Development of a small-scaleeducational workbench for Industry 4.0

André Pedro Ramalho Martins Pacheco

Thesis to obtain the Master of Science Degree in

Mechanical Engineering

Supervisor: Prof. João Carlos Prata dos Reis

Examination Committee

Chairperson: Prof. Carlos Baptista CardeiraSupervisor: Prof. João Carlos Prata dos Reis

Members of the Committee: Prof. José Barata OliveiraProf. Mário José Gonçalves Cavaco Mendes

June 2019

ii

Acknowledgments

First of all I would like to express my deepest gratitude towards my supervisor, Prof. Joao Reis, for all

his guidance, availability, support and patience during this work development. Without his constructive

feedback, this thesis would not have been possible.

Eng. Camilo Christo and Prof. Mario Ramalho gave me invaluable help on dealing with the equip-

ments and the programming language used on this work. To them, I appreciate the counselling given.

To my family, especially my mother and brother who had confidence in me and gave unconditional

support not only during the writing of this thesis but throughout my life.

To the people that I met during my years at IST and became my friends. I thank Goncalo Deus,

Carolina Pereira, Rohan Chotalal, Andre Passos, Eunice Ferreira, Miguel Bigares and Miguel Ramalho

for the companionship and presence, both during the good times and the challenging ones.

To my colleagues at the ACCAII laboratory, namely Joao Carreira, Guilherme Kano and Tiago An-

drade for creating a truly enjoyable environment.

Finally to my good friends, who have accompanied me for so many years. My heartfelt thanks to all

of them, particularly to Joana Fernandes and Flavia Forte whose kind words proved to be priceless in

the hours of most need.

iii

iv

Resumo

Encontra-se a decorrer uma transformacao no sector fabril, com o objetivo de diluir as barreiras entre o

mundo fısico e o digital. Apresentado publicamente na feira de Hannover em 2011, o programa Industria

4.0 planeia convergir varias tecnologias emergentes com o objetivo de desenvolver fabricas assentes

numa infraestrutura de rede, permitindo assim a digitalizacao da cadeia de valor desde o cliente ate ao

fim de vida do produto. Alem disso, a colaboracao entre sistemas ciber-fısicos torna possıvel que nas

fabricas inteligentes se produzam produtos personalizados em lotes reduzidos que sao rentaveis. A

chegada da quarta revolucao industrial tera impactos na educacao e nos metodos usados na formacao

de engenheiros.

O objetivo desta dissertacao foi o desenvolvimento de um conjunto experimental que transmita aos

seus utilizadores os conceitos mais relevantes da I4.0, assente numa metodologia de aprender fazendo.

Para tal, realizou-se um levantamento das tecnologias da I4.0 e solucoes didaticas. Uma comparacao

entre equipamentos ja existentes permitiu a selecao das suas principais propriedades. Com base nessa

analise desenvolveu-se uma bancada de aprendizagem em escala reduzida que simula os aspetos de

operacao de uma fabrica inteligente, com enfase na comunicacao em rede por parte de equipamentos

dissimilares e no produto com um papel ativo.

Embora os princıpios da I4.0 sejam representados mais frequentemente em larga escala demonstra-

se que, dentro de um certo nıvel de complexidade, estes continuam a ser validos num modelo menor

inserido num contexto de aprendizagem.

Palavras-chave: Automacao, Industria 4.0, Sistemas Ciber-Fısicos, Internet Industrial das

Coisas, Bancada de Aprendizagem.

v

vi

Abstract

A transformation of the manufacturing industry is currently underway, set to dilute the barriers between

the physical and the digital world. Publicly presented at the Hannover Messe in 2011, the Industry 4.0

directive plans on converging several emergent technologies with the aim of developing factories with

a network backbone, allowing for a complete digitalization of the value chain from costumer until the

product end of life. Furthermore, the joint operation of cyber-physical systems makes it possible for

smart factories to produce customized products in small batches that are profitable. The coming of the

fourth industrial revolution will have impacts on education and the methods used to train engineers.

The aim of this dissertation was to develop an experimental set for transmitting to its users the most

relevant I4.0 concepts, based on a learning-by-doing methodology. To accomplish this, a review on

the I4.0 building blocks and didactic solutions was conducted. A comparison between already existing

equipments enabled the selection of the more substantial properties. Based on this analysis, a small-

scale interactive workbench was developed, that simulates the aspects of the operation of a smart

factory, with an emphasis on the networking of heterogeneous equipments and on the product with an

active role.

Although the principles of I4.0 are more frequently represented on a large scale, it is demonstrated

that, within a certain complexity level, they are still valid in a smaller model in a learning context.

Keywords: Automation, Industry 4.0, Cyber-Physical Systems, Industrial Internet of Things,

Training Workbench.

vii

viii

Contents

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

Resumo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Topic Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Industry 4.0: a journey towards smart manufacturing 7

2.1 Justification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 Proposed concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3 Training systems: state of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.4 Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.5 Supporting technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.5.1 Cyber Physical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.5.2 Internet of Things . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.5.3 Big Data Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.5.4 Human-robot collaboration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.5.5 Augmented Reality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.5.6 Electronic tagging technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.6 Revisions and implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.6.1 Corporate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.6.2 Social . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.7 Topics for teaching and training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.7.1 Comparative analysis of training systems . . . . . . . . . . . . . . . . . . . . . . . 30

2.7.2 Skills for the workplace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

ix

2.7.3 Demonstrators as teaching aids . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3 A small scale electro-pneumatic workbench for Industry 4.0 training 37

3.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.2 Proposed implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.3 Workbench setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.4 Operation modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.5 Communications between desktop PC and PLC . . . . . . . . . . . . . . . . . . . . . . . 44

3.6 Webpages as information carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.6.1 Virtual Laboratory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.6.2 Equipment webpages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

3.7 Simulation of CPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4 Laboratory use cases 59

4.1 Test specimens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

4.2 Workpiece identifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.3 Manufacturing simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.4 Quality control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.5 Database Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.6 Emergency stop procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.6.1 Presence detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.6.2 Emergency stop switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

4.7 A simulated collaborative scenario for Cyber-Physical Systems . . . . . . . . . . . . . . . 65

5 Conclusions 69

5.1 Achievements and final remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Bibliography 71

x

List of Tables

2.1 Industry 4.0 demonstrators and respective features . . . . . . . . . . . . . . . . . . . . . . 30

3.1 8 bit integer representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

3.2 Correspondence between states and active programs . . . . . . . . . . . . . . . . . . . . 55

4.1 Joint movement between cylinders and elevator . . . . . . . . . . . . . . . . . . . . . . . . 61

xi

xii

List of Figures

1.1 The four major industrial eras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2.1 The automation pyramid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Industry 4.0 synergies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3 The MyJoghurt demonstrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.4 Details from the MyJoghurt demonstrator . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.5 The CiP Learning Factory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.6 The MTA SZTAKI Learning Factory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.7 The digital twin of Maserati’s Avvocato plant. . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.8 Industry 4.0 building blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.9 A CPS in action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.10 Connected devices forecast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.11 World’s data growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.12 A collaborative (human-robot) assembly scenario . . . . . . . . . . . . . . . . . . . . . . . 24

2.13 Augmented Reality assisted assembly process . . . . . . . . . . . . . . . . . . . . . . . . 25

2.14 A web based service for the aviation industry . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.15 Beckhoff Smart Factory demonstrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.16 Roboter Integrated Agent Network (RAIN) . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.1 Learning workbench for Industry 4.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.2 Industry 4.0 Automation Laboratory diagram . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.3 Matlab application for interacting with the PLC. . . . . . . . . . . . . . . . . . . . . . . . . 42

3.4 List of workbench elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.5 Continuous mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.6 Task List mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.7 GRAFCET diagram for the production system (part 1) . . . . . . . . . . . . . . . . . . . . 46

3.7 GRAFCET diagram for the production system (part 2) . . . . . . . . . . . . . . . . . . . . 47

3.7 GRAFCET diagram for the production system (partial GRAFCET) . . . . . . . . . . . . . 48

3.8 Text representation convention in Saia code editor. . . . . . . . . . . . . . . . . . . . . . . 49

3.9 Endianness for byte order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.10 Virtual Laboratory webpage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

xiii

3.11 QR codes on the workbench . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

3.12 Elevator webpage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3.13 Virtual button panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3.14 Finite State Machine diagram for simulated CPS communication . . . . . . . . . . . . . . 55

3.15 CPPS State Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.16 Activity Diagram for Conveyor Belt module. . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.17 Activity Diagram for Process Order Text module. . . . . . . . . . . . . . . . . . . . . . . . 57

3.18 Activity Diagram for Elevator module. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

3.19 Activity Diagram for Cylinders module. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4.1 Workpieces used for the laboratory demonstration. . . . . . . . . . . . . . . . . . . . . . . 59

4.2 A bar code with a printed production sequence. . . . . . . . . . . . . . . . . . . . . . . . . 60

4.3 Using a bar code to access the database . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.4 MATLAB application during the production phase . . . . . . . . . . . . . . . . . . . . . . . 62

4.5 MATLAB application during the quality control phase . . . . . . . . . . . . . . . . . . . . . 62

4.6 Workpieces testing during Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.7 Database management GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.8 E-stop demonstration, triggered by workspace breaching . . . . . . . . . . . . . . . . . . 65

4.9 E-stop button usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.10 A CPS collaborative scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

xiv

Glossary

BM Business Model

CAD Computer-aided design

CPS Cyber Physical System

DVR Digital Video Recorder

I4.0 Industry 4.0

IT Information Technology

IoT Internet of Things

MES Manufacturing Execution Systems, responsible

for the management of production activities,

from planning to shop floor.

PLC Programmable Logic Controller

PLM Product Life-cycle Management

WIP Work In Progress, refers to a partially finished

good which has yet to reach completion.

xv

xvi

Chapter 1

Introduction

Over the course of history, mankind took advantage of the Earth’s resources in order to overcome the

physical limitations of the human body, comparatively weaker in the speed and strength field on the

animal kingdom. From the use of sharp utensils, the domination of fire and invention of the wheel, the

innovations it created set the human species apart, effectively becoming a major driving force during its

evolution.

Not limited on relying solely on the technical evolution, but also taking into account new resources

that enable the reinvention of the previous, humanity has continuously improved on its industry. Histo-

rians consider that three leaps of qualitative advancements have occurred since the eighteenth century

(the transition to steam powered machinery, mass production and more recently the rise of digital tech-

nology), bringing with it profound effects to the organization of society.

Fomented by the emergence of the Internet in the beginning of the twenty first century, the fourth

industrial revolution is the first of these modernization time periods that has its origins on a technological

phenomenon, the digitalization across all levels of society. The emergent technologies of the fourth major

industrial era blur the boundaries between the physical and digital world, acting as a disruptive force to

the roots of industries and economies alike, with innovation periods happening at an ever increasing

rate.

1.1 Motivation

Since the last 30 years, technological advances in Information Technology (IT) have made computing

devices more economical and widespread than ever. From households to industrial plants, a variety

of devices allow a quick and effortless access to information. In an industrial environment a series of

sensors gather process data which can be posteriorly analyzed with the main goal of productivity in-

crease and preventive maintenance. The quick pace at which data is collected does not allow for current

manufacturing systems to handle with the former. A Big Data problem arises: data is generated at a

faster rhythm that it can be handled. To address not only data issues but also to keep businesses com-

petitive in the new digital world the Industry 4.0 (I4.0) concept was proposed. This new concept makes

1

use of cyber-physical systems, the Cloud and the Internet of Things to aid in the product’s project and

manufacturing phases. The use of Information and Communication Technology (ICT) in this industrial

setting allows for small batches of highly customized products which would not be possible in a classic

factory. Companies who adopt the I4.0 philosophy can remain competitive in a market where costumers

are increasingly demanding not only in terms of involvement during product development but also on

after purchase support. An also relevant topic is the training of professionals to dominate the complex

environment of multi-connected devices. By training engineers with adequate tools in a simulated smart

factory in an earlier formation stage one can educate professionals for the ever demanding and rapid

changing industry.

1.2 Topic Overview

The first Industrial Revolution is characterized by the transition from hand crafting methods to mecha-

nized production, being responsible for the shift between a predominantly agrarian society to one where

industry prevails. Following innovation periods consisted mainly on improvements to the industrial envi-

roment, namely the energy sources and the working aids, as seen in Figure 1.1.

Figure 1.1: The four major industrial eras since the eighteenth century, from [1].

Since the second half of the 18th century, developments on energy conversion techniques made

possible the transition from naturally available sources, such as rivers and the wind, to chemical energy

stored in fossil fuels, mainly coal. Starting in Great Britain, Newcomen’s atmospheric engine, and its

improved design, James Watt’s steam engine made use of steam to drive pumping gear, either to pump

water out of mines or to supply cities with drinking water. Textile industries were one of those who took

advantage on these new machines; mechanical energy could be directly used to power wiring looms,

and waste energy generated during the machine’s operation was used for chemical processes which

required heat. Miniaturization in the following decades allowed these equipments to be installed on

mobile platforms, enabling the transport of people and cargo at speed and convenience. Trains soon

connected major commerce hubs and by improving accessibility, gave rise to suburb’s development.

2

The changeover from hand crafting to machine assisted production with labour division on a factory is

commonly referred as the First Industrial Revolution [2].

In the late 19th century the Second Industrial Revolution started taking place. Soon after the electric

motor invention, power grids were laid out in major urban centres leading to urban electrification. While

streets became safer with public lighting, domestic life improved with household appliances, freeing the

user to seek leisure activities. Industrials plants benefited from electrification, employing AC and DC

currents to power new productions aids, namely the conveyor belt. Also, work conditions improved as

gas lighting was gradually replaced by its electric counterpart, which by comparison produces less heat,

is cleaner and substantially reduces fire hazard. The manufacturing assembly line envisioned by Henry

Ford allowed for high production rate of its Model T, bringing costs down and turning the automobile

into an affordable good for the masses. It was during this era of industrial growth that the middle class

originated, nourished by the improved standards of living.

The arrival of the 1940’s brought with it the transistor, and two decades later the micro-controller.

Early electronic computers such as ENIAC (originally devised by the United States Army during World

War II to calculate ballistic trajectories) relied on vacuum tubes for its circuits. The machine was prone

to frequent tube failure, meaning maintenance works were recurrent. The transistor’s compact design

required less power and space compared to vacuum tubes, eventually replacing the latter as its use in

electronic devices became prevalent in the 1950’s. Further miniaturization led to the appearance of inte-

grated circuits, concentrating in a small area a number of diminutive size transistors that is significantly

higher compared to the use of discrete components. These circuits are several orders of magnitude

faster than designs using discrete transistors, making possible the existence of computers with great

computational power, such as IBM’s Deep Blue which defeated world chess champion in 1997.

In the 1960’s, the rise of electronics also changed the factory’s mode of operation. Two major

types of devices were responsible for the introduction of high-level automation: robots and automa-

tons/programmable logic controllers (PLCs). In 1961, the first industrial robot, the UNIMATE was put in

operation at a GM factory, being used to handle die casts. Soon robotic manipulators became common-

place in industrial settings. Later in the decade, the need for improvement in manufacturing control led

to the creation of a compact and rugged digital computer class of devices, the aforementioned PLCs.

These equipments are designed not only to operate in the adverse conditions found in a factory but

also to function with high reliability. Telecommunications also played a major role: the World Wide Web

opened for public use in 1991, allowing for fast communication between a company’s organizational

levels and other businesses in a scale never seen up until then. Thus, the Third Industrial Revolution is

marked as the time period where electronics and automation made its way to the industry.

The quick pace of technological evolution since the 1990’s decade opened new opportunities for

companies to improve its operation. The most recent trend in the manufacturing landscape is being

named as the Forth Industrial Revolution, also known as Industry 4.0. According to this tendency,

factories will evolve to ”smart factories”; Equipments on the shop floor are connected in a web, allowing

for data exchange between themselves and effortless access to information regarding their state in real

time. The cooperative work between devices and its connection to the Internet allows for costumers

3

to place their orders for highly customized products with a reduced batch size. The flexibility of smart

factories makes it possible for batch size one orders to be profitable, unlike in an assembly line scenario.

I4.0 aims to endow equipments with decision taking capabilities, thus becoming cyber-physical systems,

capable of making choices which is not possible when following a strict algorithm. Also, these CPSs

are designed to assist human workers, and so the acquired production data should be displayed in an

adequate context. Naturally, the increase of data generation rate raises a series of IT issues:

• Potentially sensitive production data is to be broadcasted over the Internet;

• A reliable communication channel is necessary for data transmission;

• Redundant systems should be installed to minimize downtime.

