You are on page 1of 28

RESEARCH ARTICLE

A control architecture for continuous production processes based on


I 4.0: Application to Water Supply Systems

Edgar Chacóna , Luis Alberto Cruz Salazarb , Jose-Fernando Jiménezc and Juan
Cardilloa
a
Universidad de Los Andes, Venezuela; b Technical University of Munich, Germany; c

Pontificia Universidad Javeriana, Colombia

ARTICLE HISTORY
Compiled October 1, 2019

ABSTRACT
The fourth industrial revolution, or Industry 4.0 (I4.0), brings together new disrup-
tive technologies, increasing the productivity of factories of the future. Certainly, the
control of production processes is fast becoming a key driver for manufacturing oper-
ations. Manufacturing control systems have recently been developed for distributed
or semi-heterarchical architectures (e.g., Holonic Manufacturing Systems “HMS”)
improving both global efficiency and the reactiveness of manufacturing operations.
Still, previous studies and applications have not dealt with continuous production
processes, such as applications for water treatment, oil refining, or electric power
plants. The complexity of continuous production is that a single fault can degrade
extensively, and can even cause a breakdown of production. Therefore, this paper
proposes an HMS architecture for the control of continuous production processes.
For this version of an HMS, a Holonic Production Unit (HPU) is created as a holon
unit depicting resources in a continuous process. This unit is capable of detecting
events within the environment, evaluating several courses of actions, and chang-
ing the parameters aligned to a mission. The proposed approach was tested using
a simulated model of a water treatment plant, considering three scenarios: normal
condition and two disrupted scenarios (the unexpected increase of demand and water
quality degradation). The experiments described in this paper were conducted using
an agent-based software, where the communication and decision-making features al-
low simulation of HPU. The results suggest that the construction of a holarchy with
heterogeneous holons is potentially able to fulfill I4.0 requirements for continuous
production processes.

KEYWORDS
Continuous processes; Holonic Manufacturing Systems; HMS and Industry 4.0;
Intelligent industrial automation; Production unit

1. Introduction

Water Supply System (WSS) are part of critical infrastructure systems, which are
characterized by widely distributed systems forming distribution networks to satisfy
the population of a geographical region, in particular cities. The use of technology is
essential to manage those kind of systems (Nam and Pardo 2011). Those networks
have nodes that are autonomous having control procedures according the function of
the node, but each node must cooperate whit other nodes in order to accomplish a

CONTACT Edgar Chacón. Email: echacon@ula.ve


global objective. Those systems are considered as a System of Systems (Joannou et al.
2019).
The automation of those systems implies the use of sensors, local controllers to con-
trol the physical infrastructure (pumps, valves, pipes, tanks, dams, purification plants
and others) and software systems to perform supervision, and optimization of the op-
erations. The “digital transformation concept” developed for industry can be applied
to the critical infrastructure systems; that concept includes Big data, deep learning,
Cyber-Physical Systems (CPS), Cyber-Physical Production Systems (CPPS) and the
technological aspects associated to Industry 4.0 (I 4.0) specially in manufacturing.
Manufacturing systems are strictly challenged on a daily basis (Leitao et al. 2016).
To satisfy these complex needs, several architectures have been proposed to cope with
the control of manufacturing operations (Borangiu et al. 2019). In general, these ar-
chitectures have been classified into hierarchical, heterarchical and semi-heterarchical
architectures (Trentesaux 2009), based on the behavior, structure and dynamics ex-
pected during the manufacturing execution (Jimenez et al. 2017; Jimenez Gordillo
2017). Since recently, these architectures are being composed by self-organizing and
intelligent holons in a holarchy organization (Blanc, Demongodin, and Castagna 2008;
Barbosa et al. 2015; Dias-Ferreira et al. 2018). The Holonic approach for manufactur-
ing control is called Holonic Manufacturing Systems (HMS).
HMS aims to decentralize the manufacturing tasks into individual decisional enti-
ties (i.e. holons) for featuring autonomous, cooperative and responsiveness behavior
within manufacturing operations (Pujo, Broissin, and Ounnar 2009). Therefore, the
development of the holon concept and the holarchy organization have become a central
issue responding to the efficiency and the reactive manufacturing challenges.
Besides some interesting contributions (Chokshi and McFarlane 2008b; Indriago
et al. 2014, 2015; Tippett and Bao 2015; Bloch et al. 2017), the HMS research to date
has tended to focus on intermittent and discrete processes (Cruz S. 2018) rather than
continuous production processes–e.g. chemical processes, water-supply processes or oil
refining processes (Chokshi and McFarlane 2008a). In particular, the control policies
in continuous processes are established from the beginning and they are expected
to remain fixed throughout the entire execution. The main challenge of this kind of
configuration is the manufacturing system’s lack of adaptability and responsiveness,
hindering performance improvement or reaction to possible uncertain perturbations.
Consequently, the inherent characteristics of these processes and the possible need of
having adaptable configurations in manufacturing control, indicate the need to extend
the holon and holarchy notion to the continuous production process.
The WSS are continuous processes with jumps that are necessary to respond to
changes in water supply and consumption, restrictions on pumps and valves, and also
for failures in the network and equipment, which impose changes in the system con-
figuration (Kim and Smith 2004; Wagner, Shamir, and Marks 1988). Each component
of the entire system can be seen as a hybrid dynamic system (continuous and discrete
dynamic), which is autonomous and cooperative. The global system that results from
the interaction of all the components of the system is coordinated in a distributed way
using the holonic vision.
In detail, the implementation of the proposed architecture is divided into three
levels: The plant, associated with physical processes (mechanical, chemical, biologi-
cal); operation management level or image level, which is dedicated to optimize the
production process and the enterprise level. The plant level emulates a traditional
architecture based on PLC (Programmable logic controller) and its SCADA (Supervi-
sory Control and Data Acquisition). Therefore, this level contains a regulatory control

2
task and a local supervisor work together for monitoring the system sensors (i.e. water
storage, tank level, water treatment rate, among others). The Image level, similarly to
the digital twin concept, contains the HPUs that supervise negotiate, retrieve/store
data, and interact with other HPUs. These contribute to maintaining the state and
evolution of the physical process. The enterprise level, as a non-real time regulator,
is oriented to evaluate the execution, transfers information to the enterprise systems
and accept/reject the assigned objectives and to establish cooperation with customers
and suppliers.
The paper proposes a holonic architecture that integrates the physical level with the
enterprise level allowing the execution of processes associated to the operation manage-
ment: configuration, reconfiguration of physical processes as results of the scheduling
activities, and the supervision tasks to ensure the established production goals.
The document is organized as follows: a brief introduction highlighting the impor-
tance of water supply management systems and the automation trends followed for
this type of systems, section 2 describes the structure and its functions in water supply
systems, the aspects of process integration, and the information technology architec-
ture used. Section 3 describes an automation proposal in accordance with the trends in
manufacturing automation, and which allows the integral automation of the different
processes in continuous production systems. A methodology for the implementation
of the HPU concept is given in section 4 and applies to the case of automation of
water supply companies. Finally, section 5 highlights the benefits of the proposed
architecture.

2. Water Supply Systems: a brief description

The urban water has a cycle that covers from the water catching process until the devo-
lution to the environment, and by evaporation returns as surface water or groundwater
to the rivers and aquifers Marsalek et al. (2006) as is shown in Figure 1

Figure 1. Urban water Cycle (From Marsalek et al. (2006))

A WSS that make part of the Urban water cycle, it is collection of connected
units for collecting, storing, transporting, purifying, distributing drinking water to the
population of a geographical area, particularly cities, and it has an organization that
allows to accomplish the objectives in supply water to the population, maintaining
economic viability and sustainability with the environment.

3
The value cycle of the whole process is given by the Figure 2, where the product
model correspond to the central chain. The first stage of the product model is as-
sociated with the water catching from sources and it is specific to the water source.
Transportation of the raw water (surface or ground water) is common to the raw wa-
ter and the raw water treatment depends on the water quality and the procedures
are different. The quality of the water can be improved by mixing groundwater with
surface water. Finally the rest of stages are independent of the water sources.

Logistics and
Maintenance

WW
Sources Catch Transport Purification Distribution WW Recover Transport Treatment Discharge

Billing
Service Requirement Connection Measurement and
Collection

Estimate and
Customer Projections

Finance
Human Resource
Planification
Legal Consultancy

Communication and Marketing

Figure 2. Value chain for the water cycle and its associated process

The WSS may have several sources (superficial and groundwater), several purifica-
tion units, a main network that transports water to storage tanks for final distribution
and the distribution network. See Figure 3.

