Available online at www.sciencedirect.

com

ScienceDirect
Procedia Manufacturing 11 (2017) 198 – 206

27th International Conference on Flexible Automation and Intelligent Manufacturing, FAIM2017,

27-30 June 2017, Modena, Italy

A New Model of Modular Automation Programming in Changeable

Manufacturing Systems
Tarek AlGeddawy*
Mechanical and Industrial Engineerig, University of Minnesota, Duluth, 55812, United States

Abstract

Manufacturing systems in Industry 4.0 are changeable, smart, connected and more autonomous. The structure of a changeable
manufacturing system allows for physical reconfiguration, however, reprogramming controllers has been always performed
manually for each new system configuration. The presented model combines different ladder logic codes corresponding to
different system configurations, modularizes them and produces smaller pieces of code, which automatically get merged and
downloaded to the different system controllers. The model uses Cladistics and Design Structure Matrix (DSM) to prepare the
modular codes. A case study of a changeable robotic assembly system is presented.

©
© 2017
2017TheTheAuthors. Published
Authors. by by
Published Elsevier B.V.B.V.
Elsevier This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the 27th International Conference on Flexible Automation and
Peer-review under responsibility of the scientific committee of the 27th International Conference on Flexible Automation and
IntelligentManufacturing
Intelligent Manufacturing.

Keywords. Changeable Manufactring; Modularity; Modular Codes; DSM; Cladistics; Automation; Granularity

1. Introduction

Two relatively new industrial paradigms are shifting manufacturing systems toward more adaptability,
profitability, integration and smartness. The first is ‘changeable manufacturing’, which allows manufacturing
systems to physically change their structure to adapt to changes in product design and market demands. A

* Corresponding author. Tel.. +1-218-726-6810

E-mail address. [email protected]

2351-9789 © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the 27th International Conference on Flexible Automation and Intelligent Manufacturing
doi:10.1016/j.promfg.2017.07.224
 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206 199

manufacturing system has to be flexible, reconfigurable, modular and scalable to be called a changeable system [1].
This paper uses the already existing wealth of research knowledge related to Changeable Manufacturing Systems on
the logical level, when it comes to system and product design, and more specifically in the field of design data
management and design methodologies, and extend that knowledge to link software and codes to system design.
The second paradigm is ‘Industry 4.0’, which allows bidirectional communication and data flow between
different manufacturing domains from machine operations data to product data at the end-customer. The enabling
infrastructure technologies is the ‘Industrial Internet of Things (IIoT)’ within the premises of the manufacturing
facility and the ‘Internet of Things (IoT)’ outside the manufacturing facility [2]. When both paradigms are
integrated, the result is a smarter manufacturing system with more adaptable capabilities to change.
One way of using the IIoT is to automatically download automation codes to the different hardware components
in the system to match the new system configuration and the updated production plans. However, these codes have
to be prepared and become ready for download a priori to perform that targeted automated coding preparation. This
paper is presenting a new Automation Modular Coding (AMC) methodology specially tailored for ladder logic
codes to automate the preparation process of those modular codes.

2. Literature Review

2.1. Changeable manufacturing

Changeability can be seen as an umbrella concept that enables manufacturing systems quantitative (capacity) and
qualitative (capability) changes in response to market demands on many levels of the system, from supply chain
management to shop floor control [3]. The main drivers for changeability are universality, scalability, modularity,
mobility and compatibility from process planning to enterprise management Manufacturing system. Flexibility and
Reconfigurability are the two main facets of changeability [1]. Flexibility is a common practice in industry through
the use of Programmable Logic Controllers (PLC) in automation, Computer-Numerical-Control (CNC) machine
tools, robots and sophisticated material handling systems such as Automatic Storage and Retrieval Systems [4].
System flexibility reacts to anticipated product changes and is built into the system a priori, causing a high capital
investment and sometimes underutilized capabilities [5]. Reconfigurable Manufacturing Systems (RMS) on the
other hand have the advantage of adapting reactively to changes with minimal effort through the addition or
removal of single functional elements [6]. The common element in most of RMS applications is the modular
architecture of the system, which also requires a unified interface between system components. However,
reprogramming for each configuration is still performed manually either a priori in anticipation of that specific
configuration, or after physically forming this configuration.