Fundamentally, the Fourth Industrial Revolution has the objective of creating an efficient supply chain

between suppliers and customers, supported by digital data. The information related to this activity is to

be processed, easing its access, while at the same increasing its transparency.

1.3 Objectives

The technological elements brought by the fourth industrial revolution call for new approaches regarding

the training of professionals that will handle the tools to effectively manage an automated and inter-

connected working environment. However, when compared to previous innovation cycles where the

learning rate of some subject was nearly static (an acquired set of skills could provide a worker for life),

the current rate of technology progress occurs in a scale which is non-comparable. Skills obtained in a

short years span could soon become obsolete. Therefore, in order for an engineer to remain relevant it

essential to adopt a continuous learning scheme, not only limited to the technology field but embracing

aspects such as creativity and team work.

In light of this, the objective of this dissertation is to identify the main features of I4.0, their mode of

integration and interaction in a digital and flexible production system and also the basic requirements

for their implementation. Following a selection of the most relevant elements in order for them to be

inserted into a didactic scenario, an experimental setup is to be developed. The workbench simulates, in

a reduced scale, the operating mode of an industrial production facility, implementing the main concepts

of I4.0.

1.4 Thesis Outline

This dissertation is divided into five chapters. Chapter 1 is a preface to the subject that will be later

explored in deeper detail in the chapter that follows. The introductory text to this thesis topic includes

an historical context and a synthesis of the subject, as well as the objectives that are intended to be

achieved.

4

Following this initial chapter, Chapter 2 presents the panorama of events and the motives for the in-

troduction of new technologies into the industrial workspace. After the essential I4.0 ideas are presented,

a literature review ensues, where some testbed platforms for I4.0 are analyzed in terms of their function-

ality and objectives. Following this, the research effort done by different countries and the theme’s key

technologies are presented. Some forecasts about the impact of I4.0, both in terms of business/social

organization and methods of education close the chapter.

Chapter 3 describes the developed experimental set, as well as the academic context where it should

fit. The interfaces and the workbench modes of operation are explained, along with the developed

algorithm for establishing network communication between a PC and the PLC.

In Chapter 4 it is showcased the laboratory usage, with different illustrated scenarios to demonstrate

in detail how the system reacts.

At last, Chapter 5 draws some conclusions on the developed work and suggests some ideas for

follow-up work.

5

6

Chapter 2

Industry 4.0: a journey towards smart

manufacturing

In a recent past, a series of transformations went underway in the methods used for the production

of goods, heavily reliant on the digitalization of manufacturing. Picking on the third industrial revo-

lution, characterized by the introduction of computers and automation technology into the industrial

environment, the fourth industrial revolution plans on improving the digital environment by bringing in

autonomous entities that have a heavy focus on interconnectivity, real-time data and machine learning.

Although the outcome of Industry 4.0 still isn’t entirely clear, major challenges and changes are ex-

pected to occur in society, namely in the organization mode of enterprises and new methods to transmit

knowledge.

2.1 Justification

Ever since its debut in the 1980’s decade the World Wide Web has undergone an intense growth,

connecting people and businesses in a worldwide computer network where information is readily avail-

able. Earlier in the 2000’s decade, in pair with developments in the microelectronic and embedded sys-

tems fields, an extension to the Internet was proposed, designated by Internet of Things. Devices are

equipped with sensors for data gathering and communication modules, allowing them to be remotely

monitored and controlled. This concept has already been applied for a domestic market; A house

automation/domotic system responsible for lighting systems, HVAC and appliances, being controlled

through a user interface on a terminal.

For the industry case, this vision has yet to be fully deployed. Industry tailored equipments, specially

the ones related with aviation, are subject to strict testing and regulations compared with the domestic

ones. Naturally, transition to a new paradigm is more gradual, combined with the fact that upgrades to

existing equipments tend to be expensive and done in a span of time up to 10-20 years.

Production line equipments typical of Industry 3.0 gather data locally and report it to the SCADA

control system, responsible for supervision and process management by issuing commands which are

7

consequently performed by actuators. All acquired data has to first reach a hub, be processed and, if

necessary, a correcting command is sent back. It is one of the most commonly used control systems,

running the productive system through a top-down communication scheme [3]. The automation pyramid,

represented in Figure 2.1, is a visual example of the different levels of automation in a factory, and its

respective integration with one another.

Figure 2.1: Typical control scheme in the form of the automation pyramid, adapted from [4].

As the number of equipments on the shop floor and their complexity increases, information manage-

ment becomes increasingly challenging. A proposed solution for this scenario is an IoT implementation,

with a significant addition. Instead of data processing to be done in a dedicated system, each equip-

ment is fitted with a microprocessor, turning it into a smart device capable of autonomous operation

while coordinating with other devices, thus decentralizing decision making.

Another relevant issue is information display. As manufacturing systems become increasingly elab-

orate, the number of variables to administer escalates. It is in the best interest for raw sensor acquired

data to be converted to context relevant info to ease its analysis.

In addition, a firm driving force for change is the consumer base. As the product’s requirements

increase, and competition intensifies, companies have to innovate in shorter periods of time in order

to remain competitive on the market. Again, the IoT can be used to establish a digital link between

consumers and the company. Orders are to be placed via Internet, and since the smart factory’s equip-

ments are also connected to the same network, production data can be sent directly to them. The

existent interoperability between these devices adds flexibility to the shop floor, allowing the manufac-

ture of customized products which would not be possible under a standard production line.

A tighter integration between businesses and costumers, supported by the increasingly higher levels

of digitalization is a staple characteristic on the Industrie 4.0 program, Figure 2.2.

8

Figure 2.2: Synergies of Industry 4.0, from [5].

2.2 Proposed concepts

Industry 4.0 encompasses a series of principles that an organization should comply in order to succeed

in the Digital Age [6]. At its core are the Cyber-physical systems (CPS), a merge of a physical entity

and intelligent embedded software that is capable to connect to networks and communicate with other

devices, gathering data and processing it to determine the most appropriate action in the context in

which it operates.

• System’s modularity: Designing a line while taking in account modularity allows for enhanced

flexibility. In case requirements are altered, its components can be swapped or configured without

interrupting production;

• Interoperability: People, organizations and machines are connected through the IoT to satisfy

orders of highly customized products. Inside the working environment machines communicate

with one another, not restricted to the ones located at the shop-floor level but also upper levels on

the organizational pyramid, such as management and supervision;

• Virtualization: Physical processes can be monitored through sensors installed on the factory

equipments. Generated data is relayed to a cloud storage service, where it is used to build a virtual

world copy. This model allows for a holistic view of running processes, where information can

be conveniently accessed. Supervision work improves as engineers are notified when a certain

component needs to be intervened. This virtual environment is also a suitable platform to study

alterations to be made at the production line since it can be simulated without interruptions and

the consequent costs associated with it. Heuristic models are used with previously collected data

to predict likely failures, thereby allowing an effective prevention.

9

• Real time capabilites: With cloud technology and interoperability combined, machines collect

production data autonomously, therefore eliminating human error. Besides being used for virtu-

alization purposes, collected data can be useful for entities external to the organization, such as

suppliers. Through the Internet managers have seamless access to information, assessing inven-

tory needs in real time and plan deliveries accordingly.

• Information transparency and contextualization: Data gathered from the sensor set is most of

times not conveniently formatted in a form that is useful for analysis. Acquired raw data is often

incomplete and lacking clarity, which impairs a fast diagnosis in case of an equipment malfunction.

This raw data undergoes a transformation, becoming more relevant depending on the context

where it is required. Using as example a machine’s diagnostic, from its multiple operation variables

a selected few are narrowed down to be more likely as the malfunction origin. Their related data is

to be presented in a useful way such as graphs, in contrast with an array of numbers.

• Decentralized decision making: Equipments undergo a digital transformation, being comple-

mented with communication and decision making abilities. Limited intelligence can be achieved

with embedded computers and accompanying algorithms, which is put to use with the objective of

placing the control responsibility across the devices themselves, therefore eliminating the need for

a central control hub. The forecasted high demand for customized products requires an intricate

central control system, whose complexity would prove to be a management challenge;

• Working aids: In a smart factory, people and machine’s workspace blend, in contrast with cur-

rent production sites where the robot’s working envelope is out of reach from engineers. Moving

equipment such as robotic arms are to be fitted with vision sensors that allow them to perceive

the surrounding environment. This allows for cooperative work between the two entities; a robotic

arm can assist an individual in physical demanding tasks such as lifting heavy loads. Assistance is

also to be provided to the engineer himself, based on augmented reality technology and wearable

devices. Information is displayed with the aim of assisting the engineer during the assembly pro-

cess, guiding him step by step. In case an error is made it is indicated and a solution is provided

to correct it. Besides displaying production data, the engineer’s health status is continuously mon-

itored with two purposes: physical injury prevention and data gathering that will be used to design

an ergonomic work environment. [7]

• Services as products: The IoT fostered the growth of a new market type, one where companies

offer their goods as a digital utility contrary to the traditional business mode of selling a physical

item. Both the data storage and its processing is done on the company’s side. Such solutions

are designed to deal with Big Data, a consequence of the workspace digitalization. Data manage-

ment/analysis can be done via Web pages, acting as the interface between the costumer and the

company.

10

2.3 Training systems: state of the art

Mainly located in Germany, several education institutions are now starting to develop and implement

their own smart factory, albeit in a smaller scale when compared to their real counterparts. One of these

premises is the MyJoghurt demonstrator, located at the Technical University of Munich [8]. This facility

acts as a test-bed for custom yoghurt manufacturing, being part of a wider network that aims to simulate

the several stages of yoghurt preparation, depicted on Figures 2.3 and 2.4.

Figure 2.3: The MyJoghurt demonstrator, adapted from [8].

(a) (b)

Figure 2.4: a) A robotic arm picks up containers to and from the belt b) Details of the units filling stationSource: [8]

11

The system on Fig. 2.3 corresponds to the aforesaid demonstrator, while the packaging and sales

processing facilities are located in Hamburg and Magdeburg/Stuttgart respectively. Being geographi-

cally distant, an robust network linking the facilities is required in order for data to be exchanged without

disruptions. The facilities location and their accompanying network serves as an example of one of

the possibilities enabled by I4.0: Interoperability of these plants. Even if they don’t belong to the same

organization, different systems connected through the IoT can communicate with one another and col-

laborate with the aim of delivering a customized product for the client. The first interaction with the

demonstrator consists of ordering a product, by a costumer, for an yoghurt with filling and toppings

according to his preference. By allowing the costumer to place their orders directly through a digital

platform, value is added in the horizontal chain, since the order placement is more straightforward as

opposed to contacting a sales representative by verbal or written means which require manual process-

ing. Another advantage of this system is the deeper integration of business partners into the company’s

environment. Interactions can be made more dynamic via the digital supporting infrastructure, allowing

the costumer to track its order as it progresses through the required production stages. Moreover, hav-

ing a digital link between the company and external entities ensures that data flow is continuous from its

source until where it’s required, greatly improving processing times and dismissing intermediary human

intervention, liable to errors.

After an order is received, and before production starts, several queries are sent to the modules that

comprise the system, such as if the amount of available ingredients is enough, if the route chosen for

these is adequate as to avoid blocking the distribution tubes and if the conveyor belt speed is suitable

in order to comply with the deadline. After all these requirements are met, yoghurt production can then

start. An example of real time data usage is seen during the pre-production stage, as the different

modules notify the MES about their current status, which allows for a more efficient resource planning.

The MyJoghurt demonstrator was designed with a modular approach in mind. The modules that

make up the system, such as a robotic arm, a barcode scanner and the conveyor belt, all have access

to the information regarding the current job and their own technical data. Thus, the modules can operate

reliably since its limits are known and therefore shouldn’t be reached. Since operational data of the

current task is available in real time, the modules are capable of cooperating in order to make intelligent

decisions which improve productivity. While the system is occupied with the current task, the following

order is analysed so that to check if the incoming one can be processed as soon the current order

finishes. If not, required tasks that would only be carried out after the current order finishes can start

while the system is still operating, therefore reducing idle time. As an example, while the fruit dispenser is

active the fruit pieces required for the following order may require an additional processing step(cut down

in smaller pieces) so as not to block the filling tubes. A significant advantage of having a modular system

is that equipment changes or software updates are considerably simpler to achieve when compared to

a regular industrial installation. In the latter case the whole system usually has to be entirely shut

down and an interface for the new component to be brought in has to be developed so that the legacy

software already installed on the plant can work with the incoming one. If a plant has a modular design,

its components are hot swappable, consequently they can be removed or installed without interrupting

12

the line operation. Whenever a new component is introduced, the other modules analyse the information

about the newly introduced one and integrate it automatically.

In addition to the demonstrators that aim to showcase the several features an I4.0 facility should

possess, there exist another type of premises with the aim of training engineers that have the skills

required to tackle the challenges brought in by this new industrial age. One of these training grounds

is the Process Learning Factory CiP located at the Technical University of Darmstadt [9]. 500 square

meters are available to simulate a realistic production environment that covers all stages of manufacture,

not limited to shop floor activity, but also extending to the IT sector. Two products are assembled, a

pneumatic cylinder and an electric motor, with up to 10 and 4000 variants respectively. The above items

are to be produced in 4 lines, 2 equipped with machine tools while the remaining 2 consist of assembly

oriented equipment.

A variety of digital aspects, fundamental in an I4.0 company, have been incorporated into this learning

factory. The product variant is defined a priori by the client in a configurator, and the respective data is

stored in the parts that make up the product, thereby acting as an information carrier. The way this is

accomplished is through a RFID tag that accompanies the product throughout the assembly process,

uniquely identifying it. Storing its data online allows for trainees to retrieve the manufacturing data, as

well for costumers who can track their order. As the WIP travels through the line, upon reaching each

one of the workstations the RFID tag is read and according to the product’s variant the most adequate

tool is indicated to complete the production steps on that workstation. The WIP acts as a Smart Product,

capable of displaying its production status, interacting with its surroundings and altering it by selecting

pertinent working aids so that the engineer workspace is context oriented to the item currently at work.

After production is complete, a quality control is conducted and its resulted is stored to the aforesaid

RFID tag. A totally digital value chain is established between the costumer and the company, originating

with the order placement and terminating with its delivery.

Since smart products and the CPS that form the smart factory make available the data related with

its own and the production status, it is possible for a digital copy to exist on the Cloud. With the support

of this virtual twin, trainees can manage in real time the factory’s shop floor activity. If a line needs to be

modified, such as when an issue occurs, changes can be first studied on the twin so that the transition

to the new configuration is done with a minimum downtime.

Being CiP more directed towards training, a focus is given not only to the digital features but also

on problem solving which could be applied in a real production environment. Alike with the MyJoghurt

demonstrator, the CiP also adopts a modular structure, facilitating the task of exchanges or adjusting

the equipments layout, Fig 2.5. Trainees are given the opportunity to apply Lean philosophy methods

in order to reduce waste in a setting where they are free to test their solutions with no risk nor cost

associated.

A demonstrator variation with a hybrid objective of acting as a research and demonstration test bed

is presented in [11]. The Smart Factory laboratory at MTA SZTAKI is a facility that performs physical

and virtual processes in industrial manufacturing such that these can be explored in scenarios close

to real conditions. IT based solutions can be tested on this platform that reproduces a scaled-down

13

Figure 2.5: Trainees changing the stations layout on the CiP Learning Factory, from [10].

version of a manufacturing site, including those that involve I4.0 concepts. The platform also acts as a

promotion agent by exposing CPSs and their relevance to the public. Academic usage is also suitable,

providing students with technical higher education and hands-on experience. The laboratory is prepared

for further expandability through the contribution of students projects.

A manufacturing scenario is simulated, where each uniquely identifiable workpiece undergo pro-

cesses of stamping, drilling/punching and a human aided operation. Products are made diverse by dif-

ferent stamping patterns, and extra variation can be added in the manual step. If product diversity can’t

be exhibited, custom data is stored on an RFID tag that accompanies the piece as it travels through the

platform conveyor belt.

The site is comprised of four workstations with identical physical configurations, being connected by

the conveyor belt, Fig. 2.6. Users can place orders, track the production progress and check, upon

delivery, if the production steps where executed correctly. The IT systems may be tested for robustness

and resilience through the introduction of disturbances such as a shortage of resources. Each one of

the workstations is controlled by a Festo PLC which is acessible via a local network; communication

with auxiliary equipment is possible through the use of free I/O channels. A future upgrade is planned,

consisting on the installation of RFID readers and human-machine interfaces.