Figure 3. Water Supply System

4
2.1. The WSS enterprise organization
Figure 4 shows the functional organization for a hydrological enterprise. Those func-
tions includes administrative management functions, responsible to accomplish the
objectives set by the administration; operative management that is related to to the
planning of the physical operations and the physical operations to intake, transport,
purification, distribution of tap water, and the sewage collection, treatment and return
to the environment.
Production state

General Directives
Management Operational condition Plans
Economical state
Planing Operational restrictions
A1 Engineering &
Development
A3
A4
Major
requirements

Materials
Administration Water consumption
Requirements Operational conditions
Maintenance
A2 Restrictions Customers
Management
A7
A6
economical information
Operational restrictions
Consumers satisfaction
Expected demand
Raw water Operation Fails Drinking Water

A5
Facilities
Maintenance labor

Figure 4. IDEF 0 for a hydrological enterprise

To performs those functions, most of hydrological companies have an infrastructure


of Information & Communication Technologies and Operational Technologies based
on the classical CIM approach as it is shown in Figure 5. New analysis-based func-
tionalities (analytics) are incorporated to improve leak detection; also, functions to
support the management of interactions between maintenance and operation person-
nel to establish configurations when maintenance work is performed and to establishes
operation sequences when new network configurations are needed.
Functions expected in each level are:
• Business Processes level
◦ Long term planning. Studies of water sources, population growth, new in-
dustries, changes in water use. Development of new facilities
◦ Supply chain management.
◦ Customer management, billing.
◦ Finances.
• Optimization level for operations
◦ Short and medium term planning of operations. Forecasting
◦ Network analysis, Event management, Optimization of the operation se-
quences based on the condition of the facilities. Establishing new configu-
rations to satisfy population needs.
◦ Maintenance
◦ Customers support for new connections. Consumption measurement.
◦ Fixing of controllers parameters. Sequencing of operations. Supply of the
service to the population.
◦ Event detection.

5
ERP Server CRM Server SCM Server Other Servers

Business Processes level


Enterprise Service Bus
Firewall

Optimization level Planif. / Prog. Monit. / Super Prod & Proc Mod.

Non strict RT. Coordination Production Service Bus

SCADA Server RT−DB Quality lab Resources/Maint.

Plant level Real Time Control Area Network

PLC PLC PLC

IIoT Dev. IIoT Dev.

Process Process Process

Information & Control Flow


ERP: Enterprise Resource Planing
CRM Customer Relation Management
SCM Supply Chain Management
PLC Programable Logic Controller
RT−DB Real Time Data Base

Figure 5. Classical IT / OT Architecture

2.2. Information integration for WSS: classic approach


The decision-making process at different levels depends on the quality of the informa-
tion and its availability. Currently, most organizations do not have a standardization
of information, and the decision-making process becomes complicated.
In the case of operations management, attempts have been made to integrate the
different sources to provide at the level of optimization the information necessary for
the management of the operation. These sources are shown below.

• Automation systems
◦ SCADA Systems: Flows, levels, pressure, consumption of pumping systems.
◦ Results of laboratory analysis for water quality determination.
• New technologies (IoT) and enterprise integration.
◦ Manual consumption measurements associated with billing systems.
◦ New technologies for leak detection (Acoustic, magnetic sensors) BenSaleh
et al. (2013); Sun et al. (2011)
◦ Failure detection systems in pumping equipment.
◦ System user information about low pressure, water leaks, supply failures.
(Usually using a call center)
• Process and infrastructure description and equipment information (Engineering
systems).
◦ Network modelling systems. Geographical Information Systems
◦ Maintenance systems.
◦ Engineering drawings.

Operation management implies having network models, mainly generated by the


design department, of behavioural models that can allow valid configurations to be
established, using real values of the infrastructure condition.
In the case of serious failures, the system must allow the establishment of emergency
configurations that make possible the supply of water to priority establishments, such
as hospitals.

6
The integration of all systems, built in isolation, is very difficult and in most cases
it is not possible to obtain. This is why the holistic vision of industry 4.0 gives a way
to the necessary integration in water supply systems.

3. Industry 4.0 concept & the Holonic Production Unit concept

For existing automation systems, it is essential to respond quickly to disturbances and


maintain a stable state for efficient use of available resources operation (e.g. real-time
execution of industrial controllers). These facts promoted above are making sure that
manufacturing systems are present within the dynamic and competitive global mar-
ket. Therefore, from the origins of manufacturing systems until the recent Industry
4.0 (I4.0)) paradigm, their control has some similarities with cognitive characteristics
of human behavior have been encountered (Galán et al. 2000). Indeed, contempo-
rary manufacturing systems show a new generation of approaches or paradigms trying
to meet new demands from different perspectives: Multi-agent Systems Trentesaux
(2009); Leitao et al. (2016); Cruz S. et al. (2018); Leitão and Karnouskos (2015);
Lüder et al. (2017), Services Oriented Architecture for Manufacturing Garcı́a Valls,
López, and Villar (2013); Gamboa Quintanilla et al. (2013), Holonic Manufacturing
Systems HMS Leitão and Restivo (2008); Hsieh (2009); Borangiu et al. (2014); Bar-
bosa (2015), HMS for Process Industry Chacón, Besembel, and Hennet (2004); Chokshi
and McFarlane (2008a); Chacón et al. (2009); Zapata (2011); Indriago et al. (2016),
Cyber-Physical Systems CPS Lee, Bagheri, and Kao (2015); Monostori et al. (2016);
Monostori (2018) among others.
As a result, the need for application of standards requests is increasing (Cruz
and Vogel-Heuser 2017) and the development of intelligent manufacturing systems
(IMS) (Dumitrache and Caramihai 2010). From IMS, the distributed manufactur-
ing paradigms (non-hierarchical systems), which are introducing multi-agent system
(MAS) (Thomas, Borangiu, and Trentesaux 2017) and HMS (Cardin et al. 2017) also
arise and gain importance (Indriago et al. 2016; Simón-Marmolejo et al. 2017). In
(Arboleda C. 2011; Barbosa et al. 2015; Chacón, Carrasco, and Cardillo 2012; Pala-
cio B. 2013; Salazar V. 2009; Zapata 2011) there exist academic and industrial re-
searches, that demonstrate an increasing interest with HMS, regarding I4.0 require-
ments (Cruz S., Rojas A., and Chacón 2019). The I4.0 generates the Cyber Physical
Production Systems (CPPS) concept and their validation of multiples architectures
(Lu, Morris, and Frechette 2016; Ribeiro and Hochwallner 2018), following the future
needs of the automation (Cruz S. and Rojas A. 2014; Dai and Vyatkin 2013).

3.1. Requirements regarding Intelligent Manufacturing Systems for I4.0


For HMS to be aligned (Cruz and Vogel-Heuser 2017; Ribeiro and Hochwallner 2018;
Vogel-Heuser et al. 2014), there are five essential requirements. These are fundamen-
tal properties aligned with the I4.0 paradigm based on the CPPS concept. There is a
list (Cruz and Vogel-Heuser 2017) for CPPS Minimal Conditions identified and they
are grouped into four main categories: CPPS minimal conditions (Req1), Smart char-
acteristics attributes (Req2), Formalized Modelling Terms (Req3), and Systems and
Human integration needs (Req4). This contribution adds another requirement regard-
ing the Reference Architectural Model Industrie 4.0 “RAMI 4.0” (Platform Industrie
4.0 2018). Following are the summary of these requirements, adapted from Cruz and
Vogel-Heuser (2017):

7
3.1.1. CPPS Minimal Conditions (Req1)
Major I4.0 studies have been conducted to investigate the underlying conditions and
determinate least characteristics for CPPS. Besides, these fundamental properties
could be classified into four main items: i) Independent architecture model, which
implies that modules are simple to be integrated with open architecture and platform
for independent implementations. ii) An Open communication protocol for Industrial
Internet of Things (I2 oT), which is capable of being easy and quick to switch between
open networks. iii) All levels of automation for ISA 95 are available depending on the
scenarios in which the CPPS will be applied. iv) The usability and adaptability of the
system for future products, which is called Smart Products.

3.1.2. Smart characteristics attributes (Req2)


Previous works showed that it is elementary to investigate the conditions to discover
“intelligent” characteristics for CPPS, and their fundamental properties could be sum-
marized into four main items: i) Agents’ Autonomy and the updated processing ex-
perience are achieved by deducing behaviors of the CPPS based on goals. ii) An open
communication protocol for the I2 oT concept that is capable of being easy and quick
to switch between open networks. iii) Cooperation and developing mutually acceptable
goals. iv) Pro-activity, which means that the agent is capable of achieving his assigned
goal.

3.1.3. Formalized Modelling Terms (Req3)


It is necessary for innovative approaches to abstractions (formalisms) and architec-
tures to enable control, communication and computing integration. For this reason,
Formalized Modelling Terms requirement could help the rapid design and implemen-
tation of CPPS, based on three items: i) Modelling CPPS using standard language
even the fundamental elements for the vertical expansion (between SCADA, MES and
ERP levels). ii) Different degree of abstraction for applying the model in CPPS. iii)
Implementation of CPPS through Integrated Development Environment (IDE) and
platforms.

3.1.4. Systems and Human Integration Needs (Req4)


This is a significant requirement since CPPS consider the human behaviour into the
control systems. It is categorized into three items. i) Open systems to different systems
domain (e.g. smart grid, automobile manufacturing, oil & gas). ii) Hybrid topologies to
enlarge or downsize the production system. iii) Social norms regarding human factors.

