Exploring the Use of MES Functionalities in Smart Manufacturing

(1)

Exploring the Use of MES Functionalities in

Smart Manufacturing

A Multiple-Case Study on Their Functional Distribution

Dissertation towards the Degree Master of Science In the Study Program

MSc/MSc Operations Management (Dual Award)

Presented by

Name Lukas Klaiber

Student ID S3903672 (Groningen) S180620962 (Newcastle)

Presented to

1st_Supervisor _{Dr. J.A.C Bokhorst}

2nd_Supervisor _{Dr. Adrian Small}

Submission Date 09.12.2019

(2)

Abstract

(3)

Acknowledgement

And this is it – The last chapter of my studies comes to an end. On completion of this dissertation, I will be graduating with two degrees from Newcastle University and University of Groningen. I am thankful for all that I have been able to experience over the past six years, and I wish to thank all the people I have met along the way, personal or academic. Each and every one has somehow or other left a mark on this work.

(4)

List of Figures

Figure 1 The plant information model ... 3

Figure 2 Generic activity model of manufacturing operations management ... 4

Figure 3 Data granularity and planning horizons ... 10

Figure 4 Functional assignment in component manufacturing and assembly plants. ... 12

Figure 5 Manufacturing process types imply different characteristics ... 14

Figure 6 Conceptual Model ... 15

Figure 7 Research process path ... 16

Figure 8 Classification of cases ... 17

Figure 9 Adjusted functional distribution of case company A ... 22

Figure 10 Implication for application functionalities at case A ... 26

Figure 11 Adjusted functional distribution of case company B ... 29

Figure 12 Implication for application functionalities at case B ... 33

Figure 13 Adjusted functional distribution of case company C ... 37

Figure 14 Implication for application functionalities at case C ... 40

Figure 15 Determination of manufacturing process types ... 78

Figure 16 Procedure for determination of adjusted functional distribution ... 79

Figure 17 Implied AFs of standard MES ... 80

Figure 18 Distribution of MES functionalities based on interview data (case A) ... 89

Figure 19 Distribution of MES functionalities based on interview data (case B) ... 97

(8)

List of Tables

Table 1 Activities and their technical integration according to IEC 62264-3:2007 ... 4

Table 2 Typical modules of ERP systems in a manufacturing company ... 5

Table 3 MES functionalities as defined by MESA International ... 9

Table 4 Summary of application functionalities (AFs) relevant in MES ... 11

Table 5 Manufacturing parameters of case companies A-C ... 17

Table 6 Case descriptions ... 18

Table 7 Manufacturing characteristics at company A ... 20

Table 8 Case A – Description of functional distribution and implications for AF ... 23

Table 9 Manufacturing characteristics at company B ... 28

Table 10 Case B – Description of functional distribution and implications for AF ... 30

Table 11 Manufacturing characteristics at company C ... 35

Table 12 Case C – Description of functional distribution and implications for AF ... 37

Table 13 Summary of manufacturing environments ... 42

Table 14 Summary of distribution of MES functionalities after discussion ... 42

Table 15 Summarized implications of application functionalities per function ... 43

Table 16 Summary of how factors influence distribution ... 51

Table 17 ERP functions supporting planning and execution activities ... 60

Table 18 SCM system functions ... 62

Table 19 Five main DSS applications relevant for manufacturing systems ... 63

Table 20 Time planning for case study ... 67

Table 21 Overview data collection tasks ... 69

Table 22 Overview data collection methods ... 70

Table 23 Detailed function definitions (verified through expert interviews) ... 73

Table 24 Case A - Manufacturing characteristics (original data) ... 81

Table 25 Case A - Explorative factors (original data) ... 82

Table 26 Data collection by sub functions - case A ... 85

Table 27 Case B - Manufacturing characteristics (original data) ... 90

Table 28 Case B - Explorative factors (original data) ... 91

Table 29 Data collection by sub functions - case B ... 93

Table 30 Case C - Manufacturing characteristics (original data) ... 98

Table 31 Case C - Explorative factors (original data) ... 99

(9)

List of Abbreviations

AI Artificial Intelligence

AF Application Functionality APS Advanced Planning Systems B1 Case B – Chemical Department B2 Case B – Assembly Department BI Business Intelligence