The workpieces can be tracked by means of NFC tags, making each one of the pieces uniquely

identifiable. In addition, 752 bytes of extra memory are allocated for storing product data. The chosen

tags are smartphone compatible, facilitating the use of apps for product data inspection. A deeper

involvement with production information is attained by means of a projector installed over the desk

shared between the operator and a manipulator robot. In this configuration the working surface and

a visual interface are merged as a single entity, as opposed to the typical setting consisting of a fixed

screen with limited size close to the operator’s area of action.

The Smart Factory at MTA SZTAKI covers a simplified version of relevant processes commonly found

in the manufacturing industry, while retaining a close physical representation. Its components are aware

14

Figure 2.6: The production cells at MTA SZTAKI, from [12].

of context, autonomous, and are able to interact both with physical or virtual entities present in the IT

infrastructure. These components are an expression of the CPS paradigm, with its behaviour being

adjusted according to the surrounding environment; Interaction develops in two directions: sensors

acquire states in real-time, and the processes are affected (virtually) by actuators commanded by the

virtual subsystems.

PLCs are currently present at the core of industrial automation solutions. The predicted changes

brought by I4.0 aren’t expected to alter the importance of these controllers, which will remain a key

piece in manufacturing facilities. Nonetheless, new production requirements call for a revision on these

controller features [13]. PLC controllers operate with proprietary software, some standard within the

automation sector, such as Modbus TCP and Profinet which are used in networking. Technologies

extensively used in the Web have limited integration in these equipments; In recent years manufacturers

begun to adopt some of these technologies like web servers and HTML pages in order to provide PLCs

with services necessary to tackle new requirements. Existent web browser access (via HTTP) allows

to read and write data to the control program. However, solutions are adapted to each controller (with

proprietary software) and therefore using an open-source web interface directly on these equipments is

not feasible. Systems interoperability is also limited, as additional modules that have to be added to the

PLC also use proprietary technology.

The service paradigm was implemented on this testbed, where the PLC controller acts as a I4.0

component by providing an interface (by means of its own web capabilities) to a web-oriented automation

system. A miniature production line (comprised of four stations) for automotive processes was upgraded

with this technology, where each PLC is allocated to one station. Additionally, a diagnostics webpage

15

that can be accessed by a mobile device was developed. Using this configuration, local process data

can be made available to the Internet and shared with other devices, effectively turning the PLC into a

CPS.

A small-scale production system with wireless and product customization capabilities was tested on

[14]. This compact equipment consists on a serial line equipped with sensors to gather data concerning

each individual machine and at system level (proximity sensors based on sound, electromagnetic induc-

tion and capacitance sensing). A PLC, which has the function of serving as the system controller, is

connected to a wireless programmable sensor, a tool for building prototypes which rely on sensors for its

operation or applications for the IoT (usage as a service). Data generated by presence sensors and the

ambient variables(temperature, pressure and humidity) are integrated in a network for data exchange

(with Cloud storage) and monitoring. Attached to this network is a RFID communication module, made

available to smart devices through a wireless router. This setup allows for the PLC to be accessed by

these devices for diagnosis or to read controller data. Two modes of operation were implemented, unit

identification mode and unit selection mode. In the first mode the product proprieties are retrieved by its

accompanying RFID tags, having been previously subject to analysis by the system sensors. Parts can

be more easily distinguished as information is readily available to the production system. In the second

mode the user is queried to input the production order, consisting on the product type, and respective

quantity and orientation. While production is on course the unit proprieties are compared to those stored

in a Excel file (corresponding to the order details) for manufacturing decisions. Upon delivery, a RFID

antenna writes a note to the tag stating that all manufacturing steps where executed correctly. Defective

parts are rejected, with an nonconformity note being also written to the tags. By implementing RFID

technology, information becomes decentralized, enhancing the manufacturing setting intelligence. Hav-

ing the workpieces carrying their own information eases computing resources, freeing them to facilitate

production processes.

The Kaiserslautern smart factory, detailed on [15], is presented as being an I4.0 production plant

that is manufacturer-independent, made possible by defining uniform interface standards that enable a

link between each manufacturer’s systems. The standards allow for a flexible system expansion. These

include a structured data scheme for the product RFID tag description, the usage of OPC UA communi-

cation that allows for machine collected data to be forwarded to a cloud platform and the standardization

of hardware, meaning the constituting modules have an identical mechanical functionality.

Regarding the portuguese efforts to experiment with new manufacturing technologies, a number of

entities have already presented some studies on this subject. On [16], the authors present a multi-agent

control system intended to be used on a shop floor assembly scenario. This system, named NovaFlex

(installed at Uninova), is composed by two assembly robots, a warehouse and a transport system that

links the aforementioned modules. The usage of agents to control each component permits to circum-

vent the limitations of each component’s control methods, that may be reliant on legacy connections.

This multi agent application enables for components to be introduced or removed to the system, and

demonstrates that through the use of a software interface legacy equipments can be introduced to the

system. Another example of this nature is presented on [17], consisting on a multi-agent manufacturing

16

control system that adopts an holonic architecture, with the experimental validation undertaken in a real

laboratorial manufacturing system at the Polytechnic Institute of Braganca. This installation is based

on the flexible manufacturing system of the Computer Integrated Manufacturing (CIM) Center of Porto.

An holonic control architecture relies on the existence of entities characterized by capabilities of auton-

omy and collaboration, being named holons. This entities can represent either manufacturing resources

(such as tools) or other items such as product orders. A global self-organization of the system is in-

tended, achieved by propagating information between individual holons, in addition to the advantages

brought by a multi-agent approach (expansibility and reconfiguration). The usage of a control archi-

tecture of this type uses adaptive production control and takes advantage of holon’s self-organization

capabilities to improve the agility and reconfigurability of production systems. INESC TEC technological

demonstrator, the iilab, is a laboratory dedicated to advanced manufacturing technologies, including I4.0

[18]. This facility aims to publicize and demonstrate concepts and the technologies used towards the

digitalization of manufacturing. One of the laboratories divulged features are its collaborative robots,

with 12 units being used to demonstrate applications in a variety of sectors, such as automotive and

aerospace industry.

The transition to the new digital era is not limited to academic environment. Several organizations

have begun in recent years to experiment with tools that allow them to not only reduce costs from

product development to the production stage, but also to expedite the whole process. In collaboration

with Siemens, Maserati remodelled its Avvocato Giovanni Agnelli plant in Turin to bring it up to date with

the latest tendencies in information technology and manufacturing digitalization [19]. Two of Maserati’s

models are assembled here, the Quattroporte and the Ghibli, with the latter being available with more

than seventy thousand variants. Naturally, mass production isn’t a suitable option for luxury vehicles,

thus these being a good case for batch size one production, where every costumer gets a customized

variant. The Ghibli components were designed in Siemens’ NX CAD software, where the entire design

was also digitally assembled. The resultant models were imported into the Tecnomatix software, a

tool that aids in projecting the production line (comprised by three sections, body, paint shop and final

assembly), Fig. 2.7. The software allows for the production processes to be simulated, subsequently

indicating areas where resource usage improvement is possible; a complete 3D model of the processes

is the program’s outcome.

By using Siemens’ PLM software (Teamcenter), the company has at its disposal a digital twin of the

Ghibli, a totally faithful computer copy of the car. Having this twin is a clear advantage: data from the

virtual and real models can be combined in order to speed up testing. A third of the total development

time was reduced, as consequently, monetary resources used during testing also diminished. One of

the development stages where this can be done is during wind tunnel testing, where data gathered with

the physical prototype can be used to perform virtual testing on the digital twin, thus reducing the costs

associated with running the wind tunnel machinery each time adjustments are made. Besides aerody-

namics testing, finer details that are associated with a sense of luxury that the brand aims to transmit can

also be tuned, including the sound of closing doors and the engine inside the passenger compartment.

Similarly to the wind tunnel testing, real world data was gathered with a microphone and later used in

17

Figure 2.7: The digital twin of Maserati’s Avvocato plant, from [20].

virtual tests. The effort to achieve a fully digital environment setting isn’t circumscribed to the factory

level. Horizontal integration is achieved as suppliers are linked to the organization digitally. A request is

made to deliver the required parts to each one of the cars with a timing window so that they arrive to the

plant just at the moment they are needed, thus reducing costs associated with storage. By converting to

digital all the activities of the manufacturing process, Maserati can deliver a new model to the market in

nearly half the time that would be required had it not adopted a digital strategy, digitalizing all functions

throughout all levels of their processes. Moreover, a batch size of one production is accomplished that

is also cost-effective.

2.4 Research

A number of countries have begun to invest into the computerization of manufacturing, with the aim of

improving industry competitiveness, with the prominent leaders being Germany with its Industrie 4.0

initiative and the United States with the Smart Manufacturing program. Up to 200 million Euro and

100 million dollars have been allocated for research in these programs, respectively [21]. Germany’s

I4.0 platform has identified crucial areas that are considered essential for a successful transition; with

a technological scope, Reference Architecture, Standardisation and Networked Systems Security. The

remaining, related to human resources and renovation philosophies are Legal Environment and Work,

Education&Training [22]. Manufacturing intelligence is the main focus topic; at its core are the CPS’s and

possible approaches to integrate them into a production system, turning each one of its components into

a CPPS: a Cyber-Physical Production System. These entities are the result of the merging of systems,

mobilized from diverse area’s knowledge, such as mechanical and electrical engineering and computer

science.

18

2.5 Supporting technologies

A number of key technologies, Figure 2.8, are fundamental for the implementation of the digital manu-

facturing paradigm, that can be considered the building blocks of I4.0.

Figure 2.8: Building blocks of Industry 4.0, from [23].

2.5.1 Cyber Physical Systems

IT systems, having profoundly shaped society behaviours starting in the last decade of the twentieth

century are now initiating a period of transition for industrial machinery. Production systems are to be

revamped with modern information and communication technologies, greatly benefiting the operation of

manufacturing plants where they are set up. These upgrades aim to outfit machines with decision mak-

ing capabilities and subsequent order executing within the system own limits, so that an autonomous

operation can be achieved. The drive for a low batch, inferior priced products calls for a flexible pro-

duction system that is able to cope with the rising complexity of information processing [24]. Dynamic

demand plays a determinant role, as the market requirements rise; costumers seek to acquire individu-

alized products (increasing the number of variants and its associated data), while remaining affordable.

To that effect, I4.0 plans on introducing CPS, intelligent machines that gather information about them-

selves and the action they are currently performing, subject it to processing software and take actions

accordingly, Fig. 2.9 a) . It is also foreseen that these devices are capable of a certain level of ”social-

ization” through the use of networks, Fig. 2.9 b) . Sharing information between machines allows for an

overall better operation of the plant, decentralizing information and thus facilitating its access.

CPS’s usage aren’t limited to production related aspects. Logistics is one of the areas that can ben-

efit by introducing these systems to management departments; a CPS employed in this field is dubbed

a Cyber-Physical Logistics System. These are tasked to apply general principles of lean production,

19

essentially promoting waste reduction. An application scenario is presented on [25]. The remaining pro-

cessing time of machines is used to forecast completion dates, and consequently, demand for materials

to be delivered by a Mizusumashi, a supplier operator responsible for delivering materials to worksta-

tions. Combining information about the machine’s occupancy state and the pick-up places, either for

raw materials or finished goods, an optimal departure time can be determined that minimizes movement

waste by the Mizusumashi operator.

(a) (b)

Figure 2.9: a) A CPPS operating autonomously, retrieving assembly information through the digitaltwin. Source: [26]; b) CPS structure and interactions. Source: [27]

2.5.2 Internet of Things

The expression Internet of Things was first used in 1999 to name an environment of objects sharing its

data across the Internet [28]. While the initial concept referred to the idea of using RFID technology to

uniquely identify items in a supply chain, it later extended to a wider scale. The concept isn’t limited

to standard computing devices, as equipments that traditionally lack networking capabilities are given

internet connectivity. These objects collect information from the environment and are capable of inter-

acting with the real world, with common Internet standards being used to transfer data [29]. A subset of

the IoT, named the Industrial IoT, is used to describe the more generalist concepts of the IoT by applying

it to a real industry scenario. While the IoT is already present to an extent in home automation the IIoT

refers to the interconnectedness between smart factories, management systems and machines, with a

wider scale compared with the household scenario.

Open radio technology, mainly used in short-range protocols, is responsible for the wider growth of

connected devices, with the number of equipments forecasted to be online in 2023 to be 70% greater

than the ones in 2017. While devices that rely in cellular connections (wide-area segment) amounted

to just over half a billion in 2017, new supportive technologies are excepted to push this number to 2.4

billions. Albeit this prevision is quantitatively reduced compared to the short-range one, a compounded

annual growth rate of 26% reveals the potential for expansion in this sector, Figure 2.10.

Contrary to previous cycles of improvements, a consumer technology is determined as major driving

force for industrial communications. Common cellular protocols in use in the wide-area segment are

20

Figure 2.10: Connected devices forecast, from [30].

GSM and GPRS, both belonging to the second generation cellular technology launched in the early

1990’s. Although their low cost and good coverage make it a popular solution for machine to machine

communication, its ageing design make it unsuitable for cope with the IoT requirements. 5G cellular

technology is expected to be a supporting asset, as its low latency and high bandwidth connections

make it an adequate candidate for supporting IoT applications [31].

Mobile Internet hasn’t had a major impact in industrial communication, as consumer oriented proto-

cols aren’t suitable for the industrial market. A partial market merge may occur as telecommunications

service providers deploy specialized solutions that are based on the ones developed for general con-

sumers. Such an example is Time-Sensitive Networking, a set of standards that aim to surpass the

limitations of classic Ethernet interfaces, being originally developed by the IEEE group for audio and

video consumer applications, and later adapted to meet professional standards.

For the IIoT concept to materialize, appropriate web standards should be defined; support for legacy

systems is also relevant, meaning that multiple protocols should coexist.

2.5.3 Big Data Analytics

In the mid 1980’s less than 1% of the worlds information was stored in digital format, amounting to 0.02

exabytes. In a time span of 30 years this value rose to nearly 5 zettabytes (5000 exabytes) in 2014, with

the digitalization of information resulting in less than 0.5% being stored in analog format. At an annual

compound growth rate of 30% (with digital storage growing at twice the speed of that of analog) only in

a relatively recent past, the year 2002, has digital storage surpassed analog, Figure 2.11.

As machines get interconnected and data acquisition becomes ubiquitous with ever lower sensor

prices, CPSs are excepted to produce and transmit a considerable amount of data related to the pro-

duction process [33, 34]. Not limited to managing large amounts of data, the challenges extend to its

storage, ease of access, privacy issues and the need for a good performance on real-time data process-

ing. As early as 2001 the advisory firm Gartner Inc. defined Big Data as datasets that embody the 3 V’s

[35]:

21

Figure 2.11: Amount of data, from [32].

• Volume, referring to the ever-increasing size of available information;

• Variety, referring to the range of types data can shape and its sources;

• Velocity, referring to the celerity of capture and processing.

Value needs to be extracted from machine generated data for it to be meaningful; a considerable

amount of it originates on sensors, that have had recently an increase on the requirements for measure-

ments quality, as well as decrease in price [36]. Methods such as predictive analysis are of particular

interest for the industry, as by using production history to construct models predictions can be made

regarding tool wear or failures, therefore enabling a timely intervention [37].

Current manufacturing systems aren’t prepared to manage with Big Data as they lack appropri-

ate analytics methods. Traditional data management systems are based on relational databases, with

non-existent support for storing unstructured data. Online processing limits the velocity of industrial ap-

plications; a continuous scaling isn’t a solution as the expensive hardware they depend implies that a

large scale expansion isn’t economically feasible. Network related technologies such as cloud comput-

ing present themselves as service that offers the required flexibility and cost-effectiveness for supporting

decision making. Although cloud technologies perform fairly on Internet related sectors, additional ad-

justments are to be made in order to deal with the complex industry requirements [38, 39].

Following the data analytics phase, results are to be presented on the visualization phase. An uni-

versal approach isn’t suitable as different users will have particular requirements; different levels of

22

complexity can be achieved, starting from a detailed one as low as a machine tool level up to a more

generalist one, presenting an overview of the supply chain level.

By taking advantage on Big Data valuable outcomes can be collected. Costumers are one of the

entities to benefit from data mining, as by understanding their needs and anticipating it a better product

or service can be provided. Product development also benefits from this, as information on the product’s

usage and user feedback is sent back.

An efficient data management and distribution is therefore essential for the interconnected smart

factory vision to be realized.