3.1.5. RAMI 4.0 architecture features (Req5)


There are crucial requirements for the I4.0 components concept (Platform Industrie 4.0
2018). RAMI 4.0 indicates that the I4.0 components shall support various engineering
disciplines with system boundary between three axes: i) The “Hierarchy levels” axis
for assigning functional models to specific levels of hierarchy based on the IEC 62264-
1 (ISA – 95) and IEC 61512-1 (ISA 88) standards. ii) The six “Layers” axis that
represents all relevant information to the role of the asset (cyber and physical resources
of the company). iii) The “Life-cycle” axis represents the lifetime of the assets and
their value-added process.

8
4. Holonic Production Unit architecture

Process industry, oil & gas production, energy generation systems, energy and water
distribution systems, among others, are systems that must guarantee a continuous
flow of material to satisfy production demand. Those systems are constituted by a set
of connected units (facilities, utilities, process units) forming a production network.
Each production unit has its own behaviour and the global behaviour comes from the
composition of local behaviours. Pipes give the connection between units where the
material flow is continuous. Product flows in pipes is bulk and the material streams are
transformed at the process units, adding value to the material until reaching the desired
product. We will call production units into each process unit and the transportation
systems. See Figure 6.
Eq 4 Eq 6
Eq 2 Prod.

Eq 5
Prod.
Eq 1 Eq 3
Row Mate.

Figure 6. Physical components (equipment) and the possible interconnections

To build an integrated automation approach is necessary to have models that cover:


Production Models (Product Models and Process Models for each stage of the Product
Model). Topological Models is the layout of the plant that describes the equipment
(production units) and its hierarchy and the interconnections among the production
units; an Information Exchange Model (IEM) represents the data flow from the plant
floor to the decision layers and a Plant Information Model (PIM)) (Pérez et al. 2015).
The HPU architecture covers those aspects, and satisfies Req3. A methodology is used
to obtain the models and project on the automation architecture.
The components of the HPU are given in Fig. 7. Those components have the knowl-
edge necessary to make decisions:
HPU

Mission

HPUSupervisor

Resource Engineering

Figure 7. UML class diagram of the HPU components

Knowledge based production is grouped into two elements, 1) the product model
that describes the set of services provided by intelligent resources to obtain a product
and its order of execution including the formula and 2) the model of the process, which
is the procedure used by an HPU to provide a service. Resources can be internal to the
HPU or can be another HPU. The Engineering Holon (EH) manages the knowledge
associated to the production.
Mission Holon (MH), similar to the Order Holon in PROSA, has the objective

9
to be reached or maintained by a production unit during a period, it also has the
information about the fulfillment of the mission. Goals and State of the Mission are
part of the MH.
Resource Holon (RH) is the central element in the architecture used. RH man-
ages its internal resources, and executes the physical processes necessary to meet the
production objective. RH negotiates with other RH missions, establishes production
commitments, calculates the supervisor to be used to monitor, and control the mis-
sion. An HPU can be a resource for another HPU more generic as is shown in Fig.
7. To manage the internal processes the HPU is partitioned in several resource layers
as is shown in Fig. 8. The upper resource layers perform the negotiation with other
HPU. It evaluates the possibilities to accomplish its part of the global goal and send
an expected behavior to evaluate the whole behavior. If the composition of the system
is considered viable, an agreement is achieved.

HPU
Mission
Knowledge Process image

Supervisor Res. Image

Supervision & Coordination

Local control Supervisor

HPU
Equipment &
Process Equip. & Proc

Figure 8. Resource layers in the HPU

Supervisor Holon (SH) performs management activities of resources. The planning


and programming activities start at the reception of a production goal from the busi-
ness layer. It constructs an accessibility tree by composing the an abstraction of the
behavior of the physical processes, and generates the desired path for each resource
and a “distributed” supervisor for the “distributed” resources, as shown (Chacon et al.
2012). “Local Supervisors” determine the regulators according to the state of the phys-
ical process and the desired trajectory when events happen; it commutes to a regulator
according to the state of the production process.
MH and SH are The holon mission and the holon supervisor are strongly coupled,
since for each mission a supervisor is established. The supervisor has several layers
of control as shown in Fig. 9. To achieve the autonomy, each Production Unit must
have models that stores the knowledge of the plant (behaviour, interconnections and
controllers) that interacts among them to achieve the global control of the unit. Then
the control layers that have the production unit are described below:

Regulation layer. It ensures the local control by a set of local controllers for each
operation mode.
Supervision layer. Detects events in an operation mode, to determine if a jump
exists, and decides the new controller for the new operation region according to
the transition function.
Scheduling & Optimization layer. It establishes the desired sequences to com-

10
Objectives

Process Ressource
Scheduler
Image

Supervisor Σ

E. D E. D

Controller
Process Var

Act S1 S2
Inputs Outputs
Equipment + Process

E.D. Event Detection


S : Sensor
Act. Actuator

Figure 9. Control layers in a production unit

plete a production goal, by selecting among possible configurations the optimal


one.

4.1. Methodology to implement the HPU


The proposed methodological approach shown here, it is structured based on HPU
requirements and the preliminary requirements for I4.0 (See Section 3.1). It takes into
account Product & Process models that includes control and supervision laws, Produc-
tion Units hierarchies and interconnections among them (Topology of the production
network) and the Information Technology and the Operation Technology that support
models, information exchange and decision systems.
The physical system, as is shown in Fig. 6, is composed of several units, where each
unit performs the necessary tasks in order to accomplish a service. The concept of
manufacturing service is used to facilitate the description of the product models (Di-
etrich, Kirn, and Sugumaran 2007; Colombo and Karnouskos 2009; Karnouskos et al.
2012; Gamboa Quintanilla et al. 2013, 2015). In this vision each units is autonomous,
that is, it is able to establish its own goals, supervise and control its evolution. Each
unit establish its goals by a negotiation process among the others units in order to
achieve a global goal.
In continuous production process, each unit that have autonomy can be see as CPPS
Bloch et al. (2017). In Fig. 10 it is given a relation with the CPS 5C level architecture
proposed in Lee, Bagheri, and Kao (2015).

4.1.1. Dynamics of the physical process and its supervision


The physical processes that occur within a unit can be continuous, batch or discrete.
The ability to build models that describes their behavior allows having control mecha-
nisms that ensures the goals established for a unit. The behavior of the process depends
on the equipment capacities and condition, the raw material, energy, etc. Continuous
processes are expected to be “stationary”, but they have several operation modes that

11
Objectives

Configuration
Process Ressource
Scheduler Level
Image

Cognition
Level

Supervisor Σ

Cyber Level
E. D E. D

Controller Data to Information


Process Var Conversion Level

Act S1 S2
Inputs Outputs
Equipment + Process Smart Connection Level

5C architecture for implementation


E.D. Event Detection of Cyber−Physical Systems
S : Sensor
Act. Actuator

Figure 10. The HPU in relation to the 5 C CPS level. 5C CPS is taked from Lee, Bagheri, and Kao (2015)

result from changes on inputs or outputs, or changes on the equipment condition. Hy-
brid Systems are formal techniques that allow describing the behavior of those kinds of
processes (Antsaklis, Koutsoukos, and Zaytoon 1998; Antsaklis and Koutsoukos 2003;
Branicky, Borkar, and Mitter 1998; Lemmon, He, and Markovsky 1999; Van Der Schaft
and Schumacher 2000). Similarly, a batch process are characterized by sequences of
operation modes that allows accomplishing a production task (Alvarez et al. 1999;
Lennartson, Egardt, and Tittus 1994; Lennartson et al. 2016; Méndez et al. 2006;
Tittus and Lennartson 1999).
Then, a production system has two kinds of dynamics: continuous dynamics that
describes mechanical, chemical or biological rules and a set of logical rules discrete
dynamics, that commutes the continuous dynamic and are fired by events. The con-
tinuous dynamics are represented by a set of differential equations and the discrete by
Discrete Event Systems (Cassandras and Lafortune 2009).
A Hybrid System is defined in several ways: as a Phase Transition Systems (Manna
and Pnueli 1993), as a Hybrid Automata (Alur et al. 2000; Branicky, Borkar, and
Mitter 1998; Henzinger 2000; Lygeros 2004), by means of Petri Nets (DAVID and
ALLA 2001; David and Alla 2005). We use the definition given in (Branicky 1996) for
a controlled general hybrid dynamical system.
The evolution of the system is given by a concatenation of continuously controlled
dynamics and jumps as is shown by equation 1.

Φk (. . . (Φ1 (Φ0 (x0 , (t0 , t− + − + −


1 )), (t1 , t2 )) . . .), (tk , tk+1 )) (1)

A jump implies the change of a continuous dynamic, and the jumps can be spon-
taneous or forced. A spontaneous jump appears when the system reaches a condition,
desired or not. In that condition, the system goes into a new stage, and the controller
must be commuted to one that satisfies the condition of control and achieved the
desired goal.