2.2. Granularity of a system

Granularity is the level of detail of a system, which is the antonym of system-level integration [7]. In
manufacturing systems, granularity is the level of system model details, e.g. departments at the highest level vs.
tools at the lowest level [1]. Determining the best system granularity level affects the entire system architecture,
which is related to its design modularity or integration [8]. At one extreme the system may be completely integrated
and very coarse-grained. At the other extreme every single system component is treated individually resulting in an
extremely fine grained system and un-integrated design [9]. This Granularity notion will be adopted in this paper to
produce modular codes. At the highest level, there is that single massive code that gets downloaded to all
controllers, and the lowest level there is a large number of tiny codes each consists of a single ladder logic rung.
Granularity is also associated with modularity, since system modules represent either coarse or fine system
grains. The Design Structure Matrix (DSM) is the preferred engineering tool to express system elements and
develop their modular architecture. It can be used to describe system components, project activities, design team
members, etc. and their relationships. A component-based DSM is a special case of DSMs, in which a product or a
system components and their interactions are described in a symmetric matrix [10]. Component-based DSM
elements commonly have a 0-1 representation where ‘1’s refer to the existence of a component-to-component
200 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206

interaction while ‘0’s indicate the lack of it. They can also consist of other numbers representing the strength of
interrelationships between system components. Clustering system components into modules is the ultimate goal of
processing a component-based DSM, other than understanding component interactions. Clustering DSMs into
blocks consisting mostly of ‘1’ elements is performed using techniques which differ in the details of the clustering
objective function, but are similar in the main goal. The goal is to maximize the number of ‘1’s and minimize the
number of ‘0’s inside the blocks, or equivalently, minimize the number of ‘1’s and maximize the number of ‘0’s
outside the blocks [11]. This would lead to decoupling of blocks and strong interactions between elements within
formed blocks.

2.3. Cladistics analysis

The common DSM clustering techniques result in a planar scheme of clusters which does not capture how those
clusters are arranged with respect to each other in the system architecture. Cladistics is a hierarchical classification
method (Fig.1) mainly used in the field of biological classification. However, they have been used in organizational
systematics [12], studying evolution in manufacturing [13] and product family design [14]. Cladistics was originally
developed by Hennig [15] to construct cladograms, which are tree-like structures that provide relationship evidence
between a group of entities based on their characters (Fig.1). The objective of construction is to generate the shortest
cladogram length, i.e. best parsimony. A cladogram length is the number of character changes appearing on a
cladogram tree. Better parsimony suggests a better representative hypothesis of the relationships, hence having the
least information content. A handful of specialized software packages are dedicated to cladogram construction such
as Hennig86, PAUP, NONAME, PeeWee and Phylip, and can cluster large data sets very quickly. The NONAME
software is used in this paper to perform all Cladistics analyses. Cladistics has been also used in DSM clustering by
AlGeddawy and ElMaraghy [9] not only to group components into clusters but also to establish a hierarchical
structure of these components. A similar technique will be represented in this paper to cluster large automation
codes and prepare smaller modular codes.

Fig. 1. A Characteristics Table and the Corresponding Best Cladogram

3. Automation Modular Coding Methodology

The Automation Modular Coding (AMC) methodology has five steps:

3.1. Ladder logic codes consolidation:

In this step, Ladder logic codes are collected from the controllers of the changeable system at every single past
configuration. Since most inputs and outputs will be shared among those configurations, it is convenient to refer to a
specific input or output using a fixed notation, i.e. motor 3 on conveyor 2 is referred to as Y32 and should be kept
the same for all subsequent codes. Then ladder logic rungs referring to the same output are consolidated (Fig.2) into
a single rung, since logic equations can be added together and simplified.
 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206 201

3.2. Output DSM Preparation

Consolidated code is then converted into a simple incidence matrix, in which all outputs inputs (Xs) and outputs
(Ys) are related to other outputs (Ys). Each ladder logic rung ends with an output and contains many inputs, each
existing input will be given a ‘1’ in relation to rung output, or ‘0’ otherwise (Fig.3). Sometimes, other outputs may
also exist in the rung as inputs, therefore they have to be also represented in the incidence matrix. The Ys to Xs part
of the incidence matrix is then multiplied to its transposed matrix to remove the Xs and obtain a symmetrical matrix
that only contain Ys to Ys relationships (Fig.3). The Ys to Ys part of the incidence matrix is then added to the
previously obtained symmetrical matrix. If one Y exists as an input in a rung, its incidence matrix element should be
multiplied by the number of Xs in its own rung. In the shown example in (Fig.3) Y3 exists in Y4 and Y5 rungs, and
originally has three Xs (inputs) in its own rung, therefore its elements (Y3 to Y4, Y3 to Y5) contain ‘3’. The final
Output DSM (ODSM) matrix is still symmetrical and represents the DSM relating all the outputs and showing the
strength of their relationships.