2.5.4 Human-robot collaboration

Cobots, an abbreviation of collaborative robots, are robots which are designed to operate alongside a

human worker, sharing workload between the two. Although the concept isn’t new, with the first collab-

orative robot being introduced in December 2008 by Universal Robots, the production rhythms growth

and increasing demands from consumers have stimulated the interest in this class of manipulators [40].

Traditional industrial robots are designed to perform a repetitive task in a fast pace and high precision,

without the need for collaboration in a shared physical workspace. As such, safety measures are needed

to ensure human safety; barriers that prevent workers from entering the robot’s working envelope are

a common solution. Limited interaction is possible, provided that the robot’s speed and torque are

reduced, and its workspace (working envelope) is restricted [41].

Cobots act as an alternative to industrial robots, rather than a competitor, and a complement to

workers activity. On an ever demanding market, having a hybrid workforce is a major solution for ensur-

ing high productivity. The adoption of collaborative robotics in the workspace allows the coexistence of

automated and manual processes, useful in situations where complete automation is not possible. More-

over, keeping the worker as an workstation element is an opportunity for introducing flexible intelligence

[42, 43].

Collaboration between robots and workers enhances productivity of the later by freeing him of repeti-

tive tasks, while simultaneously reducing fatigue and stress[44]. Although operating speeds are typically

lower compared to traditional robots, preventive measures have to be devised to ensure the operators

safety. Common solutions rely on acquiring periphery awareness, through the use of cameras that

detect the operator’s location. By assessing the environment, including the robot’s status, appropriate

responses can be taking before a potential dangerous situation takes place.

New interaction methods that replace the traditional joystick and buttons are proposed, some being

hands-free. Haptic control recreates the sense of touch by transferring forces and torques felt by the

robot to the user. Speech recognition can also be used to verbally control the robot. Hand movements

for joint manipulation and head movements to confirm actions are a possibility on gesture recognition

[45].

A CPS application scenario where the operators flexibility and the robot’s accuracy are combined is

depicted in Figure 2.12, along with possible interactions.

23

Figure 2.12: A collaborative (human-robot) assembly scenario, from [46].

2.5.5 Augmented Reality

Contrasting with Virtual Reality, where the user is immersed in an artificial environment without the

possibility to see or interact the real world, Augmented Reality (AR) technologies provide an interactive

experience with the real world by overlaying information upon it, either in the form of graphics, sound or

touch [47].

Different types of user interfaces have succeeded over the years, with the aim of turning interac-

tion more straightforward and thus easing access to information. Command line interfaces gave way to

Graphical User Interfaces (GUI), exempting the user from memorizing sets of commands. Instead, the

user interacts with a visual representation of commands and objects, enabling for a more comprehensi-

ble visual feedback.

Although the ease of use of these interfaces have helped to ensure a long presence on industrial

sites, changes on the way information is dealt calls for a review on human-machine interfaces. The

information that is displayed on a GUI interface is removed from its context in relation to the real world,

forcing the user to interact with the graphical interface as opposed to the item that the information relates.

The dynamics of an industrial environment combined with the ever increasing amounts of data related

with the production process justify the introduction of new support systems on the workspace that aim

to turn the high data volumes into useful information to guide engineers.

24

Augmented reality technology can be used as a tool to improve the methods used in industries, by

delivering real time information depending on the action that it’s being performed. Common deliveries

of this technology are achieved through eyeglasses or by a mobile device equipped with a camera.

Depending on the user’s location and spatial orientation, these devices display the information directly

over the object of interest, therefore removing the need for an intermediary interface and enabling a

more direct interaction.

The technology’s potential has spurred the interest of companies who are studying the introduction

of AR to support workers. On Figure 2.13, Audi demonstrates the use of AR on its Smart Factory on

guiding a worker during an engine assembly operation. By providing information through schemes and

drawings over the region of interest as opposed to printed or static media, understanding of the tasks at

hand is made more intuitive. The capacity for flexibilization and adaptation is also enhanced, freeing the

worker from the task of having to search for or interpret new instructions.

Figure 2.13: A worker being given instructions in an Augmented Reality environment, from [48].

2.5.6 Electronic tagging technologies

A dependable IT implementation is decisive to support I4.0, since the existence of a digital network

connecting objects of a diverse nature is at the core of I4.0 principles. One of the communication tech-

nologies currently being considered to connect a smart factory components, and the goods within it, are

RFID chips. Data stored in these chips enables for individual products to exist in a mass production

environment, since each products characteristics can be sent directly to machines, as opposed to a

predefined set of actions. Product data is readily available by performing a tag scan, making catalogu-

ing more effective [49]. After the production stage, it can be tracked through the supply chain until its

designated destination thanks to real-time updates. In some cases the RFID tag acts as bridge between

25

costumer and supplier, enabling the later to monitor its usage and provide additional services such as

support, maintenance and proper disposal when its useful life terminates. RFID and NFC (the latter be-

ing a subset of the first, characterized by possessing peer-to-peer communication) usage is growing on

the market, as its wireless capabilities, versatile distance communication and high transmission rate are

valuable assets that can address the different requirements demanded by the application or environment

[50].

RFID chips rely on electromagnetic fields detection for information exchange. An antenna present

on the reader emits a radio signal that activates the tag’s integrated circuit, making possible for data

to be read or written to the tag. Due to using radio waves, the tag need not to be on the reader’s line

of sight; as long the tag is present in the electromagnetic field produced by the antenna the tag will

be activated, a clear advantage against optical readers. Naturally, operating rage is dependant on the

antenna emissive power and used frequency, with common range reaching 3 meters [51]. Following the

tag’s activation, a transponder is responsible for modulating the tag’s information, which will be received

by the reader. Subsequently, the decoded information can be sent to a computer for further processing.

Active RFID tags, powered internally by a battery, are capable of transmitting at a higher distance when

compared to passive tags that rely on the energy transferred from the reader solely. These long range

tags, known as RAIN RFID have a range of up to 15 meters and higher recognition rates than passive

tags [52].

2.6 Revisions and implications

The ongoing phenomenon of manufacturing digitalization is currently the study object of several com-

mittees, responsible for clarifying the current transformations and its expected impacts, along with rec-

ommendations for implementing a series of policies that were identified by the governments under a

research and development programs. The first dedicated report was released by the I4.0 platform on

April 2015, and presents the importance of this program to the economy. This program’s road map spans

until the 2030’s, and being a long term strategy, the industry transformation is still in its initial phases.

2.6.1 Corporate

Based on the current industry status, the following findings were presented by the I4.0 Working Group

and the ITRE committee [21, 22]:

• Only a minority of businesses are prepared to manage the technological challenges brought by

I4.0. SMEs are particularly vulnerable, due to the lack of specialized staff and the adoption of a

conservative attitude towards new technology; Moreover, the participation in digital supply chains

by SMEs could prove difficult due to the associated costs and risks, and the reduced flexibility and

independence this entry would cause;

• Relevant areas such as digital security, systems operability and health and safety present signifi-

cant challenges, with the accompanying costs and risks. Companies may not be willing to invest

26

if their resulting work could be duplicated by third parties who didn’t have to spend resources into

research and development;

• Although research is well supported, its outcomes lack in implementation.

It is notorious that the high investment cost serve as a major barrier towards transforming manufac-

turing, particularly for companies of small dimension that many not be able to afford such expense. One

solution lies in changing the business model, by delivering a digital good integrated in value chains that

build relationships with the environment.

The increased relevance of data is predicted to shift some organization’s orientation, basing their ac-

tivity on the idea of selling a service (as opposed to base its revenue on product sales), or a combination

of a product and a service, resulting in a hybrid solution. An example of a business extension to ser-

vices while maintaining its current model is presented on [36]. A supplier for the aviation sector provides

solutions for aircraft surveillance, assisting crew members in monitoring the cabin during flight. Video

captured in-flight is stored in a DVR video unit that as already been certified by aviation agencies, being

subjected earlier to strict standards and extensive testing. An add-on for the aircraft surveillance system

is offered through a ground station, that allows users to view and export earlier captured streams. As

this extension isn’t part of the aircraft it need not to undergo rigid trials, with the associated added mone-

tary expenses; therefore, development costs are lower and alterations are easier to perform. Interfacing

the ground station with the aircraft’s DVR conceives a CPS which has the possibility of providing a web

service, enabling video access on demand through the Internet. Figure 2.14 presents the intervening

agents and their interactions.

Figure 2.14: A web based service for the aviation industry, from [36].

To support the product-service system, a tighter integration between engineering and computer sci-

ence is foreseen. Software engineering plays a major role on fulfilling the operating requirements for

software systems [53]. It is recurrent that these requirements are not fully defined; CPS’s own nature

27

characterized by the interconnectedness of systems hinders the task of development. By separating

knowledge departments the solution is often deemed as fragmented, as a CPS propriety isn’t the sum

of its individual property; the system interaction with their results also has to be considered [36].

The interconnectedness of organizations and systems calls for an intervention of IT work on guar-

anteeing a appropriate orchestration of services, applications and their corresponding CPS platform.

The supply isn’t limited to web services to the end costumer, as the existence of a supply for services

setup and applications with the aim of being used in inter-company processes exist in the context of

I4.0. Being service based, and real-time enabled, CPS platforms ought to be regulated as to serve as

a foundation to a collaborative industrial environment, ranging from industrial processes to the life-cycle

support. To this effect, an effort for standardization should be undertaken, where relevant international

standards, supported by policy makers point towards the harmonization of services and BM.

A dynamic business network is also expected to have an impact on current BMs, as companies

depart from a single business network and inter collaboration deepens. For this purpose, continuous

monitoring plays a major role. Documenting processes and reporting statues allow for an clearer veri-

fication if a contractual and regulatory conditions satisfy. As product’s nature shift to servitization, a life

cycle prediction should be laid out as to assure an extended proficient service of quality. New business

partners can be brought to the business network if an adequate arrangement is made, with the proper

accompanying license models. A further characteristic enabled by new BMs is enhanced competitive-

ness, since prices can be dynamically adjusted by analysing the costumers and competitors positions.

Forecasts places industrial components manufacturing, automotive and aeronautical as the main

sectors that will benefit from the I4.0 program. A 6% increase of employment during the first decades is

expected, with low skilled workers being initially displaced, resulting in the increase of unemployment in

the short term; a higher demand for mechanical engineers and IT personnel is predicted [22].

The potential benefits brought by I4.0 on industrial development will need to be weighted against the

risks and the resources invested, since not all entities will benefit equally; particular adjustments are

needed, not limited to a company characteristics but extending to the country conditions their activity is

based on.

With the adoption of its Industrie 4.0 program, Germany seeks to adopt a dual strategy with the pur-

pose of consolidating both the suppliers equipment industry and the user companies [54]. As a provider,

the country seeks to maintain the leading role on the market as a supplier of machinery and plant engi-

neering, combining it with ICT in order to become a provider of smart manufacturing technologies. As

to the market clients, the objective is to combine the producing companies with the ones that act as

providers of equipment/services, thus creating a value creation network.

2.6.2 Social

Regarding social findings, the entities referred on 2.6.1 presented the following:

• Larger firms tend to have a positive attitude toward the I4.0 program;

• Trade unions are more inclined to have reservations about I4.0 outcomes, displaying concern on

28

employment levels, working conditions and workers’ rights;

• An existing skill gap, and increasingly growing requirements for the I4.0 panorama;

• Uneven distribution of skills across the EU, being mainly concentrated on already established

centres of knowledge.

A number of changes in the human resources department are excepted. As the nature of activities

shift from routine to creative, work itself becomes more flexible, allowing for workers to remain productive

for longer. This introduces new challenges, namely in training. Following the trend of ageing popula-

tion, and the shortage of specialized workers, new lifelong learning schemes to cope with highly specific

technologies will have to be devised, as well as additional adjustments on the workplace: health man-

agement, team working skills and work organization. The flexibilization of work has the consequence of

enabling for work hours to be adapted to an individuals scheduling, thus achieving a balance between

work, its private life and the company requirements

Workplace dynamics are expected to be revised, placing greater importance on the engineers posi-

tion in respect to its influence on the process at hand. Workers take an active position, being involved in

participative work design, being themselves responsible for their work load. Naturally, the competence

profiles sought after by employers will change, namely on an higher focus for individuals who actively

engage in the innovation process.

Given the recent nature of I4.0, it still isn’t totally clear what the consequences of its implementation

will be. Current legislation may come across as an obstacle to the quick pace of reforms. Shorter cycles

of measures require an adjustment on the frequency of releases on updated legislation, meaning that

if the mode of applying laws isn’t revised, I4.0 measures may not be applied efficiently. Nonetheless,

these must be in compliance with the legislation, creating a legal challenge by coordinating liability and

the disruptive measures.

2.7 Topics for teaching and training

A paradigm shift requires an education method overhaul, placing greater importance on the practical

aspects and implementation of theoretical concepts. It follows that supplementing delivered content with

practical activities learning efficiency is improved [55].

To complement the theoretical education, in an effort for it to adopt new methods that allow to follow

up with the recent technological changes, new learning setups have been devised that aim to incite

trainees to develop their capabilities that will be useful later while pursuing a career on the industrial

sector. A didactical teaching scenario brings together a variety of different technologies, exposing the

complexity that comes from handling different equipments in the same setup. Depending on their in-

tended use, these I4.0 demonstrators have divergent levels on complexity; the more generalist ones

cover a larger span of I4.0 concepts, while more specialized setups tend to focus on a reduced number

of concepts, commonly related.

29

2.7.1 Comparative analysis of training systems

Numerous manufacturers are now starting to offer industrial didactic equipments which allow to illustrate

most relevant Industry 4.0 concepts. Bringing these new equipments to universities teaching facili-

ties allows to train students for the industry future demands. While some of these demonstrators are

designed to interact with workers, others are configured to work in an autonomous way that requires

almost no human intervention. Therefore, Industry 4.0 concepts that can be illustrated on each one of

these demonstrators differ due to its operating condition. Some of the studied demonstrators are shown

in Table 2.1.

Table 2.1: Industry 4.0 demonstrators and respective features

MyJoghurt Beckhoff RIAN Festo RexrothSmart Factory Didatic MS 4.0

Vertical Integration x x x x xHorizontal Integration x x

Modularity x x xInteroperability x x x x x

Simulation x x xDecentralization x x

Workspace personalization xRFID/NFC x x x x

Collaboration with robots x xAugmented reality x

The MyJoghurt, Beckhoff Smart Factory and RIAN demonstrators all run automatically [8, 56, 57].

These are designed to illustrate a lights-out factory which do not require human presence on site. A

miniature version of such idealization, developed by Beckhoff, can be seen of Figure 2.15. A larger

scale of the same concept was presented at the Automatica 2014 fair trade, shown in Figure 2.16.

Figure 2.15: Beckhoff Smart Factory demonstrator, from [58].

30

Figure 2.16: Roboter Integrated Agent Network, from [59]. The RIAN demonstrator is characterized byconnecting heterogeneous plants in order to integrate them in a cooperating production line. An

autonomous vehicle is responsible to physically connect the plants.

The equipments Bosch Rexroth and Festo supply are intended to be part of a smart factory simulator

where human workers (students or future employees) play a essential role in its operation [60, 61]. An

entity wishing to coach its trainees can build a factory adjusted to its needs due to the modular nature of

this didactic sets, similar to the CiP Learning Factory detailed in Section 2.3. Moreover, these learning

factories may be used for research in a technological ambit or in management domain like organizational

motivation [9].

Following the information digitalization trend, all factory organizational levels are connected by a

network allowing a seamless information flow. Machines on the shop floor are equipped with sensors and

CPUs which monitor the machine’s state and report it to the upper factory levels, namely a maintenance

department. On a descending direction, orders arriving from a project department can be distributed to

the manufacturing equipments. By eliminating the need for printing media, information can be retrieved

quickly and in a variety of personal portable devices, such as a smartphone or tablet. Also, digital data is

less prone to handling errors (paper sheets can be misplaced and lost). All the examined demonstrators

present functionalities that reflect this aspect. The MyJoghurt demonstrator checks if all its modules are

in a normal state and if there are enough ingredient quantities before starting the yoghurt production.

The Beckhoff Smart Factory emphasizes the importance of data collection and its posterior treatment

with the purpose of process optimization. Several human-machine interfaces permit the interaction

between workers and machines, delivering information by request and thus reducing the necessary

effort for information lookup. This data generated by plant operation should be compared with the

estimated production plan, requiring critical thinking by human workers followed by establish conclusions

on plant activity and adequate solutions if necessary. This tendency of combining automation with new

ways of delivering information is designated, according to Beckhoff, by Social Automation. Vertical

integration and interoperability combine so that equipments communicate and coordinate themselves.