4.2. Methodology to implement the HPU


The implementation of the holon production unit implies: i) the definition of the phys-
ical waste models and the equipment hierarchy, ii) the establishment of the product
route for the production order and the definition of a generation scheme of the super-
visors for the production orders and iii) the deployment of the modules on the IT /

12
OT infrastructure available in the company.

4.2.1. Defining the HPU


To achieve the implementation of the HPU concept, the following steps are carried
out:
• Definition of the production units. To do this task, it is necessary to decompose
the system into strongly coupled subsystems. The following are taken into ac-
count: the geographical distribution, the material return flows and, the type of
processes carried out by each subsystem. The industry value chain guides the
initial approach.
• Construction of product and process models. Describe the products in a similar
way to the ISA–95 reference, associating the stages of the product to manufac-
turing services provided by the HPU. The internal procedures in each HPU must
include steps, events, and control laws in each step and / or mode of operation.
• Description of the equipment hierarchy within the production unit. Establish-
ment of equipment capabilities and the hierarchy of equipment within each pro-
duction unit. Parameters to be measured, expected performance. It is established
in a similar way to the ISA–95 reference.
• Description of the topological model of the plant. The units have physical ports
through which different materials and energy circulate. Some ports are contin-
uous flow; while others are associated with moving batches of product. The
topological map represents the possible connections and the products or energy
that circulate on these connections.
• Description of materials. The materials, their physical characteristics and the
laboratory tests to verify their quality must be described in order to validate the
workflows in the production processes.
• The relation among models, HPU, materials. Enumeration of manufacturing ser-
vices (including laboratory tests) provided by a production unit and the manu-
facturing services necessary for a product, materials that circulate through the
ports, location of the equipment in a unit.

4.2.2. Mapping the HPU concept over the OT / IT Architecture


The functions of the HPU must be projected on the platform of technologies of op-
eration (OT) and information technologies (IT) available in the plant. For a classic
infrastructure, like the one shown in Fig. 5, the regulation functions go directly in
the PLC, the supervision and coordination functions in the equipment in the non-
strict real-time layer, and the programming and planning functions in the information
technology layer, which store knowledge and images (digital copies) of plant processes
and equipment. These images of the process are the basis for determining the state of
the system and generating the scheduling of plant floor activities and determine local
supervisors and coordination mechanisms amongst the units.
To implement supervisors, it is necessary to have engines that receive a supervi-
sion model and execute the supervision tasks defined for that model. In our case, the
motor used is based on a meta-model described using Petri nets. At the lower level,
events at the plant floor are detected directly by sensors and/or by detection mecha-
nisms regarding the operation regions. Other plant events come from the equipment
monitoring. Other plant events come from the equipment monitoring. Supervisor re-
ceives these events to determine the following operation mode. External conditions are

13
managed similarly. The occurrence of the events updates the images.
All the events defined to describe the behavior of the physical system must have an
associated sensor or a detection mechanism based on the measurements of the system.
This allows to reconstruct the discrete dynamics, and to have an updated image that
can be used to reprogram the system.
The functions of optimization and production programming are deployed using in-
frastructure information technologies. Servers oriented to maintain the HPU informa-
tion, maintain the updated image of the process through update mechanisms by events.
Plant floor events travel to the management level of the HPU and a motor updates
the status. If the state does not correspond to the expected state, the reprogramming
mechanisms are triggered.
The product models and process models are maintained on specialized servers, which
can be updated by the engineering and development staff. The relationships between
the resource holons, and the product holons are similar to those proposed by McFarlane
and Bussmann (2000, 2003).

4.3. Contribution regarding the HMS paradigm


The use of behavioral models facilitates the vertical integration between planning
processes and execution processes. The supervisor through the detection of events
knows the evolution of the physical processes. The event detection mechanisms are
implemented in the plant floor equipment and are sent to the supervisor by those
devices.
This architecture is reactive, by adapting to failure events in the floor, such as the
failure of a machine, or inputs for which the procedure is not optimal, and it can be
mapped over different OT/IT architectures. Holarchies are created and managed at
the architectural image level and are created to satisfy a production order. The created
holarchy to satisfy the global order is compound by HPUs that are resources selected
for the order (in the holarchy). Each resource has a local order that ensures the global
order. The global supervisor generates goals for internal holons in the same way as
HPUs for their internal resources.
The holonic paradigm facilitates the tasks of supervision and planning of produc-
tion in continuous processes, as well as batch processes, by allowing to establish the
production objectives in an agile way, taking into account the state of the process
and condition of resources and react quickly to changes in the physical process. Its
implementation requires having a discrete abstraction of the continuous evolution of
the physical process, and of mechanisms that allow determining the changes of modes
of operation, and mechanisms of commutation of laws of control for each mode of
operation.
At the optimization level (image level) to perform the monitoring of operations, the
events captured on the plant floor must be transferred immediately, to maintain the
image of the process in real time and the condition of the resources. In this way, the
process and equipment images are updated, and the decisions are performed, according
to the state of the physical process.

4.4. Design pattern for the Holon Resource


The HMS architecture presented has as a central element, the HPU, that must have
the ability to control, supervise and plan its tasks by working cooperatively, when the

14
tasks are performed by several HPUs. To describe an HPU, we must then consider the
holonic components of it, and its scheme of interaction with the other units. Finally,
establish how the functions of the HPU will be displayed on the technology that is
available in the company.
As shown at the beginning of this section, the procedure for the implementation
follows a pattern, which is based on the definition of the production services provided
by the HPU. The services will be provided through internal procedures, which are
modeled as Discrete Event Systems and that can be used externally to calculate com-
pound procedures that allow obtaining a product through the cooperation of several
HPUs.
The models are used to establish supervisors and coordinators. For control of the
continuous part, the controllers must be determined in advance for each mode of
operation. Supervisors switch between controllers using real–time knowledge of the
current operating mode.
Events must be detected on the floor of the plant, and an efficient transport mecha-
nism must be implemented, so that the events reach the supervisor - coordinator, and
the image of the process to generate decisions in an agile manner.
An HPU must act as an autonomous entity when it is in cooperative mode, or as a
resource when acting as a resource.

5. Case of study: The evaluation of the HPU in a continuous production


process

The continuous production processes use, classically, a hierarchical automation ap-


proach. In those cases, the regulation of the plant is carried out by algorithms such
as PIDs often carry out the regulation of the plant. The procedures to carry out the
optimization set a production objective whose restriction on the optimization method
is the production system (complete model), where only the physical element is con-
sidered via models and the conditions of the resources and equipment are left aside.
We emphasize that the complete model results from a composition of all the intercon-
nected models of the process. In these conditions, obtaining a global optimum is an
arduous task and whose solution is possible at times that are not in accordance with
the requirements of the applications or entities that require it (planning, programming,
supervision, forecasting).
The notion of behavior as a model treats all system variables equally (Willems
2007). After analyzing the model, and depending on the purpose for which the model
is used, it may be convenient to divide the variables of the system according to their
importance, as well as to define operating regions, delimited with thresholds or ranges
of variables, which determine a mode of behavior or state. In our approach, to set
a production goal it is required a behavior model that reflects the state of the units
(condition, reliability, dependability, etc.). Thus, of natural form, an optimum (local)
is given from the selection of one of the concatenations of units that meet the require-
ment: competence-ability and availability on the different units offering. If there is
only one route, the objective is fixed according to the state of the units. Each unit
uses its knowledge (models) to determine its state that involves resources, equipment;
and, quantity and quality of the product flow to be processed and obtained. Thus,
the management of the configuration, in this unit, for a production objective, is the
responsibility of the unit itself. Thus, the (complete) process model conceived in this
way is a valid image of its behavior and its size is relatively small with which on–line

15
decision making can be made.
To show how the HPU control architecture works, we will follow the methodology
of Section 3, and show the area of operations in a Water Distribution System as
a case study. Each one of the units is a links in the value chain (product route)
and corresponds to an HPU. The description of each unit will contain: the model of
behavior and operational conditions, handling and characteristics of the product flow
both at the entrance and exit, as well as special conditions if the unit requires it,
all (the models) are seen as models of –interconnected– hybrid systems (David and
Alla 2005; Branicky 1996). These models (hybrids) provide the state of the resources,
products and equipment generating the unit state, that make up the process view
(image), which adjusts and follows the process in accordance with the measures and
actions carried out in it. Thus, the planning is given by a realization obtained from the
behavior model of the area of operations from the particular composition of the discrete
events dynamics of each unit involved. The programming (generation of supervisors)
both for the coordination between units as the supervisor in each unit is a realization
that incorporates the established model in the planning adding the degraded and faults
conditions that can happen in the system, using Gemma description (Garcı́a 2004),
as a discrete event systems. The projection on the units or the equipment of the unit
establishes the operating modes (controllers and control parameters).

5.1. Water Distribution System


A Water Distribution System (WDS) aims to provide drinking water to a population
from sources of raw water. All activities are carried out so that this is a self-sustaining
company, maintaining both the quality of the service (water supply and quality) and
the cost thereof. The main processes that define the basic business areas in a WDS
are: i) study and characterization of the sources for a given population, ii) water
purification and distribution, iii) collection of sewage with the necessary treatment,
and iv) return to the environment and commercialization of drinking water. These four
macro-processes increase the social value for a population. The “area of operations”
in the WDS is associated with the water collect – potabilization (make drinkable) –
distribution - recovery of waste water, which is a product model that contains nine
links, five for drinking water and four for sewage (Water Supplies Department 2017).
See Fig. 11.

