 3 a sketched representation of the TM8 has been given in which the three Buffer Zone BZ8,1, BZ8,2 and BZ8,3 have been indicated.

Figure 3: TM8 structure

3. THE LOW LEVEL CONTROL SYSTEM (LLCS)

3.1 LLCS basic concept The control system of the plant has been designed according to a multi-level, hierarchical

structure [7]. At the higher level, a coordinator has to manage the movement of the pallets along the transportation line in order to optimize the plant performances and to fulfill a set of logical constraints imposed by the transport line structure. At the lower level, a set of Programmable Logic Controllers PLC, one for each Transport Module, acquire the sensor signals and drive the actuators. In order to implement the pallet movement through the different Buffer Zone according to the transport line topology and the Transport Modules structure, LLCS functionalities need to be specifically designed and implemented.

The Buffer Zone definition and the regarding actuators and sensors involved into the pallet movement have led to identify a set of thirty-six Control Sequences (CS) capable to perform all possible pallet movements on the transport line, see Table 1 (Appendix). This means that it has not been necessary to develop specific control software for each Transport Module PLC but just only download those Control Sequences necessary to implement the pallet movement for that specific Transport Module configuration.

In Figure 4 the Control Sequences related to the Transport Module TM8, identified by arrows and labeled Sn, nϵ{1,…,36}, are showed. For example S21 is used to move the pallet from BZ8,2 to BZ8,3 while S28 is activated to laterally move the pallet from the adjacent Transport Module to BZ8,3.

Some Control Sequences are similar in terms of implemented pallet movement implementation but they are different regarding the devices or sensors or actuators involved. Then both S15 and S28 are used to laterally move a pallet from an adjacent Transport Module to TM8 but S15 do not have to manage any Setback while S28 has to control the Setback Right.

Figure 4: TM8 Control Sequences

The LLCS design approach allows in general to download a specific Control Sequence

into more PLC. This possibility allows to reuse the Control Sequences by simplifying the design, the development and the maintenance of the plant control software. In order to highlight such concept, in Table 2 (Appendix) the sets of Control Sequences related to each Transport Module are listed while in Table 3 (Appendix) the Transport Modules which uses a specific Control Sequence are grouped.

By representing the Buffer Zones and the Control Sequences related to all the fifteen Transport Modules, it is possible to represent the whole transport line as showed in Figure 5. From the schema it is possible to deduce how to combine the Control Sequences in order to obtain the pallet movements. For example in case a pallet lay in BZ15,3 and has to move to BZ1,1 then S3 must be applied to TM15 and, at the same time, S19 must be activated to TM1.

Figure 5: Buffer Zone and Control Sequence transport line representation

3.2 LLCS implementation Each Control Sequence has been implemented into the ISaGRAF platform [13], by

adopting the IEC 61131-3 international standard for Programmable Logic Controllers, in particular the Sequential Function Chart (SFC) programming language.

In Figure 6 the Control Sequence S10 is showed as example.

Figure 6: S10 implementation in the ISaGRAF platform

In order to deeply analyze the Control Sequence software implementation and use, two

examples of pallet movement are given, both related to the TM8. The first one regards the pallet coming from the previous Transport Module TM6, in

particular from BZ6,3 (S3) and moving to BZ8,1 (S9). The M_Tr_F of TM6 and TM8 must be switched on: the first one for a certain interval time while the second one until the pallet has reached the proximity sensor Sx_P_Sc2. A delay time is used to postpone the M_Tr_F switch off in order to guarantee the right pallet positioning into BZ8,1. It is calculated according to the Main Track speed (0.167 m/s) and the proximity switch physical position. In particular the sensor Sx_P_Sc2 is located 25 cm far from the edge of BZ8,1 which means that the pallet takes about 1.5 s to fully enter into BZ8,1 itself.