31

The transition to a paperless environment needs also to take into account not only data travelling within

the factory, but also information that originates from the value chain established in the horizontal plan.

This value chain arises from the factory need to have external entities to supply it, either by raw material

vendors, subcontractors or costumers orders. On the RIAN demonstrator, costumers can place an order

for a customized bottle opener, which they configure on a personal smart device.

With data acquired through vertical integration it is possible to create the factory’s virtual copy on

the Cloud, known as digital twin. This entity, that can be built based either in real time or in recorded

data, allows to simulate new configurations without the need for physical alterations. Also, heuristics

application on the factory model reveals not only potential sites for efficiency improvement, but also

allow for failure prediction. Festo and Bosch Rexroth cyber-physical solutions are prepared for data

acquisition and solution deployment in a virtual environment. In the scenarios that only require tweaking

certain equipment parameters (e.g.,situations where line efficiency is being studied), the solution can

be readily implemented without human intervention. Besides acting as a maintenance tool, Festo digital

twin can also be used for resource saving while developing new machines for consequent installation on

the shop floor.

Following the simulation and validation phase, alterations can be made on the production line. In the

majority of cases, only sections of the line are being renovated at a time. These updates are designated

by brownfield applications; new equipments are introduced without interrupting older machine’s opera-

tion. For a simpler transition on a brownfield setting it is essential for elements that comprise the factory

to be as modular as possible. The module’s degree of independence impacts the response celerity when

it is necessary to replace a faulty module. Both Festo and Bosch Rexroth factory model are designed

so that the assembly line offers flexibility; By setting up each workstation on a wheeled platform mobility

is enhanced and thus setup times are reduced.

The proliferation of sensors on the working environment allows for interconnectedness of not only

between machines but also between machines and the product. Industry 4.0 places the product as an

integral part on the production line. The product is equipped with an RFID tag and thus is communication

enabled. Instead of defining manufacturing instructions on a MES, manufacturing data is stored within

the product, which feeds it to the equipment who queries for it. Smart products are not limited to storing

manufacturing instructions. Information relevant for the end user can also be recorded, such as sched-

uled maintenance or the product’s respective state (e.g, wear). On MyJoghurt demonstrator costumers

place their order on their personal devices for a customized yoghurt. After filling and topping flavour are

set this information is printed on a barcode. This barcode is scanned while the jar travels through the

conveyor, guiding it to the correct filling units. The ability for smart products to communicate with its

surroundings it’s a key requirement for achieving batch size one, i.e, manufacturing a single product for

a single costumer.

On a cyber-physical factory, each worker workstation is personalized according to its preferences,

such as language on screens or ergonomic aspects like tools placement. Each worker can be uniquely

identified by a bracelet with an NFC enabled wristband or through biometric signatures such as a fin-

gerprint or retina scanner. Additionally, information is delivered according to context. For instance, if a

32

worker is adopting a position or movement that is ergonomically harmful a warning is issued and so-

lutions are proposed to correct it. Each workstation may be equipped with a touchscreen displaying

procedures for the worker to follow in order to complete the current task. Information is context sensitive,

depending on the part’s assembly status or on the workers education level. In recent years new worker

aiding systems have been proposed, as is the case of augmented reality. A worker equipped with virtual

reality glasses can be guided through its task in a more interactive way when compared with instruction

shown in a screen. Since the glasses are also communication enabled, information regarding product

assembly can be retrieved from the product itself and displayed to the worker at a relevant time, such as

tool parameters and correct placement.

In most industrial plants in operation today, robots operate in a fenced off area separate from workers

for security reasons. Industry 4.0 plans to include collaborative robots in the shop floor, assisting workers

in arduous tasks but also in situations where a higher precision degree is required. The presence of a

robot in close proximity to a human worker is only possible due to the use of computer vision technology.

In addition to detecting objects of interest, image processing algorithms can also detect a person’s

location and its respective distance, thus eliminating the collision and injury risk with a worker. Robots

already replace workers in repetitive tasks freeing time for them to pursue added value activities, either

for personal competences or to participate in team projects.

After an analysis on Table 2.1 it is perceived that the didactic solutions directed towards trainees are

the ones that express the most of Industry 4.0 aspects. By using these equipments, students acquire a

holistic view of this new industrial age. In contrast, the automated production systems are projected to

emphasize some particular aspects of a smart factory. The RIAN demonstrator is focused in modules

interoperability and decentralized decision making. This bottle opener production line is comprised

of a Schunk warehouse, KUKA robotic arms for handling purposes, a Reis laser cutting tool and a

FANUC injection moulding machine. All these separated departments communicate with each other

via the Internet. Mobile transportation robots transport the parts in between the different production

sectors. This demonstrator highlights the importance of industrial manufacturers adopting a standard

communication protocol so that all the plant’s components link between themselves as seamless as

possible.

2.7.2 Skills for the workplace

The arrival of the fourth industrial revolution will bring with it a departure from the current industrial set-

ting. A company’s physical structures aren’t the only characteristic that needs to be overhauled, as the

increased usage of digital technologies calls for new skills to the workspace. The trend to increased

automation is predicted to displace some of the low-skilled workers that are accustomed to perform sim-

ple and repetitive tasks. Flexible workforce which is experienced with mechatronics and IT is favoured,

as the tools to gain an holistic and diversified system’s view are present. Abilities related with problem-

solving and resource management need to be addressed by manufactures so that their engineers can

handle with the new situations brought in by the digitalization trend. On [7], the authors aver that I4.0

33

promotes the involvement of workers on activities that emphasise innovation, creativity and communica-

tion, as opposed to routine activities that should at least be partially assured by machines. Therefore,

engineers should refrain from thinking in a black box perspective about the technologies present in the

production systems, instead a critical mind is a crucial requirement for an engineer to succeed in a

dynamic environment.

One of the consequences of the manufacturing digitalization is the tighter integration between stan-

dardized processes and its respective people, responsible from them. Know-how that is restricted to a

limited number of individuals, a concept dubbed as islands of knowledge is to be demolished, as it is

expected for workers to have a wider understanding of all processes and the interconnections between

them, including potential issues and respective solutions. Social competencies are worthwhile in the

described scenario, as teams with dissimilar backgrounds join with the purpose of discussing measures

to improve production and to propose solutions for problems. Naturally, envisaged procedures are ex-

pected to be pragmatic and to be able to integrate into existing systems. A strong analysis capability is

needed, not only to devise a solution but also to envision how adequate it will integrate into the whole

ecosystem. Thus, adopting a viewpoint that grants a glimpse from the top levels until the shop floor

is advantageous. Having multiple systems connected, a number of times legacy equipment operating

in sync with new ones, can lead to a difficult whole problem visualization. Engineers are expected to

break down a complex problem into simpler ones to be assigned to teams later on. Teams composed of

diverse interdisciplinary subjects are more willing to adopt an out-of-the-box attitude, useful in problem-

solving as a part of the issues likely to happen in an I4.0 facility can’t be given a solution through the

exclusive use of computers; human creativity still is a clear advantage in problem solving.

To enable future workers to become familiarized with the environment and the challenges I4.0 will

bring, dedicated places are starting to make an appearance where their main objective is to consolidate

knowledge and serve as a learning platform, directed for industrial subjects. Learning factories, although

a relatively new concept, are becoming increasingly relevant for education and training in industry and

academic communities. One of its goals is competence development: by simulating a truthful factory

environment, future workers can be given specific know-how and are free to experiment by themselves

on situations where they’re not unacquainted with.

As the authors on [62] proclaim that modernising the learning process by providing a hands-on

experience helps to preserve knowledge and broaden the spectrum of found solutions and its respective

applications. As industry requirements become increasingly demanding, namely with knowledge related

with IT, trainees need to enhance its interdisciplinary capabilities and have a broader understanding

with digital technologies. Two main competencies fields are increasingly requested from industries to its

employees: professional competencies, related with the ability to handle complex situations and propose

feasible solutions, and supporting competencies, associated with social communication and personal

performance. Having a learning factory at their disposal grants an opportunity for their interpersonal

and technical skills to be developed, albeit with some limitations, such as the organizational culture and

dynamics existent in real workspaces that are absent in this setup.

Promoting engineers involvement on the company’s decisions and conceding freedom to pursue

34

projects of their own interest (which could have the potential for benefiting the company) has the ben-

efit of increasing motivation and consequently enhancing productivity. Several success products have

emerged from this 20% free time policy, such as 3M’s Post-It notes and Google’s Gmail [63].

2.7.3 Demonstrators as teaching aids

Current teaching methods have not accompanied the swift advances felt in industrial fields, such as pro-

duction technologies and techniques [64]. The classic classroom approach where students are exposed

to the subject through lectures seem to be less successful in fostering engineering competences and in-

stigating interest for students to have a critical mind and a rich multi-disciplinary background. By shifting

learning methods in order to bring it closer to the industrial practice students are given the knowledge

and tools required to thrive in a dynamic industry environment [65, 66].

The academic institutions referred on [9, 11] have shortened the education gap between academia

and industry through Learning Factories, allowing for trainees to better tackle real world situations in a

contained environment. Asides from modernising the teaching process, Learning Factories also have

the intent to open a path where research results can be more readily known by industry organizations,

spurring innovation. This integration effort between academia and industry is referred in [64] as the

knowledge triangle, a single initiative which combines education, research and innovation under a single

activity. The rapid pace of change on manufacturing technology should be taken into consideration.

Dedicated equipment may become obsolete after a few years, becoming limited in the opportunities

they can provide. Renovation is not only limited to physical assets but also needs to be extended to

the knowledge content and the methods of delivery [65, 67]. Hence it is essential to have a good

understanding of the subject’s technical contents and be aware of the potential for new ideas to be

commercialized so that the synergies between academia and industry continue to be up-to-date and

relevant. Maintaining a modern teaching method and relevant knowledge is crucial for improving trainees

performance, and consequently that of real manufacturing industries.

35

36

Chapter 3

A small scale electro-pneumatic

workbench for Industry 4.0 training

Following the literature review on Chapter 2, containing I4.0 aspects and their occurrences, this Chapter

has the role to describe one of the possible interpretations for the implementation of a small-scale

learning factory. Focusing on the product as an active element during the production process and

interoperability between equipments are the main points of this implementation.

3.1 Scope

The workbench detailed in this Chapter was developed with the goal of being used in a context of

active learning. The devised experimental set is based on a hands-on approach, giving students the

opportunity for them to try in first-hand a set of the most relevant I4.0 facets. It is pertinent to mention

that not all characteristics that were first mentioned in section 2.2 can be demonstrated.

The transformations that are occurring in manufacturing, at a much higher pace compared to the

first three industrial revolutions, call for major revisions on the roles played by PLC on a smart factory.

Nevertheless, their importance at field level means these control equipments will remain relevant, and

thus a PLC will also have a fundamental function on this set. By keeping the PLC in the miniature smart

factory, its users have a double possibility for learning. Not only can they improve on their integrated pro-

gramming languages but also on envisaging methods that introduce flexibility to the production system.

This aspect is particularly important in a panorama where several CPS are interconnected. Another skill

development opportunity arises, as an exchange of ideas and discussion in a group setting is beneficial

on selecting possible implementations for this subject or others.

This concept was implemented at the Industrial Automation laboratory of the Department of Mechan-

ical Engineering at IST, the location where the practical sessions of the course with the same name take

place. This curricular unit is lectured on the third year of the Mechanical Engineering degree, and is

aimed towards supplying the fundamentals of automation systems in use at industrial plants. The cur-

riculum has an heavy focus on the study of programmable automata and their programming language. If

37

adopted in the course curriculum, the result of this work could allow to bring the practical implementation

sessions in-line with some of the most relevant I4.0 concepts.

The implementation of this concept is naturally constrained by the available equipments in the labo-

ratory at the time the work was developed. The PLC’s proprietary architecture limits the possibility that

can be achieved with the equipment, as all interfaces need to be created through the manufacturer’s

software. Support for open-source solutions is lacking, which would allow for additional features. The

dynamic environment of a smart factory means that operations optimization is a challenging process

if done through the traditional means. A single unifying model is not adequate, as there is a need for

constant context updating. Therefore, it is necessary for the participating elements to possess adaptive

negotiation capabilities, which is an issue that PLCs are not prepared for. This scenario requires an

higher level of computational intelligence such as the one that exist in CPS, characterized by a tight

integration between the computational (software) and its physical (hardware) components. Thereby

although cooperation between PLCs or even other equipments in other workbenches is possible, the

heuristics that guide the decision-making process existent in a smart factory will usually be executed by

a third party equipment.

To materialize the digital backdrop that I4.0 disseminates, it is relevant to equip a workbench beyond

the standard PLC and actuators. Given the importance that I4.0 places on customized products and

traceability, it would be pertinent to install a reader (either based on RFID technology or the more stan-

dard barcodes) that would be tasked to examine incoming orders and subsequently transmit the relevant

instructions to the manufacturing service, which in this case is controlled by the PLC.

The interoperability of machines is a theme that can be approached by taking advantage of the net-

working capabilities of a PC and a PLC. The traditional operation mode of a PLC is to either operate

isolated from other control systems (used for localized operation) or inserted into a control loop via a

SCADA/CDS system (used in situations where a series of processes are underway, requiring more than

one PLC). By connecting a series of PLC into the same LAN, data can be sent between them using

specific communication protocols. Adopting the IoT philosophy, that besides improving interoperability

and increasing integration between dissimilar systems, one can take advantage of the open communi-

cation protocols and form a miniature production system comprising a PLC and a computer that runs

a program that is responsible to communicate with the industrial computer. To that effect, any program

that supports TCP or UDP can be used for exchanging messages with the PLC. MathWorks MATLAB is

a suited candidate to act as the link between the PC and the PLC, as its broad support for networking

toolboxes aid the task of sending and receiving message packets with the traditional TCP and UDP

protocols.

During the simulated manufacturing procedure there’s an opportunity to introduce an interpretation

on the real-time capabilities/information display aspect. Instead on relying solely on visual information

by manually inspecting the activity or having to actively check variables associated with the operation on

the PLC software, students could have access to an interface that gathers the most relevant information

regarding the process at hand. By taking into account the variable’s state, new clearer information can be

deducted from it and presented to the user. This aspect can be fused with the duality of the workbench’s

38

nature, having two versions of the same laboratory. A physical one, where students can have a hands-

on experience, and a virtual version of the physical laboratory, allowing for students to simulate the

operation of the physical laboratory without the need to interact with it directly. This version could be

classified as the digital twin of its physical counterpart, being available online through a webpage that is

hosted on the PLC.

The previously existent laboratory was comprised of:

• a four-level elevator;

• a conveyor belt;

• a panel with three pneumatic cylinders;

• a set of presence sensors;

• a set of binary light indicators;

• a webcam pointed towards the conveyor;

• a PLC;

• a desktop PC;

The first three components are wired to the PLC, which in turn is connected by a USB cable to the

desktop PC, the same applying to the webcam. Programs for controlling these first three elements are

written and downloaded through the PLC’s manufacturer editor software that runs on the desktop PC.

3.2 Proposed implementation

A didactic I4.0 workbench, based on the Industrial Automation laboratory, is proposed in this section.

Most equipments already present on the workstations are kept and are being integrated in an Industry

4.0 setting, where information is easily accessible. In order for students to be able to visualize informa-

tion regarding the processes in course and also the workstation’s equipment status, a web interface is

proposed. By uniquely identifying each component (with a QR code for example) students can access

the equipment data on their personal devices.

The value chain comprised of costumers and suppliers is also of special relevance, as are the chal-

lenges introduced on PLC programming by smart products. Due to the existence of customized prod-

ucts, the production sequence is not established by default. Instead, the product identifies itself upon

arrival and transmits the manufacturing steps to the PLC. Therefore, it is required for the GRAFTEC

program to handle a variety of possible input orders, without having to explicitly list every combination on

its program code. Costumers can place their orders through a MATLAB application that acts as interface

for communicating with the production system. A product can be identified by a bar code, setting the

manufacturing specifications, or selected from a list, where these are already known.

39

3.3 Workbench setup

As explained in section 3.2, most of the components from the current Industrial Automation Laboratory

are to be kept. The traffic lights and button panels are to be removed from the workbench physical

version and transferred to its virtual analogue. A barcode scanner is to be added above the conveyor

belt so that product codes can be read as they pass underneath it. In addition, a QR code is placed on

the workbench so that devices can retrieve information relating to the its operation, as will be detailed

later in section 3.6. An overview of the physical laboratory is shown in Figure 3.1.