Figure 11. Fresh water distribution system. (From Water Supplies Department (2017))

The application schema of the HPU architecture for this process is shown in Fig.
12. Each stage is a link in the value chain and on this function that the HPU is built.
The characteristics that allow the construction of the models of each of the units will

16
be described, as well as the interaction of the product flows between them through the
pumping system or transfer. In detail, we will show the conformation of the knowledge
model (hybrid model) of the impounding reservoir unit. A simulation in the “NetLogo”
tool (Barbosa and Leitão 2011; Da Silva et al. 2016) of the interaction of the units
will be displayed at the end of the section. See 5.2.

Figure 12. Water Distribution System as a set of HPUs

5.1.1. Impounding Reservoir


The Impounding Reservoir (IR) has an area of 100 ha, with an average capacity of 120
million m3 and a maximum capacity of 169.79 million m3 . For simplicity, it is assumed
a trapezoidal geometric shape of equal depth at each end of 38 m, the suction head at
640 masl. Properties and conditions for the operation of the reservoir are shown in Fig.
13. The characteristics and operating conditions of the raw water entering the reservoir
are the maximum flow rate of 6.3m3 /s, and minimums of up to 2m3 /s, the ecological
flow (ambient return) is 1.18 /s. The permitted water quality conditions are: turbidity
< 50 Nephelometric Turbidity Unit or N T U (turbidimeter, max 1000 N T U ), apparent
color < 100 Platinum-Cobalt Unit or P CU (spectrophotometer, max 500 P CU ),
pH(6.5 − 9) (ph-meter, range 0 − 14), determined by laboratory. For this case, the
type of substrate, anions and cations present, as well as the dissolved oxygen are not
included.
With previous knowledge of both quality and flow of the raw water, at the entrance
of the water treatment plant, the Jar Test is carried out. This test consists of emulating
(applying) the clotting, flocculation, sedimentation process to several samples of rough
water (four usually) and provides the optimal ranges of the coagulant dose according
to the pH of raw water in additional the operational conditions in each unit mentioned.
Table 1 shows the operating conditions for the development of the behavior models
for the management of the IR.

17
Figure 13. Schematic for the reservoir

Table 1. IR operational conditions


Cond. Normal Degraded Failure
Var
3
Qi f low(m /s)
6.3 < Q < 4 4<Q<3 Q<2

Level 672¡C <664 664¡C¡654 C < 644,


C > 672
(masl)
Turbidity T < 15 15 < T < 49 T > 50

(NTU)
pH 6.5 ¡pH¡ 8.5 5 < pH < pH < 5,
6.5, 8.5 < pH > 9
pH < 9

The reservoir behavior model is given by the composition of three discrete events
models: the first one corresponds to the behavior of the entry gate, the second one is
the reservoir water level and the last one is the pumping system. The entry gate model
is established according to the following conditions: the reservoir level and the quality
of the input raw water (see table 2). For the reservoir level model, only three states
are considered: high, medium and low (regions in figure 8) and the pumping model is
established according to the following conditions: inflow and reservoir level (see table
2), as shown in Fig. 14.

Figure 14. Behavior of the reservoir in term of Discrete Event Systems

The composition of these models defines the reservoir behavior, and it has four oper-
ating modes, each one of them has a differential equation associated: filling and pump-
ing (dhi /dt = 1/ρA(hi )(qmi −qm0 )), filling and no pumping (dhi /dt = 1/ρA(hi )qmi ),
no filling and pumping (dhi /dt = 1/ρA(−qm0 )) and, no filling and no pumping
(dhi /dt = 0)) corresponding to the evolution given in Fig 14, and implements the
equation 1.

18
In Fig. 15, it is shown the UHP architecture for knowledge model (hybrid model)
of the impounding reservoir unit.

Figure 15. HPU Architecture for the Reservoir Unit

5.1.2. Transport
The flow of the pumping system is fixed at 4.5 m3 /s. Thus, depending on the conditions
of the IR and the consumption requirements, in the planning activity, it is determined
the number of pumpings by day and the pumping time (T p) according to purification
capacity.

5.1.3. Water purification plant


This process consists of transforming the raw water, with turbidity and pH parameters
given by laboratory test (see Table 1), into drinking water with the following param-
eters: turbidity < 2, apparent colour < 15, 6.5 < pH < 8.5. The purification process
consists of 6 phases (units): Coagulation, Flocculation, Sedimentation, Filtration –
Disinfection, and Storage (service reservoir). Fig. 12 highlights the part of purification
through a box containing basic HPUs for the processes mentioned above. Each unit
performs tests to determine the quality of the water at its exit.
Next section presents the application of the HPU concept for the continuous pro-
duction process. For validation in this paper, it is presented the application of the
proposed approach for the control of an agent-simulation model of the Water Treat-
ment Plant located in Dos-Cerritos, Venezuela. The Dos-Cerritos Plant supplies clean
water for human and industrial consumption to a set of cities within the region of
Barquisimeto and provides services to roughly 900.000 inhabitants in Venezuela. The
structure of this section is two-folded. Firstly, it describes the water treatment plant
used in this case study and it is presented the application of the HPU concept for this
scenario. Secondly, it presents the agent-simulation of the water treatment process
and it is carried out by a set of experiments of this model for validating the proposed
approach.

5.2. HPU simulation of the implementation


For validation of the proposed approach, the HPU concept and the test cases were
implemented in a proof-of-concept software called NetLogo (Barbosa and Leitão 2011;
Da Silva et al. 2016). NetLogo is a MAS programmable modeling environment built
to enable the exploration and analysis of the emergent behavior of natural, engineer-
ing and social phenomena. In this paper, the NetLogo software allows modeling the
HPU with the multi-agent capabilities and permits testing different scenarios for the

19
validation. Then, for this work, three testing scenarios are considered in the agent-
model of the water treatment plant. In specific, these testing scenarios are conducted
in order to demonstrate the vertical integration of the planning and execution process,
the reactiveness given in the control task and the pertinence of this approach for the
continuous production processes.

Figure 16. NetLogo interface for the Reservoir HPU representation

5.2.1. HMS architecture with the HPU concept


The case study presented in this paper is the purification of raw water in a water treat-
ment plant. The potabilization of water, considered a continuous production process,
is a complex process where the raw water passes throughout the water plant to be
purified to improve the quality measurement (i.e. Turbidity). However, this process in
a complex system as even though the water is a fluid processed in a continuous pro-
cess; it also contains also characteristics of a batch production process. The detailed
description of this process is in section 5.1. In this case of study, the water treatment
process is composed of a reservoir, four treatment tanks and a service reservoir. Fig.
16 illustrates the simulation of the water treatment plant, the storage capacity of the
reservoirs and tanks, the flow rates and the minimum quality of water to pass to the
next component. The tanks are connected through a pipeline that pumps the water
between the different components. The pumping of water starts in the reservoir passes
through the coagulation, flocculation, sedimentation and the filtration tanks of the wa-
ter treatment plant, and finishes in the service reservoir for distributing to the human
and industrial consumption.
The proposed HMS architecture is divided into three levels: The plant level, the
image level and the enterprise level. Fig. 12 illustrates HMS architecture of the stated
case study. While the plant and image levels are responsible for the execution of the
water treatment process, the enterprise level is responsible for the planning process.
The plant level contains the physical assets (aka physical object) from the water treat-
ment plant. In this paper, the physical assets are the physical function of the reservoir,
the four treatment tanks and the service reservoir. The image level contains the HPUs
that, similarly to the digital twin concept, controls the physical assets function. The

20
enterprise level is oriented to evaluate the global performance of the execution, transfer
information to the enterprise system and accept/reject assigned objectives.
For the components of the HMS architecture, 6 HPUs were created for each tank
of the case study. As a general description, each HPU is responsible for monitoring
events and triggering actions of the corresponding physical asset. On one side, the HPU
monitors the water level of tank/reservoir (measured in m3 ), the flow in/out of the
tank/reservoir (measured in m3 /s) and the quality of water (measured in NTU). On
the other side, the HPS could trigger the tank/reservoir gates to next physical asset,
the quantity of chemical for the purification process (i.e. flocculation and chlorination)
and the timing of each process in the tank.

5.2.2. Validation of the HPU in a normal condition (scenario 1)


The first testing scenario is related to the normal operation of the water treatment
plant. This scenario is explored in order to test the proposed approach without any
perturbation. It aims to understand the dynamics and normal functioning of the water
treatment process. Furthermore, this scenario works as a control scenario for the other
testing scenarios and it will allow testing the reactiveness in case of a perturbation.

5.2.3. Disrupted condition of the HPU-water demand (scenario 2)


The second testing scenario is regarding water demand where a perturbation is de-
tected during the water treatment process. Specifically, the total demand of the water
treatment plant increases further than usual surpassing the forecasting and corre-
sponding threshold. This event is considered a perturbation for the normal conditions
and requires that the control executed by the proposed HMS architecture reacts to
the new situation. Therefore, this scenario aims to devise the functioning of the HMS
architecture when it recovers from a disrupted scenario.