The second example concerns the pallet movement from BZ8,1 to BZ8,2 performed by the command u14,15 and translated into the Control Sequence S10. In this case it is necessary to move down the Stacker Crane 2 by switching on the regarding actuator Ev_Sc2_D and by switching on the M_Tr_F. As soon as the Stacker Crane 2 has reached the lower position and a pre-defined delay time, equal to 0.5s, has elapsed then the Stacker Crane 2 rises up again in order to allow stopping a possible pallet coming from the previous Transport Module. The pallet moved by the Main Track is stopped by the pallet block Piston (Ev_P) and detected by the regarding sensor Sx_P_Ps. Then a delay time equal to 0.5s elapses after that M_Tr_F can be switched off because the pallet is correctly laid into BZ8,2.

3.3 LLCS implications The Buffer Zone and Control Sequence transport line model showed in Figure 5 is strictly

dependent from the specific LLCS design approach that means that, in case of any Control Sequence redefinition, then the transport line representation will change and then the HLCS design too. Moreover, from the HLCS point of view, it is useless to know what Control Sequence has to be actuated in order to move a pallet; on the contrary it is necessary to know what are the two Buffer Zone involved in the pallet movement. In this way it is possible to define specific variables uniquely associated to the pallet movement between two specific Buffer Zone. In practice these defined variables could be considered the control actions of the HLCS which have to be sent to the LLCS in order to manage the field plant. Hence the de-manufacturing plant can be represented as a directed graph with nodes and arcs, see Figure 7. The nodes represent both the Buffer Zone (circles) and the machines Mi (rectangles) where the pallet can lay, while the arcs are associated to the HLCS commands used to move the pallet from a node to an adjacent one. For convenience, these nodes are progressively numbered and labeled N1-N35. Specifically the BZi,j are labelled N1-N31 while the machines M1-M4 are labelled N32-N35. Associated with the arcs there are the labels ui,j, i,j=1,…, 35, which just represent the available HLCS commands allowing to move the pallet from node Ni to node Nj, according to the specific transport line topology. For example the command u2,3 is used to move a pallet from N2 (BZ1,3) to N3 (BZ2,1).

In order to translate these commands into the regarding Control Sequences then a C code software interface module has been implemented, which means for example that the command u2,3 will be converted into S3 (applied to TM1) and S1 (applied to TM2).

Figure 7: Direct graph pilot plant representation

4. THE DE-MANUFACTURING PLANT DES MODEL In order to create a software tool for the HLCS testing phase, a de-manufacturing plant DES

model has been developed by means of the SIMIO [14]. The methodology used to implement such a model is described in [15]. In particular the generic Transport Module has been considered as a Process Equipment characterized by its mechanical behavior, see Figure 8, and by its control behavior described by means of Finite State Machine (FSM).

Figure 8: Transport Module DES mechanical behavior

In particular the generic Transport Module DES mechanical structure is based on the

following elements: a) A set of paths that allow the pallet flowing through the Transport Module model; b) A network of nodes that represent the Buffer Zone; c) A network of nodes that represent the proximity switches placed on the Transport

Module; d) A set of SIMIO processes, see [15], used to manage the pallet movement through the

paths and the network of nodes; e) A FSM control behavior translated into C# code and implemented into SIMIO

custom steps in order to be integrated with the mechanical behavior, see [15]. Specifically for the elements b), the mechanical behavior has been modelled by means of

three nodes which represent the three possible Buffer Zone. Each node associated to an existing Buffer Zone has been set on blocking state or in order to stop the pallet movement. Each time a specific command is set by the FSM then the regarding node let the pallet flow through it so to reach the following node that is to reach the following Buffer Zone. For what concern the elements c), as the pallet reaches a node corresponding to a proximity sensor then the regarding simulated variable is set high, so to allow the FSM changing its state.

Let consider for example the Control Sequence S10 used to move a pallet from BZ8,1 to the BZ8,2. It consists of the following control steps, see Figure 9: A HLCS function starts the Control Sequence (Start_Seq_10); Both the commands to lower the Stacker Crane 2 (Ev_Sc2_D) and to start the Main

Track motor (M_Tr_F) are switched on; After the Stacker Crane 2 has reached the lower position, identified by the sensor

Sx_Sc2_D, then a predefined delay time (Delay_Time) equals to 0.5s elapses; After that delay time the Stacker Crane 2 rises up again by means of the command

Ev_Sc2_D switched off; As the pallet is detected in the BZ8,2 by the regarding proximity switch Sx_P_Ps then a

delay time (Delay_Time) equals to 0.5s elapses after that the Main Track motor command M_Tr_F is switched off.