Figure 3.1: Learning workbench for Industry 4.0

In respect to network connections, the PLC (a PCD3 M330 from Saia-Burgess) is now connected by

Ethernet cable to a wireless router, where the PC and other WiFi enabled devices are connected under

the same network. By supplying the PLC and other devices the same range of IP addresses devices

can exchange messages between them, which wasn’t possible with the previous setup. A scheme of

the elements setup can be seen in Figure 3.2.

The desktop computer represented in Figure 3.2 serves as the main interface between the user

and the production system. A Matlab application is responsible for exchanging messages with the PLC

and a database housed in the Sigma cluster at IST, containing information regarding the objects to be

manufactured.

The developed Matlab application is depicted in Figure 3.3. Items to be produced can be identified

40

Figure 3.2: Industry 4.0 Automation Laboratory diagram

either by their alphanumeric string (stored in its respective barcode) that is to be inserted on the Product

Code field or through a drop-down menu of pre-loaded items. A display area labeled Task List allows for

the user to check what is the item currently in production and the ones that are on queue. For the cases

where a quality control is needed after production, a view area with a frame captured from the webcam

and two text boxes are enabled. The camera display warns the user if the recently finished part is within

standards or is defective, a decision made with the measured top object’s area and its expected value.

Both values are displayed on the text boxes.

A diagram of the elements that comprise the developed laboratory is represented on Figure 3.4, both

physical and virtual ones.

41

Figure 3.3: Matlab application for interacting with the PLC.

Figure 3.4: List of workbench elements. Components that are present in the physical panel arecoloured in light blue, whereas those that exist digitally are coloured in dark blue.

3.4 Operation modes

The system can work in two modes, ”Continuous” and ”Task List” Modes. In ”Continuous” mode the

system processes products orders as they arrive, i.e, when the user places a part identifier on the

conveyor belt, activating the left sensor. By doing so the belt starts to move, driving it towards the

barcode scanner, where the product identifier is read. The alphanumeric string stored on the barcode

can correspond directly to the production sequence to be carried out by the pneumatic cylinders, named

A, B and C. In this situation the text string only contains the characters A, B or C, either in capital or

small letters. A capital letter corresponds to an advance movement, while a small letter is associated

42

with a retreat movement. For example, when Matlab sends ”ABab” to the PLC, cylinder A advances,

followed by cylinder B. When cylinder B has reached its end of course, cylinder A retreats, followed by

cylinder B.

If the previously described situation ensues, the text string is directly sent to the PLC, and no access

to the database is performed. If characters other than A, B or C take part in it, then the database needs

to be accessed. A SQL query is sent to the database, retrieving the production sequence and the top

object area for quality control purposes.

After cylinder movement is complete, the quality control step ensues. The webcam takes a picture

of the object, and after binarizing it a pixel count of the object’s top area is performed. If the area is

proximate to the expected area retrieved from the database, within a specified tolerance, then the object

is classified ”OK”. The belt movement is resumed, and after the object activates the belt’s right sensor

movement ceases, and the system goes into standby until the belt’s left sensor is active again. An UML

sequence diagram of this operation mode can be seen in Figure 3.5.

On ”Task List” mode the user has to firstly scan all the product identifiers. When a barcode is scanned

Matlab verifies if the code is ready to be sent to the PLC, otherwise the database is accessed. While the

codes are being scanned, respective product information is shown in the Task List display on the Matlab

interface, being the earliest scanned codes located on top. After the ”Start Task List production” button

is pressed the system awaits for the user to activate the belt’s left sensor. Upon activating it productions

develops in a similar scheme to that of Continuous mode. When an item production finishes, the top row

from the Task List is removed, an the system awaits for the next item production to start. A sequence

diagram for Task List mode is depicted in Figure 3.6.

A GRAFCET diagram that describes the behaviour of the PLC program is shown on Figure 3.7.

43

Figure 3.5: Continuous mode.

3.5 Communications between desktop PC and PLC

Saia Burgess PLC’s support two modes of data exchange through a network. S-Bus is a proprietary

protocol that allows for data exchange between Saia PLC’s in the same network. Multiple master devices

can coexist in it, being these the entities that have the capability to copy and receive data from a slave

station upon request. For the situations where communication is to be established with a device which

doesn’t support S-Bus, the Open Data Mode can be used instead. Messages are sent via UDP, whereby

a confirmation that messages were delivered to the recipient needs to be implemented. Also, in contrast

44

Figure 3.6: Task List mode.

to S-Bus, data cannot be directly requested from the remote station; stations that request data must

wait until a message is delivered to them. Devices that don’t support the S-Bus protocol aren’t able to

transmit the data types commonly used in Instruction List programs such as flags or registers. Therefore

only unformatted data can be sent between devices, such as a vector of numbers or characters.

Saia PLC’s registers use 32 bits (8 bytes) to store integers, in the range of [-2,147,483,648:2,147,483,647].

Texts (an array of characters) are represented as vectors of numbers in the unsigned integer 8 bit for-

mat, where each vector entry (a character) corresponds to a number in the ASCII table. As an example,

”Micro SI prefix: µ” is coded as seen in Figure 3.8.

Matlab supports sending/receiving UDP messages consisting on 32 bits through the DSP System

Toolbox, but since this toolbox isn’t available on the Industrial Automation laboratory PCs this possibility

was discarded. Instead, a function from the Mathworks website was used [68], which uses Matlab’s Java

interface to send and receive messages (by UDP packets) in the 8 bit signed integer format. The devel-

45

Figure 3.7: GRAFCET diagram for the production system (part 1)

46

Figure 3.7: GRAFCET diagram for the production system (part 2)

47

Figure 3.7: GRAFCET diagram for the production system (partial GRAFCET)

48

Figure 3.8: Text representation convention in Saia code editor.

oped algorithms for sending/receiving data make use of this function, and are detailed in the following

subsections.

Sending numbers

Using the UDP communication function to send a number directly to a register in the PLC is disadvan-

tageous since only numbers belonging to the int8 format range of [-128;127] can be sent. Instead, a

string of 4 characters is sent in the form of an int8 vector, received by the PLC in a text variable, and its

decimal value is consequently copied to a register. By using this method any number belonging to the

int32 format range can be sent using the previously mentioned function. Table 3.1 gives an example of

the correspondence between the int8 vector sent by Matlab, the 4 character text received by the PLC

and its corresponding numeric value.

int8 vector Binary (Least significant register byte) Decimal value ASCII character (Least significant byte)

[0 0 0 1] 0000 0001 1 SOH (start of header)[0 0 0 2] 0000 0010 2 STX (start of text)

[0 0 0 127] 0111 1111 127 DEL (delete)[0 0 0 -128] 1000 0000 128 e

[0 0 0 -127] 1000 0001 129 -[0 0 0 -1] 1111 1111 255 y

Table 3.1: 8 bit integer representations

It should be noted that Saia’s PLCs use two’s complement for representing signed numbers. By

doing so the value zero is uniquely represented in this notation, whereby in one’s complement zero can

be positive and negative. By eliminating this redudancy the value zero has a unique representation, and

numbers can be represented in the int8 range. In comparison, a signed 8 bit integer which uses one’s

complement has to be within the range of [-127;127].

In order to determine the 4 character sequence which represents a given number first it’s 32 bits

binary representation has to be computed. Due to two’s complement, an important detail should be

noted: contrarily to positive numbers where its representation in binary form is direct, negative numbers

should be subtracted one unit from the input absolute value, and its binary representation be inverted.

This is due to the conversion algorithm to this notation:

1. Perform one’s complement (invert all bits)

2. Add one unit

49

The 32 bit binary vector resultant from this operation needs to be split into 4 blocks of 8 bits each, so

that each byte represents a character in the ASCII table. An intermediary conversion to uint8 is made as

Matlab’s typecast.m returns the same numbers of bytes as were in the input, and no direct conversion

from binary to int8 is available. As both int8 and uint8 use eight bits, there’s no overflow risk. With each

character now established, its decimal value is stored in the correspondent entry in the int8 vector, that

is now ready to be sent.

An important remark is to be done regarding the order in which bits are stored into memory, named

endianness. In big-endian format, the most significant byte is stored first on the lowest memory address

and the following bytes are subsequently stored by decreased significance order, being the least sig-

nificant byte stored on the highest memory address. On contrary, on the little-endian format the most

significant byte is stored last on the highest address and the least significant byte is stored firstly on the

lowest address. It’s essential that two devices communicating over a network have the same byte order

storage, otherwise the message becomes unintelligible upon arriving on destination. While Intel/AMD

CPUs use little-endian, networks and Motorola processors (such as the one used in Saia’s PCDs) use

big-endian. Therefore, an operation similar to the one depicted in Figure 3.9 has to be performed, either

before sending the message through Matlab or swapping the byte order on the PLC upon receiving the

message.

Figure 3.9: Endianness, from Saia’s PG5 Help

Receiving numbers

Using the function in [68] configured to receive a packet results in an 4 entry int8 vector as an output.

Each one of the entries is converted to uint8, since this format has the same range as an unsigned 8 bit

binary number. Each entry is consequently converted to binary format and are concatenated so that the

result is a 32 bit binary vector, which is then converted to decimal.

Sending text

Text can be seamlessly sent, it is only required to convert a string or single character to an int8 vector.

50

Receiving text

Similarly to the case of receiving numbers, a text received by Matlab arrives in the form of an int8 vector.

After receiving the variable, excess spaces (identifiable as the number 32 in the ASCII table) at the end

of the message are deleted, and running the char.m function returns the corresponding characters.

3.6 Webpages as information carriers

The factory of the future is expected to generate and exchange a considerable amount of data. Having

the relevant information readily available is crucial for a satisfactory management of equipments and

orders. Webpages that serve as human-machine interfaces were developed using PG5’s Web Editor,

allowing an holistic view of the workbench equipments and its respective data.

3.6.1 Virtual Laboratory

In addition to the physical laboratory, a virtual version was also developed. The virtual laboratory can

work either as a mirror of its real counterpart or as a standalone version, capable of running programs

without the need for the user to be present at the physical lab, since the user’s interactions are done via

a webpage hosted in the PLC, which is connected to a network. The developed interface is depicted in

Figure 3.10.

Figure 3.10: Virtual Laboratory webpage

In an effort to improve the workbench organization some elements were removed from the lab phys-

ical version, such as the button panel which now only exists in the virtual one. A benefit of having a fully

digital laboratory is that common components existent in production lines can be added to the Automa-

51

tion Laboratory with zero expense. A three color stack light and an emergency button is now part of all

workbenches without the need for additional hardware configuration.

3.6.2 Equipment webpages

Each equipment on the physical workbench has a dedicated QR code. When read, it redirects the user

to a webpage where information regarding that component can be conveniently accessed, such as its

status and useful statistics. The elevator’s QR code, Figure 3.11 b), redirects the user for the elevator’s

webpage, Figure 3.12. On this webpage the user can verify the amount of time the elevator has been

stationed at a certain floor and the percentage of use of said floor.

Similar to the equipment frames, the virtual laboratory also has a QR code on a frame placed at the

workbench, Figure 3.11 a). This allows for the physical workbench user to have a condensed view of

its status, as well as giving access to additional buttons that aren’t accessible through the laboratory

physical version. In addition to the button that are available in the Virtual Laboratory webpage, an

additional button panel is made available through another webpage hosted on the PLC. This panel

(Figure 3.13) hosts six switches, an editable text box for the user to input its name and input fields where

the user can read/write from/to registers.

(a) (b)

Figure 3.11: a) Virtual Laboratory QR code b) Elevator QR code

3.7 Simulation of CPS

In an ideal set, a plant that is intended to operate according to I4.0 principles should have its constituting

elements acting as modules. In such arrangement, the communication between participating elements

occurs by sending messages directly to the intended recipient, unlike networks which are reliant on

52

Figure 3.12: Elevator webpage

Figure 3.13: Virtual button panel

having a hub responsible for intermediary information exchange. This communication scheme that is

characterized by not relying on additional hardware to bridge devices is labelled as decentralized, where

each element has the ability to work independently. Abstracting from the interior details of the whole,

the combination of the individual CPS can be interpreted as a single mechanism, which in the context

of I4.0 is normally associated with a production system.

53

The learning workbench that was depicted earlier in this Chapter uses a program written in GRAFTEC

(Saia’s graphical programming method as an implementation IEC Sequential Function Chart standard)

to establish communication between the PLC and MATLAB running on the PC. IEC’s SFC (defined by

the 61131 standard from the same organization) itself is based on the GRAFCET norm, a graphical

representation methodology oriented towards controlling a sequential process. The GRAFTEC program

contains all the necessary communication functions, activities and variables of the production system.

This way, the activities of all the elements that comprise the system are all controlled in the same block of

code, as opposed to being distributed through each element’s block. This first approach is not in line with

the characteristic autonomy of CPS. Thus, in order to simulate the concurrent operation of several CPS

(each workbench element amounts to one cyber-physical system) a distinct approach was undertaken,

being demonstrated in this section.

The test platform used for testing the modularity and interoperability of a smart factory’s elements

was a simplified version of the previously shown virtual laboratory in Figure 3.10. Some changes were

made; the removal of the stack light and the virtual button panel as they are deemed unnecessary

for simulation purposes. Furthermore, the MATLAB application (showcased in Section 3.3) and the

Quality Control procedure were equally dismissed as it is only of interest to simulate the operation of

the production elements that can be contained only within the PLC. To replace the MATLAB application

a text box was added to the virtual laboratory webpage. Through this graphical element the intended

production sequences can be directly inserted into the PLC.

In order to embrace the modularity aspect of a CPS, the code written for each one of the workbench

elements was isolated from the previously referred GRAFTEC program and inserted into a block in an

Instruction List file. The Instruction List language is a low level (non-graphical) programming language

developed by Saia for programming its PLC, with programs being formed by a series of instructions

arranged in successive rows. This language is structured such that each simultaneous running process

is defined by a Cyclic Organization Block (COB), a block of instructions that runs cyclically from its first

to the last instruction. Saia’s PLC models can run up to 16 cyclical blocks, being successively executed.

Two main reasons exist for choosing to implement this CPS simulation in Instruction List language:

• Each CPS activities can be clearly defined inside its respective COB;

• Some of workbench elements have its behaviour described in GRAFTEC programs, which are not

possible to call from another Sequential Block.

The developed workbench presented on this Chapter uses a function in MATLAB to send and re-

ceive messages in UDP to the PLC. This messages controls the GRAFTEC program flow, causing it to

go through stages that are associated with production elements activities. On this simulation environ-

ment, this communication aspect between elements is mimicked through the use of flags. These binary

variables were used to condition the execution of a set of instructions contained in each COB, i.e., the

program in each of the cyclic blocks only runs in the situations where one of these variables has its

value set as true. Likewise, if the message-replacing flag is set as false then the program to which it is

associated does not execute.

54

The communication aspect of this setup can be represented by a finite state machine diagram, shown

in Figure 3.14. The overall system (comprised of four entities) undergoes through several states during

operation, being each one of them associated with a program. The correspondence between a given

state an its respective program is displayed on Table 3.2. In order for the system to change between

states certain inputs must be verified; these correspond to the aforementioned flags that appear on the

diagram’s transition arcs.

S0

S2

S3

S4

S1

Call Conveyor Belt

Call Process Order Text

Call Elevator

Call Cylinders

Call Process Order Text

End of production

Figure 3.14: Finite State Machine diagram for simulated CPS communication

State Active program

S0 InitializationS1 Conveyor Belt COBS2 Process Order Text COBS3 Elevator COBS4 Cylinders COB

Table 3.2: Correspondence between states and active programs

As seen on Figure 3.14, four elements were considered to be depicted as a CPS. The conveyor belt

and the pneumatic cylinders were modelled in Instruction List, while the elevator and the Process Order

Text module where programmed in GRAFTEC. The first three act as a virtual representation of their real

counterparts, being responsible for the animations in the simulation webpage. The Process Order Text

module has the function of receiving the production sequence and dividing it into characters used to

command the elevator and cylinders, as explained in Section 3.4.

The complete system’s behaviour, modelled with an UML state diagram, can be seen in Figure 3.15.

The underlying operation mode of this setup is similar to the one implemented in GRAFTEC format.

Upon start-up, the group of CPS initialize their own variables. This is done by using an Exception

Organization Block, a block of code in Instruction List which is only executed once, and corresponds to

the Restart Parameters state. Following the initialization, the system’s progresses to the Conveyor Belt

COB state. As the transition between these two states has no associated event, it will auto-trigger when

the Restart Parameters state has all its internal actions executed. As the system reaches the second

state, the first CPS becomes active, corresponding to the conveyor belt. The COB associated with this

CPS, COB 0, runs continuously while the simulated production system stays in this state. The activation

55

Figure 3.15: State Diagram for the Cyber-Physical Production System in study.

of the left side conveyor belt sensor puts the later on motion, as represented on its UML activity diagram

on Figure 3.16.

