SciELO - Scientific Electronic Library Online

 
vol.24 issue4Escape in a nonideal electro-mechanical system author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand

Article

Indicators

Related links

Share


Journal of the Brazilian Society of Mechanical Sciences

Print version ISSN 0100-7386

J. Braz. Soc. Mech. Sci. vol.24 no.4 Rio de Janeiro Nov. 2002

http://dx.doi.org/10.1590/S0100-73862002000400015 

Petri Net approach for modelling system integration in intelligent buildings

 

 

P. E. MiyagiI; E. VillaniII; G. D. B. GustinIII; N. MaruyamaIV; D. J. Santos FilhoV

Ipemiyagi@usp.br
IIevillani@usp.br
IIIgladysbg@usp.br
IVmaruyama@usp.br
VDepto. de Eng. Mecatrônica e de Sistemas Mecânicos, Escola Politécnica da Universidade de São Paulo, Av. Prof. Mello Moraes, 2231, 05508-900 São Paulo, SP. Brazil, diolinos@usp.br

 

 


ABSTRACT

In this paper, a Petri Net approach is introduced for modelling and simulation of control strategies in Intelligent Building. In this context, it is claimed that integration with other building systems can be achieved in a more systematic way considering a mechatronic approach (i.e. multidisciplinary concepts applied to the development of systems). The case study is the Ambulatory Building of Medical School Hospital of University of São Paulo. Particularly, the developed methodology is applied to the elevator system and to the HVAC (Heating, Ventilation and Air Conditioning) system. It is shown that using this approach, the control systems could be integrated, improving performance.

Keywords: Petri nets, intelligent buildings, modelling, system integration


 

 

Introduction

The term "Intelligent Building" (IB) was created a few decades ago, probably as a mere commercial slogan. However, in the following years, it achieved a new connotation incorporating engineering concepts for design, construction and operation of buildings [Becker, 1995; Arkin & Paciuk, 1995; Fujie & Mikami, 1991]. Nowadays, there is no unique, well-accepted definition of the so-called "Intelligent Building". A number of definitions have been adopted by different institutes and research groups around the world. In a short overview, the main common points among them are the integration among systems and the use of new information technologies. The purpose of IB is to maximise the productivity of its occupants, to allow an efficient management of resources and to minimise costs.1

There is no doubt about the relevant role of system integration in the fulfilment of IB goals. Among its advantages are the improvement of safety, system control reliability and reduction of operation costs. To fulfil this integration, a Building Management System (BMS) is responsible, among other tasks, for exchanging information among different systems, such as fire control systems, access systems, lighting systems, elevator systems, HVAC (Heating, Ventilation and Air Conditioning) systems, etc. (Fig. 1).

 

 

IB, as man-made systems, has their dynamic mainly characterised by discrete events and discrete states. Examples of discrete events are turning on lights, entering in a room, etc. Examples of discrete states are light on/off, person in/out the room, etc. Therefore, from a management point of view, IB can be considered to some extent a predominantly Discrete Event Dynamic System (DEDS). On the other hand, for some IB systems, it is also necessary to take into account continuous interactions that cannot be represented by discrete rules and should be modelled as continuous variables.

In this context, considering a mechatronic approach where multidisciplinary concepts are combined and applied, we are introducing techniques based on Petri Nets for modelling, analysis and control of buildings systems. The focus is on the integration among them.

The case study adopted is the Ambulatory Building of the Medical School Hospital of University of São Paulo, called PAMB. Particularly, in this paper, we present a methodology developed for modelling and simulation of control strategies of two building systems: the elevator system and the HVAC system. The former illustrates the methodology for pure discrete systems. The latter is a typical example of hybrid system, i.e. where both discrete and continuous aspects are considered.

 

The Petri Net Approach

Petri Net is a graphic and mathematical tool for modelling, analysis and design of DEDS. Among its main advantages are easy graphical interpretation, identifying states and action in a clear way, and the possibility of representing a system dynamic and structure in many levels of detail. Petri Nets can also represent synchronism, asynchronism, concurrence, causality, conflict, and resource sharing.

Among the Petri Net approaches for DEDS, the PFS/MFG methodology [Hasewaga et al., 1988; Miyagi, 1996] is adopted. A Petri Net based model of the system is built by using a top-down approach. Firstly, a conceptual model is obtained by using the PFS (Production Flow Schema) technique [Miyagi, Arata, Moscato, 1997]. Then, the PFS model is refined into a functional model using MFG (Mark Flow Graph) or some specific type of Petri Nets.