Fig. 2. Ladder Logic Consolidation

Fig. 3. A Simple Example of the Ladder Logic Conversion Process to an ODSM

202 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206

3.3. Modularity analyses using cladistics

Cladistics analysis is then performed on Output DSM. The result is the shortest cladogram (Fig.4) that support
relationships between the outputs (Ys). The Ys in the ODSM can then be recorded to match the Ys at the terminals
of the shortest Cladogram. Dissecting that cladogram at each node results in the different granularity levels of the
DSM. Each level represents a modularity scheme, such as the four levels represented in (Fig.4). Top level 1
represents a single module that is associated with the whole consolidated code, while bottom level 4 represents five
separate modules, which are associated with four small codes, each only contains the single rung of a specific
output.

Fig. 4. Cladistics Analysis and Modularization Process of the Simple Example

3.4. Granularity level determination based on Modularity Index

In this step, the best granularity level is determined between the two extremes, the top level in which a single
code is presented and bottom level in which every output is separated in its own code. A modularity index is
calculated following equation 1. Table 1 shows the MI for the four granularity levels of the simple example in Fig.3
and 4. It shows that level 2 is the best one with the maximum MI. This level generates two codes, one with rungs of
outputs {Y1,Y2}, and the other with outputs {Y4,Y3,Y5} according to Fig.4.
MI = Ain-Aout , Ain=Ein/Nin , Aout=Eout/Nout (1)
Where: Ain is the average of the numbers in the ODSM elements inside modules
Aout is the average of the numbers in the ODSM elements outside modules
Ein is the summation of the numbers in the ODSM elements inside modules
Eout is the summation of the numbers in the ODSM elements outside modules
Nin is the number of ODSM elements inside modules
Nout is the number of ODSM elements outside modules

Table 1. The 4 Corresponding Modularity Indices of the Simple Example

 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206 203

3.5. New code generation for future configurations

After determining the next changeable system configuration and the required hardware components (inputs and
outputs), the corresponding Ys are specified and their associated codes are pulled from the pool of prepared modular
codes in previous steps. For example, if in the simple example in Fig.3 a new configuration has only Y3 and Y5,
then second code will be downloaded to the controller. If it has Y2, Y3 and Y4 exists, then both codes will be
downloaded.

4. Case Study

The automated system in this study is called the ‘Cyber-Physical Factory’ (CP Factory) from Festo Didactic. It is
a modular, changeable and robotic assembly system that can change structure and layout quickly. Fig. 5 shows one
configuration of the system. Multiple configurations of the CP Factory have been created and their ladder logic
programs have been reduced to 28 outputs (Ys), since other dependent outputs were logically merged. In the first
step of the AMC model ladder logic rungs have been consolidated to obtain the minimum number of rungs, which is
exactly equal to the number of Ys. In step 2, all the 28 Ys have been converted into a single ODSM (Fig.6), by
converting all the rungs into an incidence matrix showing relationships between of inputs and outputs (Xs and Ys),
and then multiplying the Xs-to-Ys part of that matrix to its transpose and adding it to the Ys-to-Ys part, as explained
previously in section 3.2.

Fig. 5. An Instance of the CP Factory from Festo Didactic

Fig.6. The ODSM of the CP Factory

204 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206

The ODSM is then analyzed in step 3 of the AMC model for modularity using Cladistics analysis. The
NONAME software [16] has been used to generate the shortest cladogram, which is shown in Fig.7, presenting a
total of 14 possible granularity levels, each level corresponds to a splitting point of two branches on the tree. The
cladogram also shows a new arrangement of the outputs which is used to reorder columns and rows of the ODSM.
The new reordered ODSM is shown next to the cladogram in Fig.7. In step 4, Modularity Index (MI) is evaluated
for each level following equation 1 and its values are shown in Table 2. The highest MI value is found at level 4,
which dissects the reordered ODSM into four modules as shown in Fig.7. These modules are program-
1:{6,26,8,19,22,14,20}, program-2:{2}, Program-3:{12,13,23,1,16,15,4,3,27}, program4:
{17,7,11,5,10,24,25,9,21,18,28} and they represent the four modular programs that will be used for future system
configurations. In step 5, the new planned configuration only needs the outputs of the indices
1,3,4,6,8,13,14,15,16,19,20,23 and 26. Fig. 8 shows how these Ys are ordered according to the reordered ODSM to
pull out the relevant codes. In this case, program 1 and program 3 are selected and downloaded to the controller.