Figure 3.16: Activity Diagram for Conveyor Belt module.

The occurrence of the transition’s event (activating the right side conveyor belt sensor) makes the

former to trigger and to advance the system to the next state. The transition’s triggering by validating the

guard condition is analogous to the Conveyor Belt CPS sending a message to the Process Order Text

CPS informing the later to start its activities. After the message is sent the Conveyor Belt goes into a

waiting condition until its actions are again required. As this is done in a simulated environment, a flag

variable named CallProcessOrderText is used to request the Process Order Text COB to become active.

When the Process Order Text CPS becomes active, its first action is to read the production sequence

previously input by the user in the text box, as it can be seen on its activity diagram, represented on

Figure 3.17.

This CPS has the task of extracting sequential information from the production sequence, namely the

intended elevator floor and an order for the pneumatic cylinder located at that floor (either an advance

or retreat movement). Following this identification, two messages are sent to the Elevator and Cylinders

CPS, indicating the elevator floor and the cylinder name/movement pair. Given the simulation situation

56

Figure 3.17: Activity Diagram for Process Order Text module.

where this occurs, the message exchange is reproduced by copying the variables of interest from COB

1 to COB 2 and COB 3. It should be noted that unlike the message sent by the Conveyor Belt these

messages aren’t intended to change the production system state by calling another CPS. It is only after

these messages are received that the Process Order Text CPS sends the message for the Elevator to

become active.

When the Elevator becomes active, the cabin initiates motion from the initial floor to one of the floors

previously set by the Process Order Text CPS. When it immobilizes there, a message is sent to the

Cylinders, requesting for them to execute the order that was previously received. The behaviour of this

CPS is presented on Figure 3.18.

Figure 3.18: Activity Diagram for Elevator module.

When said motion is complete, Process Order Text is again called, as show in Figure 3.19.

In the situation where the text string has finished processed, i.e, all its characters have been read, a

57

Figure 3.19: Activity Diagram for Cylinders module.

message is sent to the Elevator for it to return to floor 0 and the system returns to its original state.

58

Chapter 4

Laboratory use cases

In this section a demonstration on the usage of the workbench setup detailed on Chapter 3 is un-

dertaken. The covered situations include scenarios in which the manufacturing process originates an

admissible and a defective part. The features of the developed MATLAB application are exposed in

here, along with its interactions with the production system. Two methods for an emergency stop pro-

cedure are also detailed. To conclude, a scenario for collaboration between cyber-physical systems is

showcased.

4.1 Test specimens

To test the production system, different versions of the same workpiece, a miniature turbine, were pre-

pared. One has the expected characteristics that the piece acquires after it clears the production pro-

cess, while the other possesses some property that makes it improper for acceptance, Figure 4.1 a) and

b) respectively. In this case, the defective part has a larger shaft diameter when compared to the regular

part.

(a) Regular part (b) Faulty part

Figure 4.1: Workpieces used for the laboratory demonstration.

59

4.2 Workpiece identifiers

In order for the parts to enter the production system each one of them need to be attributed a bar code

that allows for it to be identified by the production system. Each bar code, consisting in a sequence

of alphanumeric characters, uniquely identifies a particular order. Through the aforementioned charac-

ter sequence, titled the ID, an additional set of information can be retrieved from the online database

mentioned in Section 3.3:

• Name: A label for the object;

• Production sequence: Sequence of movement to be executed by the pneumatic cylinders;

• Area: A pixel count of the object’s top area used for quality control.

As explained in Section 3.4, if the ID contains only the characters A, B or C on small or capital letters

then the ID corresponds to the production sequence. In this case no query is submitted to the database,

and therefore there is no information available regarding the fields Name and Area. Antagonistically,

when it is required to query the database since the ID can’t be sent directly to the PLC, the production

sequence is retrieved, along with the object’s label and the Area value to be used in the quality control

step after production has finished. On this situation the ID acts as a primary key to uniquely identify the

tuple on the database. Figures 4.2 and 4.3 show respectively a bar code with an ID corresponding to

the production sequence, and another which requires a query to the database, along with the retrieved

tuple.

Figure 4.2: A bar code with a printed production sequence.

4.3 Manufacturing simulation

Following the production sequence retrieval by passing the bar code under the scanner, after pressing

the button akin to one of the two production methods explained in Section 3.4 and trigger the left belt

sensor, the conveyor belt immobilizes. With the worpiece stationed under the webcam, production

ensues. The series of cylinder movements, determined by the production sequence, are accompanied

by the elevator according to Table 4.1. This is done to simulate the workpiece transportation between

cylinders, as if they were arranged like shown in Figure 3.2.

60

(a) Bar code used to acess the database (b) Attributes retrieved from the query

Figure 4.3: Using a bar code to access the database

Cylinders movement Elevator movement

Advance/retreat A Move to floor 1Advance/retreat B Move to floor 2Advance/retreat C Move to floor 3

Table 4.1: Joint movement between cylinders and elevator

During this step the MATLAB application updates the state of the process, displaying ”Producao em

curso” as the cylinder and elevator movement progresses, Figure 4.4. Regardless of the product bar

code requiring to query the SQL database, details of the order currently being attended are presented

in the Lista de tarefas panel. If a query is necessary, the available items indicated in Section 4.2 are

displayed in this board. Additionally, the Area parameter is displayed in the applicable window, Area do

objecto (BD) Otherwise, only the production sequence field is filled.

When all characters in the production sequence are read this phase terminates, and the elevator

returns to the zero floor.

4.4 Quality control

With production complete and the workpiece stationed under the webcam, a snapshot of the first is

taken. Then the quality control algorithm processes the previously captured still and classifies the re-

cently finished part. As shown in Figure 4.5, a rectangle is placed on the bottom left corner image

delimiting the total picture area to the object of interest. Also, a message indicating the output of the

quality control program is presented over the object image.

If the part is considered satisfactory, an OK is printed. Otherwise, the string Not OK is shown. The

green or red light of the stack light turn on during 3 seconds, respectively. Figures 4.6 and 4.5 exemplify

the MATLAB application and the workbench behaviour during this stage.

After the 3 seconds have passed, the conveyor belt resumes it motion. If the workpiece was con-

sidered to be within the defined parameters, the movement is from the left to the right. This way, when

61

Figure 4.4: MATLAB application during the production phase

Figure 4.5: MATLAB application during the quality control phase

the workpiece triggers the sensor located on the conveyor’s right side, the current order is considered

finished. Depending if the system is running in Continuous or Task List mode, the MATLAB application

62

(a) Adequate part (b) Green indicator onstack light

(c) Defective part (d) Red indicator onstack light

Figure 4.6: Testing the workpieces after production (left column) and a visual flag of the result (rightcolumn)

either waits for the next product bar code to be scanned or proceeds to the next item present on the

Task List panel. If the part was classified as defective by not falling in between the required parameters

then the conveyor belt moves from right to left. When the part triggers the sensor located to the left of

the conveyor it is discarded, and the conveyor immobilizes.

4.5 Database Management

Besides being used as the communication link between the PC and the PLC, the MATLAB application

serves as an interface between the user and the products database. On the window displayed on Figure

4.7, the user is able to insert or remove items into the database. For the insertion of tuples, it is required

to insert the product ID, its production sequence and the area (the name is optional and used mainly to

better clarify to what object each line corresponds to). After the required fields are filled, the new entry

is sent to the remote database and the later is updated on the window’s top table. Simultaneously, a bar

code is generated that encodes the previously inserted ID. It can be printed and later used to identify a

63

product to be manufactured on the production system.

For the case where an object is to be removed from the database, the user has to solely supply the

ID. After requesting the deletion of the object, the table on top of the window refreshes and the product

is removed from its listing.

Figure 4.7: Database management GUI

4.6 Emergency stop procedure

While designing the workbench, it was considered pertinent to include safety measures with the task of

paralysing equipment in an hazardous situation. Both proactive and reactive solutions were taken into

account, with the inclusion of a presence sensor and an emergency stop switch.

4.6.1 Presence detector

The presence sensor is a safeguarding system, meaning it automatically acts in order to prevent an

individual to become involved with an hazard before it occurs. Their operating principle is based on

the detection of electromagnetic waves, either in the form of infrared or microwaves, or even through

sound waves with an higher frequency than the common human hearing (ultrasounds). One of the more

commonly installed presence sensing devices on industrial premises are photoelectric sensors, which

are based on the emission and reflection of a light beam from a target, thus detecting a change in light

intensity. If the light emitted by the transmitter hits the target, it reflects some of the light back to the

receiver (indicating a change on the amount of light that arrives to it).

Compared with another type of proximity sensors, such as the ones based on the inductance phe-

nomenon, photoelectric sensors have a greater sensing range, and notably can detect non-metals. This

makes it particularly interesting for detecting if an individual approaches/invades the working envelope

64

of a machine, enabling for preventive measures to be taken. In this implementation the presence sensor

is used to detect if a person’s hand traverses the moving path of the pneumatic cylinders. If such a

situation occurs, a warning is issued on the laboratory webpage, and the ongoing process is stopped.

Figure 4.8 showcases this proactive security measure.

(a) Hand triggering the presence sensor, notice the yellow lightindicating the presence of an obstacle

(b) Warning displayed on the laboratory webpage

Figure 4.8: E-stop demonstration, triggered by workspace breaching

4.6.2 Emergency stop switch

Contrary to the presence detector, the emergency stop button is a reactive stop function, which by

definition it is to be initiated by a single human action, thus is not automatic. Although it does not qualify

as a safeguarding device, it is still an important complementary protective mechanism, providing the

user with a backup to the primary safeguards. If a user detects that a potentially dangerous situation is

about to unfold the emergency stop button can be used to stop the system. Figure 4.9 exemplifies the

use of the e-stop switch, and its effect on the MATLAB application.

4.7 A simulated collaborative scenario for Cyber-Physical Systems

A demonstration of the scenario detailed in Section 3.7 is illustrated in this subsection. Following the

system initialization, a request is made for the Conveyor Belt to become active. It should be noted that

the term active used in this context means to become aware of the other elements belonging to the

production system (such as preparing for receiving messages) or to actions done by the user that act as

a stimulus for the CPS to perform an action. It does not necessarily mean that when it is active it needs

to be carrying out an action such as moving the conveyor. As can be seen on Figure 4.10 c), the flag

65

(a) Default position (b) Depressed button, which paralyses the system

(c) MATLAB application warning

Figure 4.9: E-stop button usage and warnings on MATLAB application

CallConveyorBelt is set as true, which emulates the request that was made for the transporter CPS to

activate following the system initialization.

With the Conveyor Belt now awaiting for an input, the user can press the presence sensor image

located on the conveyor’s left, simulating the placement of an object on the conveyor. This causes the

conveyor to initiate its motion, transporting the object from its left extremity to the right one as can be

seen by the green direction indicator arrow on Figure 4.10 a). Since this simulation runs entirely on the

PLC workspace there is no barcode reader installed over the conveyor belt to read the order’s production

sequence. Instead, there is a text box where the user inputs the intended production sequence. The

production sequence should be written there before the user presses the presence sensor image on the

conveyor’s right side. Failing to do so won’t allow the production system to evolve according to the state

diagram on Figure 3.15 as no character can be read. When the user triggers the right side presence

sensor the Conveyor Belt sends a message to the Process Order Text CPS in order for the latter to

perform the sequence of actions according to its activity diagram, Figure 3.17.

As depicted on Figure 4.10 d), the flag CallProcessOrderText becomes true meaning the correspond-

ing CPS is active. The Process Order Text CPS receives the production sequence previously input and

reads its first character, printing it into a text box (Figure 4.10 b)). Depending on the extracted character,

66

(a) Active Conveyor Belt CPS (b) Active Process Order Text CPS

(c) Flag for Conveyor Belt CPS activation (d) Flag for Process Order Text CPS activation

(e) Active Elevator CPS (f) Active Cylinders CPS

(g) Flag for Elevator CPS activation (h) Flag for Cylinders CPS activation

Figure 4.10: A 4 CPS elements collaborative scenario

the elevator floor and the cylinder movement are sent to their respective CPS. Then, the Elevator CPS

becomes active by setting the flag CallElevator as true, Figure 4.10 g). In this example the first produc-

tion stage corresponds to an operation to be performed by cylinder B. Since this cylinder is located at

the second floor and at startup the elevator is stationed at ground floor the latter will initiate an upward

67

movement towards that floor. Figure 4.10 e) represents this stage, with the elevator travelling between

floors. When the Elevator is active and moving its status changes from ”Em pausa” to ”Em movimento”

and no floor light is switched on.

When the elevator arrives at the intended floor it goes into a waiting state and the flag CallCilindros

is set true (Figure 4.10 h)). This has the effect of activating the Cylinders CPS, which will execute the

order that was previously sent by the Process Order Text module. Since the extracted character from the

production sequence is in upper case the B cylinder will execute a forward movement, shown in Figure

4.10 f). Since the production sequence is formed by more characters, the flag CallProcessOrderText is

again set as true so that Process Order Text extracts the next character, sends messages to the Elevator

and Cylinders CPS, et cetera, until all the production sequence as been read. When there are no more

characters to read, the elevator moves to the ground floor and the flag CallConveyorBelt is set as true,

permiting the whole process to restart.

68

Chapter 5

Conclusions

5.1 Achievements and final remarks

The work developed on this dissertation is directed towards the development of an educational experi-

mental kit, with the goal of enlightening its users regarding a set of features of the I4.0 programme. In

order for this to materialize a survey of the main aspects/underlying technologies of the fourth industrial

revolution was carried out, along with a study of the already existing training systems/demonstrators,

in the form of a comparison between these and the methods used by them in order to epitomize the

characteristics of I4.0 found earlier.

Having gathered a series of guidelines for the construction of a I4.0 didactic learning scenario, a

small scale production system was devised. Encompassing important I4.0 aspects, the designed sys-

tem takes as a starting point the current Industrial Automation Laboratory at IST to develop a more

flexible version of it, allowing for its users to introduce new interactive elements into it such as MATLAB

code and Wi-Fi enabled devices. This would not have been possible without the integration of its ele-

ments in a network and the development of a communication algorithm between the PLC and external

devices by taking advantage of the open communications protocol of the first. The developed workbench

allows for the placement of orders of customized products (identifiable by a bar code), whose production

sequence are retrieved by a MATLAB application and sent to the PLC, thus illustrating the interoperabil-

ity aspect between equipments. A series of webpages (accessible through QR codes) allow for direct

information access regarding its equipments, including a complete virtual version of the workbench. Al-

though a single workbench is both limited to the number of comprising modules and functionalities, a

more complex system comprised of several workstation could expand its capabilities, namely the joint

operation between modules.

A study of a collaborative production scenario based on CPS was also conducted. The previously

devised system (running on a GRAFTEC program) was modelled using a finite state machine, where

each state corresponds to an active workbench component. The transitions between states represents

the communications between CPS, therefore enabling the individual elements to work collectively to

become a single entity. This alternative implementation proves to be advantageous compared to the

69

previous in terms of flexibility, as the modular nature of this program allows for easier adjustments, such

as the number of participating elements or their behaviour.

Being a fairly recent concept, the implementation of Industry 4.0 is seen by some as big-budget

technology project, showing diffidence owing to the uncertainty of opportunities and threats. Instead

of considering it as the implementation of an IT project, I4.0 is to be regarded as an overall strategic

objective to materialize the smart factory of the future. An additional research effort is still needed to

better clarify the effects of its implementation, specially to minimize the negative ones. The focus of

considerable investments to be done on the design phase should be well thought over as poor planning

could lead to a waste of resources, with little results during the functioning phase.

5.2 Future Work

Due to its components limitations, it is not possible for the implemented workbench to cover exhaustively

all the studied I4.0 facets. With that in mind, in this section some suggestions are presented to continue

this work or inspire further ones. There is potential to personalize the experience when one uses the

laboratory interface, such as giving the possibility for a user to have access to webpages configured

according to its preferences, such as icons/buttons layout and display language. This could be done

through wearable pieces of identifications, like a wristband with an NFC tag that would be read upon

approaching the workbench.

Following on the interoperability aspect of I4.0, one suggestion is to devise a facility with multiple

stations. By taking advantage of the communication algorithm developed on this work or the inbuilt PLC

functions, a network could be established between stations in order to set up a multi-node miniature

factory.

Another proposal is the implementation of an augmented reality system, based on digital projection

technology that superimposes computer generated graphics into the work surface. This tool would

provide visual and audio instructions, acting as an interactive guide for a manual task.