5.2.4. Disrupted condition of the HPU-quality of water (scenario 3)


The third testing scenario is regarding the quality of water where a perturbation is
detected during the water treatment process. In this case, the perturbation detected is
related to the quality of water. As it was mentioned before, raw water from reservoirs
contains a large number of particles that are cleaned at the water treatment process.
This haziness of the fluid is measured in turbidity and represent a key test of the water
quality. Normally, it is expected that the turbidity is between a turbidity interval.
However, depending on the rain, river flow, and reservoir environment, this turbidity
could exceed the expected measure. Then, the processing time in the water treatment
plant may increase accordingly. The perturbation in this scenario in that the turbidity
of the raw water exceeds the turbidity expected. This scenario aims to explore the
functioning of the HMS architecture considering that the processing time in the water
treatment process (i.e. Coagulation, Flocculation, etc.) last longer than usual.
Fig. 17 shows an example of possible simulations of the scenarios above mentioned
via NetLogo.

6. Discussion

Stock and Seliger (2016) define that the “paradigm of I4.0 is essentially outlined by
three dimensions (Req5 ): i) horizontal integration across the entire value creation

21
Figure 17. NetLogo scenarios results (trends of the variables).

network, ii) end-to-end engineering across the entire product life cycle, as well as iii)
vertical integration and networked manufacturing systems”. Other vital elements in
I4.0 are: the standardization of information and telecommunication technologies, the
interaction of the system with humans, the use of CPPS concepts (Req1 – Req4 ).
Taking into account those recommendations and the requirements established in Cruz
and Vogel-Heuser (2017), the following analysis is elaborated (See section 3.1).

6.1. HPU architecture aligned with the Industry 4.0 requirements


In the HPU approach, a production system is a network of intelligent units responsible
for storing, transporting, transforming materials (Req1 ). HPUs must work coopera-
tively to deliver final products to consumption centers. The flow of materials acquires
value as it passes over the different units, as shown in Fig. 6; this constitutes the prod-
uct value chain. A product can be obtained by using different product routes (value
chains), so in the planning activity, one of the main tasks is to select the optimal route
for an order.

6.1.1. Horizontal Integration


HMSs describe the life cycle axis of the product as the product model (Req5 ). The
HPU describes the life cycle of the product in terms of services that must be provided
to fulfill the mission. The link between different HPUs gives the horizontal axis corre-
sponding to the value chain, to provide all the necessary services to fulfill the mission.
This route is not a predefined one for a mission, and it is determined according to an
optimization criterion. In case of failures of an HPU, when the process is in operation
mode, if there are available HPUs that can perform the same service and a possible
route, the system is reconfigured (Req2 ).

6.1.2. Vertical integration


HPU proposes the concept of intelligent resources as a central element. The HPU has
three sets of dynamic models that describe the evolution of the production process,
the procedure to obtain a product based on “stages”, and the dynamics of internal
resources in order to determine their capacity and availability. The state of the sys-
tem is obtained from measurements of variables of materials, physical processes, and
equipment, which associated with the models (Req5 ); allow determining the state of
equipment, processes and product quality. The image maintains complete information
of what is happening on the shop floor and follows the process dynamics from the

22
events. This concept is similar to the digital twin in I4.0.

6.1.3. Use of mathematical models for the process (CPPS)


Mathematical models can build the scheduling of the production systems and define
controllers and supervisors describe the behavior of the process (Req3 ).

6.1.4. Standardization
The equipment hierarchy levels and Process models follow the ISA recommendations
(Req1, Req5 ). Product models are specified by the services that are provided by smart
resources (Req2 ). The implementation of control and supervision mechanisms are
adapted to the classical OT (Req1 ). Additional, HPU can consult the OT compo-
nents (from several brands) to hybrid topologies implementation (Req3 ) with open
communication protocols (Req1 ) –e.g. using OPC Unified Architecture (OPC UA).

6.1.5. Decentralization
The HPU architecture considers each resource as autonomous - cooperating (Req1 ).
The mechanisms of control and supervision of the resource are mapped on elements of
the OT / IT architecture available in the organization. Their images, and the high level
functions of each HPU are in the cloud, where holarchies are established in high-level
interactions.

6.1.6. Digital twins


A central element in the architecture is the concept of the process image and the
component resources of an HPU. The model is similar to the concept of a digital
twin and it allows knowing the status of processes and resources. The image receives
detecting events on the plant floor, and by behavioral models define what the state of
the process (Req3 ), which allows decision making in an agile way is (Req4, Req5 ).

7. Conclusion

Following the preliminary design in (Cruz S., Rojas A., and Chacón 2019) expanded
HMS architecture for continuous processes. The HPU is based on the concept of HMS
and contains three fundamental holons (RH, MH, and SH). An additional holon, sim-
ilar to the one proposed in ADACOR, allows for the supervision and control tasks
for each HPU. In this way, HPUs can resolve unpredictable demands and implement
internal fault tolerance behavior. The mechanisms of high-level cooperation, as well as
the establishment of global objectives, are achieved through elements that model the
physical behavior of the processes and allow maintaining a coherent operation between
units. This idea is similar to that of a digital twin concept.
Cooperation between units is achieved by establishing viable global configurations
and selecting the optimal one. The formation of individual behavior models allows
determining global behaviors through simulation. Each unit is modeled separately,
and the composition of models establishes possible operation configurations. These
configurations found to ensure the continuation of the operation even if they are not
the best.

23
The article shows how the HPU architecture can meet the requirements of I4.0,
even addressing RAMI 4.0 axes.

References

Alur, Rajeev, Thomas A Henzinger, Gerardo Lafferriere, and George J Pappas. 2000. “Discrete
abstractions of hybrid systems.” Proceedings of the IEEE 88 (7): 971–984.
Alvarez, Alex, Winston Garcı́a, Oscar Camacho, and Edgar Chacón. 1999. “Automation of a
steam generation plant using hybrid systems.” ISA transactions 38 (1): 87–99.
Antsaklis, Panos, Xenofon Koutsoukos, and Janan Zaytoon. 1998. “On hybrid control of com-
plex systems: A survey.” Journal européen des systèmes automatisés 32 (9–10): 1023–1045.
Antsaklis, Panos J, and Xenofon D Koutsoukos. 2003. “Hybrid systems: Review and recent
progress.” Software-Enabled Control: Information Technology for Dynamical Systems 273–
298.
Arboleda C., A. S. 2011. “Transformación automática de requisitos representados en esquemas
preconceptuales a modelos de interacción de sistemas holónicos.” Master’s thesis, Maestrı́a
en Ingenierı́a de Sistemas, Facultad de Minas Escuela de Sistemas, Universidad Nacional de
Colombia, Medellı́n.
Barbosa, José. 2015. “Self-organized and evolvable holonic architecture for manufacturing con-
trol.” PhD diss., Université de Valenciennes et du Hainaut-Cambresis.
Barbosa, José, and Paulo Leitão. 2011. “Simulation of multi-agent manufacturing systems
using agent-based modelling platforms.” In 2011 9th IEEE International Conference on
Industrial Informatics, 477–482. IEEE.
Barbosa, José, Paulo Leitão, Emmanuel Adam, and Damien Trentesaux. 2015. “Dynamic
self-organization in holonic multi-agent manufacturing systems: The ADACOR evolution.”
Computers in Industry 66: 99–111.
BenSaleh, Mohammed S, Syed Manzoor Qasim, Abdulfattah M Obeid, and Alberto Garcia-
Ortiz. 2013. “A review on wireless sensor network for water pipeline monitoring applica-
tions.” In 2013 International Conference on Collaboration Technologies and Systems (CTS),
128–131. IEEE.
Blanc, Pascal, Isabel Demongodin, and Pierre Castagna. 2008. “A holonic approach for manu-
facturing execution system design: An industrial application.” Engineering Applications of
Artificial Intelligence 21 (3): 315–330.
Bloch, Henry, Alexander Fay, Torsten Knohl, Stephan Hensel, Anna Hahn, Leon Urbas, Sachari
Wassilew, Jens Bernshausen, Mario Hoernicke, and Axel Haller. 2017. “Model-based engi-
neering of CPPS in the process industries.” In 2017 IEEE 15th International Conference on
Industrial Informatics (INDIN), 1153–1159. IEEE.
Borangiu, T., D. Trentesaux, A. Thomas, and S. Cavalieri, eds. 2019. Service Orientation in
Holonic and Multi-Agent Manufacturing. Vol. 803. Springer International Publishing.
Borangiu, Theodor, Silviu Raileanu, Damien Trentesaux, Thierry Berger, and Iulia Iacob. 2014.
“Distributed manufacturing control with extended CNP interaction of intelligent products.”
Journal of Intelligent Manufacturing 25 (5): 1065–1075.
Branicky, Michael S. 1996. “General hybrid dynamical systems: Modeling, analysis, and con-
trol.” In Hybrid Systems III, 186–200. Springer.
Branicky, M.S., V.S. Borkar, and S.K. Mitter. 1998. “A unified framework for hybrid control:
model and optimal control theory.” IEEE Trans. on Automatic Control 43: 31–45.
Cardin, Olivier, Damien Trentesaux, André Thomas, Pierre Castagna, Thierry Berger, and
Hind Bril El-Haouzi. 2017. “Coupling predictive scheduling and reactive control in manu-
facturing hybrid control architectures: state of the art and future challenges.” Journal of
Intelligent Manufacturing 28 (7): 1503–1517.
Cassandras, Christos G, and Stephane Lafortune. 2009. Introduction to discrete event systems.
Springer Science & Business Media.
Chacón, E., E. Carrasco, and J. Cardillo. 2012. “Programación en lı́nea de procesos de