CAM Computer-Aided Manufacturing CAPP Computer-Aided Process Planning CAx Computer-Aided Technologies CPS Cyber-Physical Systems DCM Dynamic Case Management DCS Distributed Control System DSS Decision Support System ERP Enterprise Resource Planning ETL Extract, Transform and Load HMI Human-Machine Interface

HR Human Resources

IS Information Systems I4.0 Industry 4.0

IoT Internet of Things

LMS Learning Management System MES Manufacturing Execution Systems

MESA Manufacturing Execution Systems Association MOM Manufacturing Operations Management MRPII Manufacturing Resource Planning ODS Operational Data Store

OLAP Online Analytical Processing PDM Product Data Management PLC Programmable Logic Controller PLM Product Lifecycle Management PMC Process Monitoring and Control QMS Quality Management System RQ Research Question

SCADA Supervisory Control and Data Acquisition SCM Supply Chain Management

SM Smart Manufacturing TDM Tool Data Management

TPS Transaction Processing Systems TRS Time Registration System WIP Work In Progress

(10)

1 Introduction

The manufacturing industry has been undergoing fundamental changes, with smart manufacturing (SM) potentially marking the beginning towards the next industrial revolution (Kusiak, 2018). It seems that existing methods of value creation in manufacturing can no longer handle the increasing complexities (Hofmann and Rüsch, 2017). Thus, manufacturing strategies like the German Industry 4.0 (I4.0), the US Industrial Internet program and China’s Made in

China 2025 were developed to attain SM (Tao and Qi, 2017). These programs built on the idea

to virtually and digitally connect people and machines in cyber-physical systems (CPS) through internet of things (IoT), so that “human beings, machines and resources communicate with each other as naturally as in a social network” (Kagermann et al., 2013, p. 19).

SM emphasizes the collection, transmission and sharing of real-time data, which transformed into knowledge, should enable manufacturing intelligence1_{throughout the entire organization}

(O’Donovan et al., 2015). In achieving this, relevant information systems (IS) were identified, i.e., manufacturing execution systems (MES), enterprise resource planning (ERP), supply chain management (SCM) and product lifecycle management (PLM) systems (Mittal et al., 2017; Mantravadi and Møller, 2019). Except for MES, all systems are represented in the enterprise level as described by the ISA–95.00.01–2000 standard. It defines functional data flows and activities between the functional hierarchies (Unver, 2013). Accordingly, MES build a bridge between enterprise level IS (e.g. ERP) and shop floor automation systems e.g., supervisory control and data acquisition (SCADA). This enables sharing of real-time data from production and translation of global plans into detailed operational production schedules (ISA, 2000).

“As real-time data collection, integration, and analysis continue to become more critical to manufacturing business success, MES continues to evolve to address the challenge – and to become more integral to an overall smart manufacturing transformation” (Panchak, 2018, para. 6).

With MES in SM, companies become more competitive by increasing performance, quality and agility, but also by reducing lead times and cost through decentralized and real-time production control (Almada-Lobo, 2015). Yet, the most difficult challenge is the functional integration (i.e., that all components work well together) with other systems to ensure smooth interaction. This necessitates clear specification of functional boundaries (Govindaraju and Putra, 2016) to avoid isolated proprietary systems with inefficient data flows and functional redundancies. Thus, organizations must emphasize tightly integrated IS that provide direct feedback and unique data from manufacturing (Lu et al., 2016). This has led to several standards that provide guidance in defining MES, its functions and integration within enterprise IT architectures. In practice and theory, the ISA-95 and MESA (MES Association) are the most frequently used standards to define MES requirements. However, a study by Schmidt et al. (2011) points out that current standards attempt to cover all production requirements across industries at the expense of industry specific standards. The standards “are deeply rooted in the chemical industry” (p. 304) which, for instance, hinders automotive companies from applying them. MES

1_{Manufacturing intelligence is the capability of good decision-making in manufacturing, through understanding}

(11)

and ERP also support similar functions leading to redundancies between both systems. In fact, modern ERP systems incorporate shop floor functionalities (Boguski and Holzmann, 2011; Arica and Powell, 2017) which becomes an issue if e.g., ERP incorporates MES short-term planning with single item data collection (Koch et al., 2010; Romero and Vernadat, 2016). If companies deploy multiple IS, analysis and clear assignment of functions must be stressed to avoid redundancy of functions and data (Witsch and Vogel-Heuser, 2012). However, Schmidt

et al. (2009) conclude that “requirements on MES differ significantly depending on the industry

and the type of production” (p. 21) which is why an analysis on manufacturing environments and the effect on the distribution of MES functionalities2_{becomes indispensable. Therefore,}

the first research question (RQ) of this work states:

“How do different manufacturing characteristics influence the assignment of MES functions to MES and other IS?”