70

Bibliography

[1] Emerging Tech: Industry 4.0. URL https://techwinx.com/industry-4-0/. [Online; accessed

26-April-2019].

[2] C. Freeman and F. Louca. As time goes by: from the industrial revolutions to the information

revolution. OUP Oxford, 2001.

[3] A. Daneels and W. Salter. What is SCADA? 1999.

[4] What is the automation pyramid? URL https://realpars.com/automation-pyramid/. [Online;

accessed 28-April-2019].

[5] L. Monostori, B. Kadar, T. Bauernhansl, S. Kondoh, S. Kumara, G. Reinhart, O. Sauer, G. Schuh,

W. Sihn, and K. Ueda. Cyber-physical systems in manufacturing. CIRP Annals, 65(2):621–641,

2016.

[6] Y. Lu. Industry 4.0: A survey on technologies, applications and open research issues. Journal of

Industrial Information Integration, 6:1 – 10, 2017. ISSN 2452-414X. doi: https://doi.org/10.1016/j.jii.

2017.04.005. URL http://www.sciencedirect.com/science/article/pii/S2452414X17300043.

[7] S. Erol, A. Jager, P. Hold, K. Ott, and W. Sihn. Tangible Industry 4.0: A Scenario-Based Approach

to Learning for the Future of Production. Procedia CIRP, 54:13–18, 2016. ISSN 22128271. doi:

10.1016/j.procir.2016.03.162.

[8] M. Weine. Overcoming communication barriers. Faszination Forschung, (20):83–91, 2017.

[9] E. Abele, J. Metternich, M. Tisch, G. Chryssolouris, W. Sihn, H. ElMaraghy, V. Hummel, and F. Ranz.

Learning factories for research, education, and training. Procedia CIRP, 32(Clf):1–6, 2015. ISSN

22128271. doi: 10.1016/j.procir.2015.02.187.

[10] PTW TU Darmstadt - 10 years Center for Industrial Productivity (CiP). URL https://youtu.be/

H9Od8phTUdM?t=142s. [Online; accessed 25-March-2019].

[11] Z. Kemeny, R. J. Beregi, G. Erdos, and J. Nacsa. The MTA SZTAKI Smart Factory: Platform

for Research and Project-oriented Skill Development in Higher Education. Procedia CIRP, 54:

53 – 58, 2016. ISSN 2212-8271. doi: https://doi.org/10.1016/j.procir.2016.05.060. URL http:

//www.sciencedirect.com/science/article/pii/S2212827116305169. 6th CIRP Conference on

Learning Factories.

71

[12] Network of Innovative Learning Factories (NIL). The Learning Factory, an annual edition from the

network of innovative leaning factories. 2016. ISBN 9783981750805.

[13] R. Langmann and L. Rojas-Pena. PLCs as Industry 4.0 components in laboratory applications.

International Journal of Online Engineering, 12(7):37–44, 2016. ISSN 18612121. doi: 10.3991/

ijoe.v12i07.5828.

[14] H. ElMoaqet, I. Ismael, F. Patzolt, and M. Ryalat. Design and Integration of an IoT Device for

Training Purposes of Industry 4.0. In Proceedings of the 2nd International Symposium on Computer

Science and Intelligent Control, page 25. ACM, 2018.

[15] SmartFactory-KL. URL https://smartfactory.de/en/. [Online; accessed 3-July-2019].

[16] G. Candido and J. Barata. A multiagent control system for shop floor assembly. pages 293–302,

01 2007. doi: 10.1007/978-3-540-74481-8 28.

[17] P. Leitao and F. J. Restivo. Implementation of a holonic control system in a flexible manufacturing

system. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews),

38(5):699–709, Sep. 2008. ISSN 1094-6977. doi: 10.1109/TSMCC.2008.923881.

[18] INESC TEC inaugura laboratorio para a industria e inovacao. URL https://noticias.up.pt/

inesc-tec-inaugura-laboratorio-para-a-industria-e-inovacao. [Online; accessed 3-July-

2019].

[19] S. Heimbach. Maserati and Siemens drive digitalization in the automotive industry. pages 2–5,

2015.

[20] Maserati and the Future of Manufacturing. URL https://youtu.be/Hk__pjSe3a8. [Online; ac-

cessed 27-March-2019].

[21] W. Henning, Kagermann Wolfgang and H. Johannes. Recommendations for implementing the

strategic initiative INDUSTRIE 4.0. Final report of the Industrie 4.0 Working Group, (April):82,

2013. ISSN 2405-8963. doi: 10.13140/RG.2.1.1205.8966.

[22] J. Smit, S. Kreutzer, C. Moeller, and M. Carlberg. Industry 4.0. Study for the ITRE Committee,

Policy Department A: Economic and Scientific Policy, European Parliament, Brussels, 2016.

[23] BCG Analysis. Embracing Industry 4.0 and Rediscovering Growth. URL https://www.bcg.

com/capabilities/operations/embracing-industry-4.0-rediscovering-growth.aspx. [On-

line; accessed 22-February-2019].

[24] I. Graßler, A. Pohler, and J. Pottebaum. Creation of a Learning Factory for Cyber Physical Produc-

tion Systems. Procedia CIRP, 54:107–112, 2016. ISSN 22128271. doi: 10.1016/j.procir.2016.05.

063.

[25] G. Reinhart, P. Engelhardt, T. Philipp, W. Wahlster, J. Zuhlke, J. Schlick, T. Becker, M. Lockelt,

B. Pirvu, P. Stephan, S. Hodek, B. Scholz-Reiter, K.-D. Thoben, C. Gorldt, K. Hribernik, D. Lappe,

72

and M. Veigt. Cyber-physische produktionssysteme: Produktivitats- und flexibilitatssteigerung

durch die vernetzung intelligenter systeme in der fabrik. wt Werkstattstechnik (ISSN 1436-4980),

103:84–89, 03 2016.

[26] Pilot Cyber-physical Production Systems. URL https://www.youtube.com/watch?v=

wro3uoHR-ZY. [Online; accessed 22-February-2019].

[27] L. Monostori. Cyber-physical production systems: Roots, expectations and r&d challenges. Proce-

dia Cirp, 17:9–13, 2014.

[28] H. Sundmaeker, P. Guillemin, P. Friess, and S. Woelffle. Vision and challenges for realising the

internet of things. Cluster of European Research Projects on the Internet of Things, European

Commision, 3(3):34–36, 2010.

[29] J. Gubbi, R. Buyya, S. Marusic, and M. Palaniswami. Internet of Things (IoT): A vision, ar-

chitectural elements, and future directions. Future Generation Computer Systems, 29(7):1645–

1660, sep 2013. ISSN 0167-739X. doi: 10.1016/J.FUTURE.2013.01.010. URL https://www.

sciencedirect.com/science/article/pii/S0167739X13000241.

[30] N. Heuveldop et al. Ericsson mobility report. Ericsson, Stockholm, 2017.

[31] M. Wollschlaeger, T. Sauter, and J. Jasperneite. The future of industrial communication: Automation

networks in the era of the internet of things and industry 4.0. IEEE Industrial Electronics Magazine,

11(1):17–27, 2017.

[32] M. Hilbert. Quantifying the data deluge and the data drought. 2015.

[33] Average costs of industrial Internet of Things (IoT) sensors from 2004 to 2020 (in U.S. dollars). URL

https://www.statista.com/statistics/682846/vr-tethered-hmd-average-selling-price/.

Data provided by Goldman Sachs, BI Intelligence Estimates [Online; accessed 30-January-2019].

[34] L. D. Xu and L. Duan. Big data for cyber physical systems in industry 4.0: a survey. Enterprise

Information Systems, 13(2):148–169, 2019. doi: 10.1080/17517575.2018.1442934. URL https:

//doi.org/10.1080/17517575.2018.1442934.

[35] D. Laney. 3d data management: Controlling data volume, velocity and variety. META group research

note, 6(70):1, 2001.

[36] K.-d. Thoben, S. Wiesner, and T. Wuest. “ Industrie 4.0 ” and Smart Manufacturing – A Review of

Research Issues and Application Examples. International Journal of Automation Technology, 11

(1):4–16, 2017. ISSN 1883-8022. doi: 10.20965/ijat.2017.p0004.

[37] J. Lee, H.-A. Kao, and S. Yang. Service Innovation and Smart Analytics for Industry 4.0

and Big Data Environment. Procedia CIRP, 16:3–8, jan 2014. ISSN 2212-8271. doi: 10.

1016/J.PROCIR.2014.02.001. URL https://www.sciencedirect.com/science/article/pii/

S2212827114000857.

73

[38] S. Yin and O. Kaynak. Big Data for Modern Industry: Challenges and Trends. Proceedings of the

IEEE, 103(2):143–146, 2015. ISSN 00189219. doi: 10.1109/JPROC.2015.2388958.

[39] H. Hu, Y. Wen, T. S. Chua, and X. Li. Toward scalable systems for big data analytics: A technology

tutorial. IEEE Access, 2:652–687, 2014. ISSN 21693536. doi: 10.1109/ACCESS.2014.2332453.

[40] Universal Robots. How Universal Robots sold the first cobot, 2016. URL

https://www.universal-robots.com/about-universal-robots/news-centre/

the-history-behind-collaborative-robots-cobots. [Online; accessed 20-January-2019].

[41] Mitsubishi Electric. Collaborative robots and industry 4.0. URL http://files.messe.de/

abstracts/68353_Mitsubishi_Electric_Puetz.pdf. [Online; accessed 21-January-2019].

[42] M. A. K. Bahrin, M. F. Othman, N. H. N. Azli, and M. F. Talib. Industry 4.0: A review on industrial

automation and robotic. Jurnal Teknologi, 78(6-13):137–143, 2016. ISSN 01279696. doi: 10.

11113/jt.v78.9285.

[43] Cobots: the rise of the collaborative robot (cobot). URL https://www.i-scoop.eu/industry-4-0/

cobot-collaborative-robot/. [Online; accessed 28-January-2019].

[44] V. Villani, F. Pini, F. Leali, and C. Secchi. Survey on human–robot collaboration in industrial settings:

Safety, intuitive interfaces and applications. Mechatronics, 55:248–266, nov 2018. ISSN 0957-

4158. doi: 10.1016/J.MECHATRONICS.2018.02.009. URL https://www.sciencedirect.com/

science/article/abs/pii/S0957415818300321.

[45] P. Gustavsson, M. Holm, and A. Syberfeldt. Human-robot collaboration – towards new metrics for

selection of communication technologies. Procedia CIRP, 72:123–128, jan 2018. ISSN 2212-8271.

doi: 10.1016/J.PROCIR.2018.03.156. URL https://www.sciencedirect.com/science/article/

pii/S2212827118303147.

[46] L. Wang, M. Torngren, and M. Onori. Current status and advancement of cyber-physical systems

in manufacturing. Journal of Manufacturing Systems, 37:517–527, 2015.

[47] V. Paelke. Augmented reality in the smart factory: Supporting workers in an industry 4.0. envi-

ronment. In Proceedings of the 2014 IEEE Emerging Technology and Factory Automation (ETFA),

pages 1–4, Sep. 2014. doi: 10.1109/ETFA.2014.7005252.

[48] Audi Smart Factory - Future of Audi Production. URL https://www.youtube.com/watch?v=

sqCbYd8O8MU&t=613s. [Online; accessed 22-February-2019].

[49] N. Gjeldum, M. Mladineo, M. Crnjac, I. Veza, and A. Aljinovic. Performance analysis of the RFID

system for optimal design of the intelligent assembly line in the learning factory. Procedia Man-

ufacturing, 23:63–68, jan 2018. ISSN 2351-9789. doi: 10.1016/J.PROMFG.2018.03.162. URL

https://www.sciencedirect.com/science/article/pii/S2351978918304669.

74

[50] M. Papic, Z. Bundalo, D. Bundalo, R. Stojanovic, Z. Kovacevic, D. Pasalic, and B. Cvijic. Microcom-

puter based embedded SCADA and RFID systems implemented on LINUX platform. Microproces-

sors and Microsystems, 63:116–127, nov 2018. ISSN 0141-9331. doi: 10.1016/J.MICPRO.2018.

08.009. URL https://www.sciencedirect.com/science/article/pii/S014193311830070X.

[51] K. Domdouzis, B. Kumar, and C. Anumba. Radio-Frequency Identification (RFID) applications: A

brief introduction. Advanced Engineering Informatics, 21(4):350–355, oct 2007. ISSN 1474-0346.

doi: 10.1016/J.AEI.2006.09.001. URL https://www.sciencedirect.com/science/article/pii/

S1474034606000498.

[52] Advanced Mobile Group. RFID’s Role in Industry 4.0, 2017. URL https://www.

advancedmobilegroup.com/blog/rfids-role-in-industry-4.0. [Online; accessed 19-January-

2019].

[53] M. Aksit. The role of computer science and software technology in organizing universities for in-

dustry 4.0 and beyond. pages 5–11, 09 2018. doi: 10.15439/2018F002.

[54] Weidmuller Group. Industry 4.0: an overview from the perspective of a German-headquartered

firm. 2014.

[55] O. Andrs. Using industry 4.0 technologies for teaching and learning in education process. pages

149–156, 09 2018. ISBN 978-3-319-65959-6. doi: 10.1007/978-3-319-65960-2 20.

[56] Beckhoff. Industry 4.0 Forum: PC-based Control Concepts as Core Technology for the Smart

Factory. Beckhoff Automation GmbH, 2014.

[57] Industrie 4.0 and Robots - Roboter Integrated Agent Network (RIAN). URL https://www.ais.mw.

tum.de/en/research/equipment/automatica/. [Online; accessed 16-January-2019].

[58] Beckhoff Smart Manufacturing. URL https://www.youtube.com/watch?v=Z-c3DzUNYaU. [Online;

accessed 22-April-2019].

[59] Industrie 4.0 LIVE auf der AUTOMATICA 2014. URL https://www.youtube.com/watch?list=

PLGD8fyzcQSREaRe2CR6PBr6qrOiQMvLgq&v=tQxWxGYuEIc. [Online; accessed 22-April-2019].

[60] Festo AG & Co. KG. Industry 4.0 Qualification for the factory of the future Indus-

try 4.0 – the future of production What changes can industry expect? 2016. URL

https://www.festo.com/net/SupportPortal/Files/425639/Qualification_I4.0_Product_

Brochure_56745_screen_full_EN.pdf.

[61] C. Academy. Innovations and Trends from Industry for Educational Institutes Automation meets IT.

[62] S. M. Sackey, A. Bester, and D. Adams. Industry 4.0 Learning Factory Didactic Design Pa-

rameters for Industrial Engineering Education in South Africa. South African Journal of In-

dustrial Engineering, 28(1):114–124, 2017. ISSN 22247890. doi: 10.7166/28-1-1584. URL

http://sajie.journals.ac.za/pub/article/view/1584.

75

[63] Kaomi Goetz. How 3M Gave Everyone Days Off and Created an In-

novation Dynamo, 2011. URL https://www.fastcompany.com/1663137/

how-3m-gave-everyone-days-off-and-created-an-innovation-dynamo. [Online; accessed

28-November-2018].

[64] G. Chryssolouris, D. Mavrikios, and L. Rentzos. The Teaching Factory: A Manufacturing Education

Paradigm. Procedia CIRP, 57:44–48, jan 2016. ISSN 2212-8271. doi: 10.1016/J.PROCIR.2016.

11.009. URL https://www.sciencedirect.com/science/article/pii/S2212827116311623.

[65] D. Mavrikios, N. Papakostas, D. Mourtzis, and G. Chryssolouris. On industrial learning and training

for the factories of the future: A conceptual, cognitive and technology framework. Journal of Intelli-

gent Manufacturing, 24(3):473–485, 2013. ISSN 09565515. doi: 10.1007/s10845-011-0590-9.

[66] L. Rentzos, D. Mavrikios, and G. Chryssolouris. A two-way knowledge interaction in manufacturing

education: The teaching factory. Procedia CIRP, 32:31–35, 2015.

[67] G. Chryssolouris, D. Mavrikios, and D. Mourtzis. Manufacturing systems: Skills & competencies

for the future. Procedia CIRP, 7:17–24, 2013. ISSN 22128271. doi: 10.1016/j.procir.2013.05.004.

URL http://dx.doi.org/10.1016/j.procir.2013.05.004.

[68] Kevin Bartlett. A simple UDP communications application, 2010. URL https://www.mathworks.

com/matlabcentral/fileexchange/24525-a-simple-udp-communications-application. [On-

line; accessed 13-November-2018].

76