The aim of PFS is to identify the activities in a flow of discrete items (information, material) in a high level of abstraction. The PFS models have no dynamic. The components of PFS are activities, which represent modifications on the flow of items, inter-activities, which do not modify the items, and arcs (Fig. 2).

 

 

MFG models can explicitly show the interaction with external components. Contrary to the PFS, the MFG can model the system dynamic by changing the net marking. The MFG components are illustrated in Fig. 3. The detailed definition of MFG and its derivations can be found in [Miyagi, 1996; Santos Filho, 1998; Matsuzaki, 1998].

 

 

The Methodology

Generally, the methodology for modelling and analysis is divided into four main steps (Fig. 4).

 

 

Step 1: System definition

The main activity of this step is to get the necessary information about system and building. All the system requirements should be considered. If necessary, system data for the modelling step should be collected.

Step 2: Strategies definition

According to the information got on step 1, the control strategies are defined.

Step 3: Modelling

An interpreted Petri Net approach is used to build system models by using a top-down methodology. Firstly, a conceptual model is obtained by using the PFS technique. Then, the PFS model is refined into a functional model using MFG or other type of Petri Nets. At this level, the description of the activities of PFS model are preserved and its details are described in a functional level.

It is important to underline that, according to system characteristics, different types of Petri net can be used in this step.

Step 4: Simulation

Once models are built, simulation is performed to provide information about control system behaviour, allowing, for example, the detection of deadlock occurrence, of possible non-desired states, etc.

From simulation data for performance evaluation is also obtained.

Nowadays, a number of tools for Petri Nets simulation and analysis have been developed. A survey about the most important tools could be found in [Störrle, 1998], where the author underlined their main characteristics and their pro and cons.

 

Case Study

The case study adopted here is the Ambulatory Building of the Medical School Hospital of University of São Paulo, which is known as PAMB. Giving an idea of its complexity, PAMB is a building of about 100,000m2, and includes clinics, surgery centre, industrial pharmacy, among others installations. About 2,300 people pass through it every day.

PAMB has been chosen due to the large variety of conditions it offers and due to its complexity. It has discrete and hybrid systems that offer possibility to use different types of Petri nets. The aim of the project is to develop methods to improve its control based on the integration of systems.

Following, the presented methodology is applied to the elevators system [Gustin, 1999a; Gustin, 1999b] and to the HVAC system [Villani, 1999; Villani, 2000].

 

Elevator System

Step 1: System definition

The purpose of the elevator system in PAMB is to transport patients, visitants, doctors, servants, equipment and other kind of load.

The PAMB has 14 elevators for general use and 2 elevators for building maintenance. The elevators are divided in 4 circulation blocks (A, B, C and D).

Nowadays, the PAMB elevators are not interconnected. It is believed that an intelligent management system would improve the performance. The main aspects of this management systems are group control strategies and integration with other building systems.

Step 2: Strategies definition

According to the Brazilian norms for hospitals, the following requirements should be attended:

- The patient elevators that serve more than 4 floors should have collective automatic control.

- At least one elevator should be adequate for fire-fighters procedures.

- The elevator system should have special strategies for fire and blackout situation.

According to the information of Step 1 and to the Brazilian norms, the following strategies have been considered for PAMB:

- Duplex/triplex strategy (the allocation of each landing call is answered by the elevator car which has the more appropriated position).

- Up-peak traffic strategy.

- Down-peak traffic strategy.

- Fire strategy.

- Blackout strategy.

- Preventive maintenance strategy.

- Fault strategy.

Description of each strategy could be found in [Barney, 1985; Elevator World, 1990, 1991].

Step 3: System modelling

For the elevator system, this step is divided into two stages (Fig. 5):

 

 

Stage 1: Structural Modelling

All the system components, the relation among them and the control hierarchic should be identified in the structural model. Basically, the elevator system could be divided into two parts: the control object (elevator cars) and the control system. Fig. 6 displays the structural model for PAMB elevator system.

 

 

Stage 2: Functional Modelling

Each component of the structural model is refined first into a conceptual model and then into a functional model using PFS/MFG methodology.