24
producción continua mediante supervisores.” In International Society of Automation ISA
Colombia (Ed.), IIII (Tercera) Jornada de automatización de la industria petrolera JAIP,
Vol. 1a.edición, Bogotá, Colombia, 278 – 286. International Society of Automation ISA
Colombia.
Chacón, Edgar, Isabel Besembel, and Jean Claude Hennet. 2004. “Coordination and optimiza-
tion in oil and gas production complexes.” Computers in Industry 53 (1): 17–37.
Chacón, Edgar, Isabel Besembel, Dulce Rivero, and Juan Cardillo. 2009. “Embedded holonics
systems in production process: holonic unit of production.” Revista Técnica de la Facultad
de Ingenierı́a. Universidad del Zulia 32 (1).
Chacon, Edgar, Juan Cardillo, Rafael Chacon, and Germán Zapata. 2012. “Planification
en ligne pour les systemes de production distribués: une approche par les systemes
holoniques.” In 9th International Conference on Modeling, Optimization & SIMulation,
https://hal.archives-ouvertes.fr/hal-00728682/document.
Chokshi, Nirav, and Duncan McFarlane. 2008a. “A distributed architecture for reconfigurable
control of continuous process operations.” Journal of Intelligent Manufacturing 19 (2): 215–
232.
Chokshi, Nirav N., and Duncan C. McFarlane. 2008b. A Distributed Coordination Approach
to Reconfigurable Process Control. Springer.
Colombo, Armando Walter, and Stamatis Karnouskos. 2009. “Towards the factory of the
future: A service-oriented cross-layer infrastructure.” ICT Shaping the World: A Scientific
View. European Telecommunications Standards Institute (ETSI), John Wiley and Sons 65:
81.
Cruz, S Luis A, and Birgit Vogel-Heuser. 2017. “Comparison of agent oriented software method-
ologies to apply in cyber physical production systems.” In 2017 IEEE 15th International
Conference on Industrial Informatics (INDIN), 65–71. IEEE.
Cruz S., Luis A. 2018. Automatización Industrial Inteligente: Una estructura de control desde
el paradigma holónico de manufactura. (Intelligent Industrial Automation: A control struc-
ture since the holonic manufacturing paradigm). 1st ed. Beau Bassin, Mauritius: Editorial
Académica Española.
Cruz S., Luis. A., and Oscar A. Rojas A. 2014. “The future of industrial automation and IEC
614993 standard.” In III International Congress of Engineering Mechatronics and Automa-
tion, CIIMA, 1–5.
Cruz S., Luis. A., Oscar A. Rojas A., and Edgar Chacón. 2019. Service Orientation in Holonic
and Multi-Agent Manufacturing, Chap. Implementing Industrial Control Automation for
Holonic Manufacturing Systems based on the Production Unit Architecture, 1–14. Springer
International Publishing.
Cruz S., Luis Alberto, Felix Mayer, Daniel Schütz, and Birgit Vogel-Heuser. 2018. “Plat-
form Independent Multi-Agent System for Robust Networks of Production Systems.” IFAC-
PapersOnLine 51 (11): 1261–1268.
Da Silva, Robson Marinho, Fabricio Junqueira, Diolino J Santos Filho, and Paulo E Miyagi.
2016. “Control architecture and design method of reconfigurable manufacturing systems.”
Control Engineering Practice 49: 87–100.
Dai, Wenbin, and Valeriy Vyatkin. 2013. “A component-based design pattern for improving
reusability of automation programs.” In IECON 2013-39th Annual Conference of the IEEE
Industrial Electronics Society, 4328–4333. IEEE.
DAVID, RENÉ, and HASSANE ALLA. 2001. “On Hybrid Petri Nets.” Discrete Event Dy-
namic Systems: Theory and Applications 11: 9–40.
David, René, and Hassane Alla. 2005. Discrete, Continuous, and Hybrid Petri Nets. Springer.
Dias-Ferreira, João, Luis Ribeiro, Hakan Akillioglu, Pedro Neves, and Mauro Onori. 2018.
“BIOSOARM: a bio-inspired self-organising architecture for manufacturing cyber-physical
shopfloors.” Journal of Intelligent Manufacturing 1–24.
Dietrich, Andreas J, Stefan Kirn, and Vijayan Sugumaran. 2007. “A service-oriented architec-
ture for mass customization – A shoe industry case study.” IEEE Transactions on engineer-
ing management 54 (1): 190–204.

25
Dumitrache, I, and S. Caramihai. 2010. “Intelligent Manufacturing: a New Paradigm.” In IFAC
Proceedings Volumes, Vol. 43, 1–7.
Galán, Ramón, Agustı́n Jiménez, Ricardo Sanz, and Fernando Matı́a. 2000. “Control in-
teligente.” Inteligencia Artificial. Revista Iberoamericana de Inteligencia Artificial 4 (10).
Gamboa Quintanilla, Francisco, Olivier Cardin, Anne L’Anton, and Pierre Castagna. 2015.
“Process Specification Framework in a Service Oriented Holonic Manufacturing Systems.”
In Service Orientation in Holonic and Multi-agent Manufacturing, edited by T. Borangiu,
A. Thomas, and D. Trentesaux, 81–89. Springer.
Gamboa Quintanilla, Francisco, Sylvain Kubler, Olivier Cardin, and Pierre Castagna. 2013.
“Product Specification in a Service-Oriented Holonic Manufacturing System using Petri-
Nets.” In Intelligent Manufacturing Systems, Vol. 11.
Garcı́a, Emilio. 2004. Automatización de Procesos Industriales. Universidad Politécnica de
Valencia, Alfaomega.
Garcı́a Valls, Marisol, I Rodrı́guez López, and L Fernández Villar. 2013. “iLAND: An enhanced
middleware for real-time reconfiguration of service oriented distributed real-time systems.”
IEEE Transactions on Industrial Informatics 9 (1): 228–236.
Henzinger, Thomas A. 2000. The theory of hybrid automata. Springer.
Hsieh, Fu-Shiung. 2009. “Collaborative reconfiguration mechanism for holonic manufacturing
systems.” Automatica 45: 2563–2569.
Indriago, Carlos, Olivier Cardin, Odile Bellenguez-Morineau, Naly Rakoto, and Pierre
Castagna. 2015. “Performance evaluation of holonic-based online scheduling for a switch
arrival system.” In IFAC Papers Online, Vol. 48, 1105–1110.
Indriago, Carlos, Olivier Cardin, Naly Rakoto, Pierre Castagna, and Edgar Chacon. 2014.
“Application du paradigme holonique a un systeme de reservoirs.” In MOSIM 2014, 10ème
Conférence Francophone de Modélisation, Optimisation et Simulation, https://hal.archives-
ouvertes.fr/hal-01166615.
Indriago, Carlos, Olivier Cardin, Naly Rakoto, Pierre Castagna, and Edgar Chacón. 2016. “H
2 CM: A holonic architecture for flexible hybrid control systems.” Computers in Industry
77: 15–28.
Jimenez, Jose-Fernando, Abdelghani Bekrar, Gabriel Zambrano-Rey, Damien Trentesaux, and
Paulo Leitão. 2017. “Pollux: a dynamic hybrid control architecture for flexible job shop
systems.” International Journal of Production Research 55 (15): 4229–4247.
Jimenez Gordillo, Jose Fernando. 2017. “Dynamic and hybrid architecture for the optimal
reconfiguration of control systems: application to manufacturing control.” PhD diss., Uni-
versité de Valenciennes et du Hainaut-Cambresis.
Joannou, Demetrios, Roy Kalawsky, Sara Saravi, Monica Rivas Casado, Guangtao Fu, and
Fanlin Meng. 2019. “A Model-Based Engineering Methodology and Architecture for Re-
silience in Systems-of-Systems: A Case of Water Supply Resilience to Flooding.” Water 11
(3): 496.
Karnouskos, Stamatis, Armando Walter Colombo, Thomas Bangemann, Keijo Manninen,
Roberto Camp, Marcel Tilly, Petr Stluka, François Jammes, Jerker Delsing, and Jens Elias-
son. 2012. “A SOA-based architecture for empowering future collaborative cloud-based in-
dustrial automation.” In IECON 2012-38th Annual Conference on IEEE Industrial Elec-
tronics Society, 5766–5772. IEEE.
Kim, J-K, and R Smith. 2004. “Automated design of discontinuous water systems.” Process
Safety and Environmental Protection 82 (3): 238–248.
Lee, Jay, Behrad Bagheri, and Hung-An Kao. 2015. “A cyber-physical systems architecture
for industry 4.0-based manufacturing systems.” Manufacturing Letters 3: 18–23.
Leitão, Paulo, and Stamatis Karnouskos. 2015. Industrial Agents: Emerging Applications of
Software Agents in Industry. Morgan Kaufmann.
Leitao, Paulo, Stamatis Karnouskos, Luis Ribeiro, Jay Lee, Thomas Strasser, and Armando W
Colombo. 2016. “Smart Agents in Industrial Cyber–Physical Systems.” Proceedings of the
IEEE 104 (5): 1086–1101.
Leitão, Paulo, and Francisco J. Restivo. 2008. “Implementation of a Holonic Control System in