However, this RQ focuses on manufacturing characteristics only and neglects the relation to industry specific factors which, according to Schmidt et al. (2009), have significant impacts on the configuration of MES functions. To explore and analyze other factors, the second RQ states:

“What are additional (explorative) factors and how do they influence the assignment of MES functions to MES and other IS?”

An issue that arises from different distributions, is the requirement of application functionalities (AFs). AFs are usually enabled by a single underlying technology. For instance, transaction

processing is enabled by transaction processing systems (TPS) constituting the backbone of

ERP, whereas product data management (PDM) systems provide document control to integrate engineering related data within PLM (Waschull et al., 2019). The authors also found that MES are technologically heterogeneous and enable interfaced AFs, that is, they converge different AFs into one single application. For example, MES are considered decision support systems (DSS) for immediate response in production due to storage and analysis of historic process data, real-time and closed loop process control, but also prognostic of future process runs (Naedele

et al., 2015). This also implies analytical and interactive planning capabilities to MES which

is supported by business intelligence (BI) and advanced planning systems (APS) to analyze current production situation and instantly re-plan schedules and sequences (Saenz de Ugarte et

al., 2009). Additionally, Waschull et al. (2019) found that ‘document control’ and ‘performance

analysis’ require a single type of AF only. If these functions shift between IS, the requirements for AFs change or shift accordingly between the respective systems. Thus, the third RQ states:

“How do different configurations of MES functions influence the requirement for application functionalities within MES and other IS?”

(12)

2 Theoretical Background

This section elaborates on the different concepts relevant for this study. Figure 1 abstracts the hierarchical levels and IS in a classic organizational architecture. The tactical (top) level shows the relevant enterprise IS and MES at the operational manufacturing operations management (MOM) level to bridge enterprise applications and automation systems at shop

floor level. For this, the literature refers to current standards such as ISA–95.00.01–2000, which

describe the interfaces between enterprise (ERP) and execution (MES) activities, but also how common objects (e.g. material lots with unique properties) are normally exchanged between the enterprise levels.

Figure 1 The plant information model (MESA International, 1997). Extended by PLM and hierarchy indication (ISA, 2000;

Holligan et al., 2017).

The model is also used as guidance through the following sub-sections: Firstly, in (2.1) a closer look is taken at activities related to MES (dark grey box) based on the ISA–95.00.01–2000 set of standards. Such standards constitute the basis for the functionalities defined by MESA and to set the scope of functions relevant for this research. Secondly, the IS at tactical level and MES are conceptualized sequentially as they appear in the model including their underlying technologies and application functionalities (AFs). That is, in (2.2) ERP systems are explained first, followed by SCM systems in (2.3), PLM systems that constitute the overarching concept of PDM in (2.4). Holligan et al. (2017) show that product and process engineering are only relevant for design and prototyping which is why CAD, CAPP and CAM are not considered in this research. MES and how the functional distribution is affected by varying manufacturing environments is explained in (2.5). For PLM, Lastly, the manufacturing process types are defined to lay the foundation for classification of different production environments.

2.1 Manufacturing Operations Management

Part 1 of the ISA–95.00.01–2000 set of standards describes tactical and operational level functions for all departments of an enterprise and which common objects are usually exchanged. The defined models, concepts and terminologies provide standards for efficient integration of the tactical (4) and operational (3) level (see Figure 1). The characteristics and attributes of objects exchanged are defined in part 2 of the standard. In part 3 of the standard, functions and activities are described related to the management of operations in both the manufacturing and

MES

Integrated Production Data Working with Operations Management Systems, People and Practice

Controls Data Collection SCADA PLC/ SoftLogic DCS / OCS Manual Process Control

Automation, Instruments, Equipment

Tactical Level Operational Level Shop-Floor Level Enterprise Applications (Level 4) Manufacturing Operations Management Systems (Level 3) Supervisory Control Systems (Level 2) Production Process