Naturally, it is impossible to present here all models developed for PAMB. As an example, we present the model of the boxes "Management system for all groups" and "Group D control".

The box "Management system for all groups" sends orders and information that allow, for example, turn on/off elevators, set elevator to independent service procedure or to group service procedure, execute specific strategies for each group, etc. Figure 7 shows the PFS model of it.

 

 

The "Group D control" is divided into two parts: strategies and operations (Fig. 8).

 

 

As an example of strategy modelling, the PFS model for fire strategy is shown in Fig. 9.

 

 

Examples of system operations are: turn on elevator, cancel calls, etc. To illustrate the modelling of system operations, Fig. 10 shows the model of operation "Turn on elevators".

 

 

The model of the control object (cabin and actuator) is detailed in Fig. 11.

 

 

MFG model is developed from PFS model by refinement. Figure 12 shows an example.

 

 

The refinement of all activities results in a model where the interaction among components and the control signals could be easily identified.

Step 4: Simulation

The model simulation provides a way for studying the control system behaviour for different situations, such as blackout, fire, etc. From simulation is also possible to obtain data about system performance by introducing timed components, such as timed transitions where the time associated to the transition represents the event duration. Examples of possible data for strategies, such as duplex/triplex, up/down-peak traffic, are waiting time, service time, etc.

 

HVAC System

Step 1: System definition

The PAMB HVAC system includes heating and cooling. The hot and chilled water is centrally produced by 2 boilers and 8 chillers. The water is then distributed to various fan-coils. Each fan-coil conditions a zone. The zone temperature is controlled by changing the amount of water that passes through the coil. A 3-way valve determines the water flow.

The methodology embodies the HVAC control system, the HVAC equipment and installations and the conditioned environment.

The HVAC control system is divided into two parts: HVAC management system and local control systems. As most of buildings systems, HVAC management system can be considered a DEDS. It is responsible for receiving and sending information to BMS. It is composed by a number of control strategies, which are switched according to the occurrence of discrete events, such as fire alarms, chiller failure, etc. The strategies of HVAC management system are also composed by sequences of discrete events such as turn on a chiller, open a valve, etc. Local control systems are usually composed by Proportional/Proportional-Integral controllers. These local control systems have their states changing continuously and are called Continuous Variables Dynamic Systems (CVDS) [Ho, 1989].

The HVAC equipment, such as chillers, fans, coils and valves, acts on water and air flow by changing its properties, as temperature. Finally, we have to consider the interaction between HVAC and environment, considering the time evolution of environment properties such as temperature and humidity. These system variables are also changing continuously and, as the local control systems, cannot be considered in a DEDS approach.

According to what was presented, it is necessary to represent simultaneously characteristics of DEDS and of CVDS (Fig. 13). Therefore, for the HVAC system, an hybrid approach is adopted. Formal definitions of hybrid systems can be find in [Alla & David, 1998; Antsaklis & Nerode, 1998].

 

 

Step 2: Strategies definition.

The HVAC strategies could be defined for zone equipment or for central equipment such as chillers and boilers. Examples of PAMB strategies for a zone are:

Fire Strategy: is activated and maintained when fire is detected in the zone. The event sequence is:

– the mixing box is set to take 100% of outside air;

– the return fan speed is increased to prevent smoke diffusion and the supply fan is turned on;

Unoccupied Zone Strategy: is activated and maintained when there is nobody in the zone and while fire is not detected in the zone. The event sequence is:

– fans are turned off;

– the mixing box is set to take 0% of outside air and coil valves are closed;

Occupied Zone Strategy: is activated and maintained while there is people in the zone and while fire is not detected in the zone. The event sequence is:

– fans are turned on;

– the mixing box is set to take 60% outside air and coil valves are opened;

Step 3: System Modelling

For the elevator system, this step is divided into two stages (Fig. 14).

 

 

Stage 1: Control Strategies Modelling

As introduced before, the HVAC management system is discrete.

At this stage, PFS model of the strategies is built (Fig. 15a). This model is detailed in a new PFS model (Fig. 15b), in a Petri Net model, or in a mixed PFS/Petri Net model.

 

 

Sequentially, each activity is detailed by using Petri Nets. The Petri Net considered here is Place-Transition [David & Alla, 1994]. Inhibitor and permissive arcs are added to the original Place-Transition Petri Net.