26
a Flexible Manufacturing System.” IEEE TRANSACTIONS ON SYSTEMS, MAN, AND
CYBERNETICS – PART C: APPLICATIONS AND REVIEWS 38 (5): 699–709.
Lemmon, M.D., K.X. He, and I. Markovsky. 1999. “Supervisory hybrid systems.” Control
Systems, IEEE 19 (4): 42–55.
Lennartson, Bengt, Kristofer Bengtsson, Oskar Wigström, and Sarmad Riazi. 2016. “Modeling
and Optimization of Hybrid Systems for the Tweeting Factory.” IEEE Transactions on
Automation Science and Engineering 13 (1): 191–205.
Lennartson, Bengt, Bo Egardt, and Michael Tittus. 1994. “Hybrid systems in process control.”
In Decision and Control, 1994., Proceedings of the 33rd IEEE Conference on, Vol. 4, 3587–
3592. IEEE.
Lu, Yan, Katherine C Morris, and Simon Frechette. 2016. Current standards landscape
for smart manufacturing systems. Technical Report. National Institute of Standards
and Technology, NISTIR. Http://nvlpubs.nist.gov/nistpubs/ir/2016/NIST.IR.8107.pdf,
http://nvlpubs.nist.gov/nistpubs/ir/2016/NIST.IR.8107.pdf.
Lüder, Arndt, Ambra Calá, Jacek Zawisza, and Ronald Rosendahl. 2017. “Design pattern for
agent based production system control–A survey.” In Automation Science and Engineering
(CASE), 2017 13th IEEE Conference on, 717–722. IEEE.
Lygeros, John. 2004. “Lecture notes on hybrid systems.” In Notes for an ENSIETA workshop,
Citeseer.
Manna, Zohar, and Amir Pnueli. 1993. “Verifying hybrid systems.” In Hybrid Systems, edited
by R.L. Grossman, Nerode A., A.P. Ravn, and H. Rischel, 4–35. Springer.
Marsalek, J., B.E. Jiménez-Cisneros, P.-A. Malmquist, M. Karamouz, J. Goldenfum, and
B. Chocat. 2006. Urban water cycle processes and interactions. International Hydrological
Programme ((IHP) UNESCO.
McFarlane, Duncan C, and Stefan Bussmann. 2000. “Developments in holonic production
planning and control.” Production Planning & Control 11 (6): 522–536.
McFarlane, Duncan C, and Stefan Bussmann. 2003. “Holonic manufacturing control: Ratio-
nales, developments and open issues.” In Agent-based manufacturing, edited by S. M. Deen,
303–326. Springer.
Méndez, Carlos A., Jaime Cerdáb, Ignacio E. Grossmann, Iiro Harjunkoskic, and Marco Fahlc.
2006. “State-of-the-art review of optimization methods for short-term scheduling of batch
processes.” Computers & Chemical Engineering 30 (6–7): 913–946.
Monostori, László. 2018. CIRP Encyclopedia of Production Engineering, Chap. Cyber-Physical
Systems, 1–8. Springer Berlin Heidelberg.
Monostori, László, Botond Kádár, T Bauernhansl, S Kondoh, S Kumara, G Reinhart, O Sauer,
G Schuh, W Sihn, and K Ueda. 2016. “Cyber-physical systems in manufacturing.” CIRP
Annals 65 (2): 621–641.
Nam, Taewoo, and Theresa A Pardo. 2011. “Conceptualizing smart city with dimensions of
technology, people, and institutions.” In Proceedings of the 12th annual international digital
government research conference: digital government innovation in challenging times, 282–
291. ACM.
Palacio B., J. E. 2013. “Una Propuesta para Distribuir el Coordinador de una
Unidad de Producción Holónica a partir de la Teorı́a de Control Supervisorio.”
Master’s thesis, Tesis de Maestrı́a Universidad Nacional de Colombia, Manizales.
http://www.bdigital.unal.edu.co/10585/.
Pérez, Federico, Edurne Irisarri, Darı́o Orive, Marga Marcos, and Elisabet Estevez. 2015.
“A CPPS Architecture approach for Industry 4.0.” In Emerging Technologies & Factory
Automation (ETFA), 2015 IEEE 20th Conference on, 1–4. IEEE.
Platform Industrie 4.0, (I4.0). 2018. The Structure of the Administration Shell: Trilat-
eral Perspective from France, Italy and Germany. Technical Report. Platform Indus-
trie 4.0. https://www.plattform-i40.de/I40/Redaktion/EN/Downloads/Publikation/hm-
2018-trilaterale-coop.pdf? blob=publicationFile&v=5.
Pujo, Patrick, Nicolas Broissin, and Fouzia Ounnar. 2009. “PROSIS: An isoarchic structure
for HMS control.” Engineering Applications of Artificial Intelligence 22 (7): 1034–1045.

27
Ribeiro, Luis, and Martin Hochwallner. 2018. “On the Design Complexity of Cyberphysical
Production Systems.” Complexity 2018.
Salazar V., R. D. 2009. “Modelo De Holones Recurso En Sistemas Holónicos De Manufactura.
Interface.” Master’s thesis, Tesis de Maestrı́a en Automatización Industrial, Facultad de
Ingenierı́a y Arquitectura, Universidad Nacional de Colombia-Sede Manizales.
Simón-Marmolejo, I, Omar López-Ortega, LE Ramos-Velasco, and M Ortiz-Domı́nguez. 2017.
“Ontologı́a Unificada para un Sistema Holónico de Manufactura.” Revista Iberoamericana
de Automática e Informática industrial .
Stock, T, and G Seliger. 2016. “Opportunities of sustainable manufacturing in industry 4.0.”
Procedia Cirp 40: 536–541.
Sun, Zhi, Pu Wang, Mehmet C. Vuran, Mznah A. Al-Rodhaan, Abdullah M. Al-Dhelaan,
and Ian F. Akyildiz. 2011. “MISE-PIPE: Magnetic induction-based wireless sensor net-
works for underground pipeline monitoring.” Ad Hoc Networks 9 (3): 218 – 227.
http://www.sciencedirect.com/science/article/pii/S1570870510001617.
Thomas, André, Theodor Borangiu, and Damien Trentesaux. 2017. “Holonic and multi-agent
technologies for service and computing oriented manufacturing.” Journal of Intelligent Man-
ufacturing 28 (7): 1501–1502.
Tippett, Michael J, and Jie Bao. 2015. “Distributed control of chemical process networks.”
International Journal of Automation and Computing 12 (4): 368–381.
Tittus, Michael, and Bengt Lennartson. 1999. “Hierarchical supervisory control for batch pro-
cesses.” Control Systems Technology, IEEE Transactions on 7 (5): 542–554.
Trentesaux, Damien. 2009. “Distributed control of production systems.” Engineering Applica-
tions of Artificial Intelligence 22 (7): 971–978.
Van Der Schaft, Arjan J, and Johannes Maria Schumacher. 2000. An Introduction to Hybrid
Dynamical Systems. Vol. 251. Springer London.
Vogel-Heuser, Birgit, Christian Diedrich, Dorothea Pantförder, and Peter Göhner. 2014. “Cou-
pling heterogeneous production systems by a multi-agent based cyber-physical production
system.” In Industrial Informatics (INDIN), 2014 12th IEEE International Conference on,
713–719. IEEE.
Wagner, Janet M, Uri Shamir, and David H Marks. 1988. “Water distribution reliability:
simulation methods.” Journal of water resources planning and management 114 (3): 276–
294.
Water Supplies Department. 2017. “Core Businesses: Drinking Water Quality.” See: Jan-
uary, 2019, https://www.wsd.gov.hk/en/core-businesses/water-quality/my-drinking-water-
quality/index.html.
Willems, Jan C. 2007. “The behavioral approach to open and interconnected systems.” Control
Systems, IEEE 27 (6): 46–99.
Zapata, Germán. 2011. “Propuesta Para la Planificación, Programación, Supervisión y Control
de la Producción en Procesos Continuos Desde la Teorı́a del Control Supervisorio y el
Enfoque Holónico.” PhD diss., Facultad de Ingenierı́a. Universidad de Los Andes, Mérida,
Venezuela.

28

You might also like