Fig. 7. The Best Cladogram of the CP Factory and the Reordered ODSM

Table 2. The 14 Corresponding Modularity Indices of the CP Factory

 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206 205

Fig. 8. The Downloaded Programs to the new CP Factory Configuration

5. Conclusions

A new Automation Modular Coding (AMC) model has been presented in this paper to modularize existing ladder
logic codes and automatically generate the needed codes for future system configurations. The model consists of
five steps, in which it first consolidates existing codes to reduce the amount of data, then it converts codes and rungs
to a DSM that relates all the outputs, which is then analyzed using Cladistics analysis to produce a Cladogram that
represents the granularity map of the consolidated code. To determine the best granularity level, Modularity Index
(MI) is evaluated at each granularity level, then modular codes are prepared at the highest MI level. Finally codes
for future system configurations are generated by determining the corresponding prepared modular codes based on
the required outputs. Future work of this research will extend the model to handle other ladder logic codes such as
timers, counters, block functions, sub-routines, etc. It is also planned to handle other forms of automation codes,
such as flowcharts and G-codes.

References

[1] H.P. Wiendahl, H.A. ElMaraghy, P. Nyhuis, M.F. Zah, H.H. Wiendahl, N. Duffie, M. Brieke, Changeable Manufacturing - Classification,
Design and Operation, CIRP Annals - Manufacturing Technology, 56 (2007) 783-809.
[2] S. Wang, J. Wan, D. Li, C. Zhang, Implementing Smart Factory of Industrie 4.0: An Outlook, International Journal of Distributed Sensor
Networks, 2016 (2016).
[3] H. ElMaraghy, G. Schuh, W. Elmaraghy, F. Piller, P. Schonsleben, M. Tseng, A. Bernard, Product variety management, CIRP Annals -
Manufacturing Technology, 62 (2013) 629-652.
[4] B. Babic, Axiomatic design of flexible manufacturing systems, International Journal of Production Research, 37 (1999) 1159-1173.
[5] T. Tolio, D. Ceglarek, H.A. ElMaraghy, A. Fischer, S.J. Hu, L. Laperriere, S.T. Newman, J. Vancza, SPECIES-Co-evolution of products,
processes and production systems, CIRP Annals - Manufacturing Technology, 59 (2010) 672-693.
[6] Y. Koren, Reconfigurable Manufacturing Systems, in: CIRP 1st International Conference on Agile and Reconfigurable Manufacturing, Ann
Arbor, 2002.
[7] M.E. Sosa, S.D. Eppinger, C.M. Rowles, Identifying modular and integrative systems and their impact on design team interactions, Journal of
Mechanical Design, Transactions of the ASME, 125 (2003) 240-252.
[8] N. Chiriac, K. Holtta-Otto, D. Lysy, S. Eun Suk, Level of Modularity and Different Levels of System Granularity, Journal of Mechanical
Design, 133 (2011) 329-339.
[9] T. AlGeddawy, H. ElMaraghy, Optimum Granularity Level of Modular Product Design Architecture, CIRP Annals, 62 (2013) 151–154.
[10] S.D. Eppinger, T.R. Browning, Design structure matrix methods and applications, MIT Press, Cambridge, Mass., 2012.
[11] K. Holtta-otto, N.A. Chiriac, D. Lysy, E. Suk Suh, Comparative analysis of coupling modularity metrics, Journal of Engineering Design, 23
(2012) 787-803.
[12] C. Rose-Anderssen, J. Baldwin, K. Ridgway, P. Allen, L. Varga, M. Strathern, A cladistic classification of commercial aerospace supply
chain evolution, Journal of Manufacturing Technology Management, 20 (2009) 235-257.
[13] H. ElMaraghy, T. AlGeddawy, A. Azab, Modelling evolution in manufacturing: A biological analogy, CIRP Annals - Manufacturing
Technology, 57 (2008) 467-472.
[14] H. ElMaraghy, T. AlGeddawy, New dependency model and biological analogy for integrating product design for variety with market
requirements, Journal of Engineering Design, 23 (2012) 719-742.
206 Tarek AlGeddawy / Procedia Manufacturing 11 (2017) 198 – 206

[15] W. Hennig, Phylogenitic Systematics, University of Illinois Press, Urbana, (1966), republished in 1999.
[16] P. Goloboff, NONAME ver. 2, (1999).