As for the elevator system, activities of PFS model execute operations that change equipment state. Inhibitor and permissive arcs connect strategies and operation.

As an example, Fig. 16 presents the Petri Net model obtained from activity [Move mixing box to 100% of outside air] of Fig. 15. In Fig. 16 the operation is initialised by 2 strategies (transition T1a e T1b).

 

 

Stage 2: HVAC installation and equipment modelling

In this stage, a hybrid approach is necessary to model both discrete equipment state and water/air continuous variables.

For modelling the hybrid system, we extended PFS, initially defined only for DEDS. The continuous flow is considered as flow of infinitely small packets. An activity is defined as a transformation performed on these packets. Firstly, the properties of the flow should be defined. Then, an activity is any modification on these properties along the process.

Figure 17 presents an example of PFS modelling for airflow through zone equipment. Here, the variables to be modelled are flow and temperature.

 

 

Each activity of PFS can be detailed in a new PFS or in a Diferential Predicate-Transition Petri Net (DPT net) [Champagnat, 1998]. A vector of variables is associated with DPT net. Each place is associated with a system of differential equation. These systems represent the evolution of the variables when the place is marked. As an example, Fig. 18 shows the DTP net for the activity [Air in the mixing box], where Tzone, Toutside and Tbox are the air temperature in the zone, outside and just after the mixing box.

 

 

Once all the activities of PFS are modelled, it is possible to connect the DPT nets to the operation models. Once connected, operations authorise/inhibit DPT net transition firing, changing equipment states according to the evolution of management system and to the occurrence of discrete events on other building systems.

The local control system, composed by P and PI controllers, are also modelled using DPT net.

Stage 3: Zone modelling

The model of the conditioned environment is composed by differential equations. The equations determine the evolution of temperature in the room according to HVAC system, to thermal loads introduced across walls, windows, etc., and to thermal loads introduced by equipment, people, lights, etc. The latter is defined by discrete events (light is turned on, someone enter the room) and is modelled using Petri nets.

Step 4: Model Simulation

The model simulation implies in the accomplishment of two kinds of simulators: one of Petri Nets and one of differential equations. The simulators are synchronised by events. Figure 19 introduces the stages of hybrid simulation.

 

 

Firstly, the initial state is defined (the initial marking and initial values of continuous variables). Then it is verified if any transition could be fired. Both Petri net transitions and DPT net transitions should be tested. If possible, a transition is fired. When no more transition could be fired, a numeric simulation of the activated systems of equation is performed. Between every time increment it should be tested if any transition could be fired. The firing of a transition has priority over time evolution.

The simulation data can be used in performance methods such as the PMV (people average vote) and PPD (percentage of people unsatisfied) [Fanger, 1970].

 

Conclusions

In this paper, we proposed a systematic methodology for modelling and simulation of control strategies in Intelligent Buildings systems through a Petri Net approach. To demonstrate the effectiveness of the introduced methodology, a case study was conducted and satisfactory results have been observed in the control systems integration and the performance. The adopted PFS/MFG methodology provides a top-down modelling and turned to be useful for complex systems.

The methodology was successfully applied to the elevator system and the HVAC system. It allows the specification of functional model for the whole system (object and system control), including strategies for integration with other building systems. For the HVAC system, a hybrid approach is introduced in order to consider both discrete and continuous characteristics.

 

Acknowledge

The authors thank the collaboration of FSP-USP, HC-FMUSP, FAPESP, CYTED, CNPq, CAPES and RECOPE/FINEP. We also thank Prof. Emilio C. N. Silva for correcting English grammar errors.

 

References

Alla, H. & David, R, 1998, "Continuous and Hybrid Petri Nets", Journal of Circuits, Systems and Computers, vol.8, n.1.        [ Links ]

Antsaklis, P. J. & Nerode, A., 1998, "Hybrid Control Systems: An Introductory Discussion to the Special Issue", IEEE Transactions on Automatic Control, vol 43, n.4, pp 457-459.        [ Links ]

Arkin, H. & Paciuk, M., 1995, "Service Systems Integration in Intelligent Buildings ", Proceedings of IB/IC Intelligent Buildings Congress, Telaviv, Israel.        [ Links ]