Figure 9: S10 Finite State Machine

By aggregating the fifteen Transport Module DES models and by modelling in a simplified way the four machines, then the whole plant DES model is obtained, see Figure 10.

Figure 10: The de-manufacturing plant representation of the regarding DES model

It has been validated by comparing its simulation experiments results with the real pilot

plant acquisitions. In particular a specific control algorithm, running on a customized control platform developed in the C++ programming language, has been implemented in order to move a pallet according to a predefined route. By applying such control steps both to the plant DES model and to the real plant LLCS then the simulated and the real pallet movements have

been compared, by also taking into account the time needed to carry on the pallet movements itself.

5. CONCLUSIONS

In this paper the design of a multi-level control system for a de-manufacturing pilot plant has been described, with focus on the implementation of the lower level based on Control Sequences. Such development has allowed to obtain an abstract direct graph plant representation. This mathematical formalism can be used for the implementation of the advanced control policies, to be run at the higher level of the plant automation system. The pallet transport line and the regarding LLCS have been respectively modelled and emulated into a discrete event simulator in order to gain a software tool to be used for the HLCS testing phase. The so obtained integrated DES model has been validated by comparing the simulation experiments results with the measurements acquired from the real pilot plant.

Future works might be focused on different objectives like the development of a decentralized HLCS just based on the plant direct graph model besides the implementation of the LLCS according to different control methods and techniques.

6. REFERENCES [1] M. Bauer, I. K. Craig: Economic assessment of advanced process control – a survey and

framework, Journal of Process Control, Vol. 18, pp. 2-18, 2008. [2] J. Maciejowski: Predictive control with constraints, Prentice Hall, 2002. [3] J.A. Rossiter: Model based predictive control: a practical approach, CRC Press, 2003. [4] J.B. Rawlings, D.Q. Mayne: Model Predictive Control: Theory and Design, Nob Hill

Publishing, 2009. [5] A. Cataldo, A. Perizzato, R. Scattolini: Management of a production cell lubrication

system with model predictive control. Proc. APMS international conference on Advances in production systems, APMS 2014, Ajaccio, France, 20-24 September 2014, Part III, IFIP AICT 440, pp. 131-138, 2014.

[6] Alina N. Tarau, Bart De Schutter, and Hans Hellendoorn, Centralized versus decentralized route choice control in DCV-based baggage handling systems. Proceedings of the 2009 IEEE International Conference on Networking, Sensing and Control, Okayama, Japan, March 26-29, 2009.

[7] R. Scattolini: Architectures for distributed and hierarchical model predictive control – a review, Journal of Process Control, Vol. 19, pp. 723-731, 2009.

[8] R. Scattolini, P. Colaneri, Hierarchical Model Predictive Control. Proceedings of the 46th IEEE Conference on Decision and Control, New Orleans, LA, USA, Dec. 12-14 2007.

[9] Institute of Industrial Technology and Automation (ITIA) – National Research Council (CNR) de-manufacturing pilot plant, Italy, http://www.itia.cnr.it/en/index.php?sez=4#notizia42.

[10] G. Copani, A. Brusaferri, M. Colledani, N. Pedrocchi, M. Sacco, T. Tolio, Integrated De-manufacturing systems as new approach to End-of-Life management of mechatronic devices. 10th Global Conference on Sustainable Manufacturing Towards Implementing Sustainable Manufacturing, Istanbul, Turkey, 2012.

[11] C. Ghezzi, M. Jazayeri, D. Mandrioli, Fundamentals of Software Engineering. Prentice Hall, 1991.

[12] A. Cataldo, R. Scattolini, Modeling and model predictive control of a de-manufacturing plant. IEEE Multi-conference on Systems and Control, Antibes / Nice, 8-10 October 2014.