Barney, G. C & Dos Santos, S. M., 1985, Elevator Traffic Analysis, Design and Control, Ellis Horwood Limited, Chichester, England        [ Links ]

Becker, R., 1995, "What is an "Intelligent Building", Proceedings of IB/IC Intelligent Buildings Congress, Telaviv, Israel.        [ Links ]

Champagnat, R., 1998, Supervision des Systèmes Discontinus: Definition d'un Modèle Hybride et Pilotage en Temps-rèel Université Paul Sabatier, Thèse de Doctorat, Toulouse, France.        [ Links ]

David, R & Alla, H., 1994, "Petri Nets for Modeling of Dynamic Systems – A Survey" Automática, v.30, n.2, pp175-201.        [ Links ]

Elevator World, 1990, Educational Package and Reference Library, v. 1, Editor Elevator World Educational Division Mobile, United States.        [ Links ]

Elevator World, 1991, The Guide to Elevatoring, Editor Elevator World Educational Division. Mobile, United States.        [ Links ]

Fanger, P. O., 1970, Thermal Confort, McGraw-Hill, New York, USA.        [ Links ]

Fujie, S. & Mikami, Y.,1991, "Construction Aspects of Intelligent Buildings", IEEE Communications Magazine, vol. 29 n. 4, pp 50-57.        [ Links ]

Gomez, L.F., 1997, Redes de Petri reactivas e hierárquicas - integração de formalismos no projecto de sistemas reactivos de tempo-real, Doutorado Tese, Universidade Nova de Lisboa, Lisboa, Portugal.         [ Links ]

Gustin, G. B., 1999a, "Modelagem de Sistemas de elevadores no contexto de edifícios inteligentes através de Redes de Petri interpretadas", Congresso Brasileiro de Engenharia Mecânica (COBEM'99), Águas de Lindóia, Brazil.        [ Links ]

Gustin, G. B., 1999b, Aplicação de redes de Petri interpretadas na modelagem de sistemas de elevadores em edifícios inteligentes, Dissertação de mestrado, Universidade de São Paulo, São Paulo, Brazil.        [ Links ]

Hasegawa, K.; Takahashi, K.; Miyagi, P. E., 1988, "Application of the Mark Flow Graph to represent discrete event production systems and system control", Transactions of the Society of Instrument and Control Engineers, vol. 24, n.1, pp.67-75.         [ Links ]

Ho, Y.C., 1989, "Scanning the issue - Dynamics of discrete event systems", Proceedings of IEEE, vol.77, n.1, pp.3-6, 1989.        [ Links ]

Matsusaki, C. T. M.,1998, Redes F-MFG (Functional Mark Flow Graph) e sua aplicação no projeto de sistemas antropocêntricos, Dissertação de Mestrado, Escola Politecnica da Universidade de São Paulo, São Paulo, Brazil.        [ Links ]

Miyagi, P.E., 1996, Controle Programável, Editora Edgard Blücher, São Paulo, Brazil.        [ Links ]

Miyagi, P.E., Arata, W. M., Moscato, L. A., 1997, Analysis of Manufacturing Systems – Application on PFS model based analysis of Manufacturing System for Performance Assessment, Revista Brasileira de Ciências Mecânicas, vol. XIX, n.1, pp58-71.        [ Links ]

Santos, D., 1998, Controle de sistemas antropocêntricos de produção baseado em Redes de Petri interpretadas, Tese de doutorado, Escola Politecnica da Universidade de São Paulo, São Paulo, Brazil.        [ Links ]

Störrle, H., 1998, "An Evaluation of High End Tools for Petri-Nets", In: http://www.daimi.au.dk/PetriNets/.        [ Links ]

Villani, E., 1999, "Modelagem do Sistema de Controle das Condições de Conforto Térmico em Edifícios Inteligentes", Congresso Brasileiro de Engenharia Mecânica (COBEM'99), Águas de Lindóia, Brazil.        [ Links ]

Villani, E., 2000, Abordagem Híbrida para Modelagem de Sistemas de Ar Condicionado em Edifícios Inteligentes, Dissertação de mestrado, Universidade de São Paulo, São Paulo, Brazil.        [ Links ]

 

 

Presented at COBEM 99 – 15th Brazilian Congress of Mechanical Engineering, 22-26 November 1999, São Paulo, SP. Brazil.
Technical Editor: José Roberto F. Arruda.