[13] ICS Triplex ISaGRAF, www.isagraf.com. [14] SIMIO simulation software, www.simio.com [15] A. Cataldo, M. Taisch. B. Stahl, Modeling, simulation and evaluation of Energy

consumption for a manufacturing production line. 39th Annual Conference of IEEE on Industrial Electronics Society IECON 2013, Vienna - Austria, 10-13 November 2013.

7. APPENDIX - TABLES

Sn From To S1 Previous TM Stacker crane 1 S2 Stacker Crane 1 Block pallet Piston S3 Block pallet Piston Next TM S4 External Right Stacker Crane 1 S5 External Left Stacker Crane 1 S6 Stacker Crane 1 External Right S7 Stacker Crane 1 External Left S8 Stacker Crane 1 Next TM (cell 3) S9 Previous TM Stacker Crane 2 S10 Stacker Crane 2 Block pallet Piston S11 Stacker Crane 1 Next TM S12 Previous TM Block pallet Piston S13 Block pallet Piston Stacker Crane 1 S14 External Left Stacker Crane 2 S15 External Right Stacker Crane 2 S16 Stacker Crane 2 External Right S17 Stacker Crane 2 External Left S18 Stacker Crane 1 Previous TM S19 Previous TM Setback Right S20 Setback Right Next TM S21 Block pallet Piston Setback Right S22 Block pallet Piston Setback Left S23 Setback Left Next TM S24 Stacker Crane 1 Next TM (BZ4,1)

S25 External Left Stacker Crane 1 (Setback Right)

S26 Stacker Crane 1 (Setback Right)

External Left (Setback Left)

S27 External Left (Setback Left)

Stacker Crane 1 (Setback Right)

S28 External Right (Setback Right)

Stacker Crane 1 (Setback Left)

S29 External Right Stacker Crane 1 (Setback Left)

S30 Stacker Crane 1 (Setback Left) External Right

S31 Stacker Crane 1 External Left (Setback Left)

S32 Next TM Stacker Crane 1

S33 External Left External Right (through Stacker Crane 2)

S34 External Right External Left (through Stacker Crane 1)

S35 External Left (Setback Left)

External Right (through Stacker Crane 1)

S36 Stacker Crane 1 External Right (M1)

Table 1: The Control Sequences (Sn) definition

Transport Modules Control Sequences

TM1 S2, S3, S36, S19, S26, S27, S28 TM2 S1, S2, S3, S5, S7 TM3 S2, S3, S4, S33 TM4 S1, S5, S7, S24 TM5 S7, S12, S13 TM6 S2, S3, S6, S25 TM7 S1, S2, S3, S5, S18

TM8 S6, S9, S10, S15, S16, S20, S21, S26, S27, S28

TM9 S9, S10, S14, S17, S22, S23, S29, S30, S31, S35

TM10 S7, S12, S13, S34 TM11 S7, S8, S12, S13, S32 TM12 S5, S7, S11, S12, S13 TM13 S5, S7, S11, S12, S13 TM14 S7, S12, S13 TM15 S2, S3, S5

Table 2: The Control Sequences (Sn) related to the regarding Transport Modules (TMi)

Sn Transport Modules S1 TM2, TM4, TM7 S2 TM1, TM2, TM3, TM6, TM7, TM15 S3 TM1, TM2, TM3, TM6, TM7, TM15 S4 TM3 S5 TM2, TM4, TM7, TM12, TM13, TM15 S6 TM6, TM8

S7 TM2, TM4, TM5, TM10, TM11, TM12, TM13, TM14

S8 TM11 S9 TM8, TM9 S10 TM8, TM9 S11 TM12, TM13 S12 TM5, TM10, TM11, TM12, TM13, TM14 S13 TM5, TM10, TM11, TM12, TM13, TM14 S14 TM9 S15 TM8 S16 TM8 S17 TM9 S18 TM7 S19 TM1 S20 TM8 S21 TM8 S22 TM9 S23 TM9 S24 TM4 S25 TM6 S26 TM1, TM8 S27 TM1, TM8 S28 TM1, TM8 S29 TM9 S30 TM9 S31 TM9 S32 TM11 S33 TM3 S34 TM10 S35 TM9 S36 TM1

Table 3: The Transport Modules TMi related to the regarding Control Sequences (Sn)