Sensing & Manipulation (Level 1)

Level 0

SCM ERP

Product & Process Engineering

PDM

CAPP CAM CAD

(13)

process industry. These activities are shown in Figure 2 and can be assigned to the MOM (i.e. MES) level and be divided into four functional areas: (1) production-, (2) quality-, (3) maintenance- and (4) inventory operations management. The tasks of each activity however, differ depending on the functional area. This means that while e.g., “detailed scheduling” is to set up detailed and optimized production schedules in the functional area of production, it is to examine, prioritize and integrate time-related maintenance tasks into schedule within the maintenance area (Theobald et al., 2011).

Figure 2 Generic activity model of manufacturing operations management according to

IEC 62264-3:2007, fig. 6 (CENELEC, 2007, p. 17)

These generic activities were also used to define the MESA standard of functions. The MESA standard covers all of these activities and adds an additional three functions that perform maintenance-, quality- and inventory management activities. Therefore, the MESA standard provides eleven functionalities It can be seen from existing architectures that some activities are implemented at different enterprise levels and systems, that is, as function of MES or ERP, or in process control systems (Theobald et al., 2011). For this, the activities illustrated in Figure 2 can be applied uniformly across each functional area of the MES level. Depending on the functional area, activities are assigned to different levels as abstracted in the following table:

Table 1 Activities and their technical integration according to IEC 62264-3:2007 (CENELEC, 2007)

Maintenance Production Quality Inventory

Detailed Scheduling ● ○ ● ● Resource Management ● ○ ● ● Tracking ● ○ ● ● Dispatching ○ ○ ○ ● Analysis ● ○ ● ○ Definition Management ○ ○ ○ ● Data Collection ○ ○ ○ ○ Execution Management ○ ○ ○ ○ Detailed Scheduling Resource Management Tracking Dispatching Data Collection Analysis Definition Management Execution Management Operations

Definition Operations Capability Operations Request Operations Response

Level 1 and 2 Functions

Equipment and Process Specific Rules

and Instructions Commands Responses

(14)

It can be seen from the table above that all generic activities can be assigned to MES level in the production area, which signals the significance of MES systems for production. This goes along with the MESA standard that provides a functionality in MES for each activity plus three additional functions for maintenance, quality and inventory. Therefore, it can be assumed that according to MESA, all MES functions apart from maintenance-, quality- and inventory management, are mainly to support the functional area of production.

2.2 Enterprise Resource Planning Systems

The term enterprise resource planning (ERP) is generally used for transactional IS designed to integrate and manage an enterprise’s resources. ERP systems are software-driven business management tools that streamline data across an organization and integrate all business functions and information in support of tactical operations planning (Algeo and Barkmeyer, 1998; Chou et al., 2005; Bradford, 2015). Thus, cross-functional integration brings together processes and data executed in all departments of a company from core functions (e.g. manufacturing and operations) to support functions such as accounting or finance (Oztemel and Gursev, 2018). ERP systems were designed to process large amounts of business transactions and normally comprise the following main components: client server systems, centralized enterprise database as single source of truth and various application modules (Yen et al., 2002; Bradford, 2015). They integrate all business data into a centralized enterprise database, but cannot carry out extensive data analysis and decision support (Chou et al., 2005) although some functions can be perceived as decision support functions (Holsapple and Sena, 2003). In fact, ERP systems offer benefits that could not be achieved with only isolated departmental IS. This is because their functions connect and execute cross-functional transactions. The main benefits comprise (1) an overall representation of the organizational structure comprising all functions and departments but also (2) a logically centralized database where all enterprise transactions are entered, recorded, processed, monitored and reported (Algeo and Barkmeyer, 1998). The following functions3_{enable such cross-functional integration (Bradford, 2015, p. 3):}

Table 2 Typical modules of ERP systems in a manufacturing company (Bradford, 2015, p. 3)

Operations and Supply Chain

Plant Maintenance Purchasing Quality Management

Sales and Distribution Shop Floor Management Transportation Management Manufacturing Warehouse Management Advanced Planning

Financial Accounting

General Ledger Cash Management Accounts Payable

Accounts Receivable Fixed Assets Financial Consolidation Management Accounting

Cost Center Accounting Product Costing Budgeting

Profit Center Accounting Activity-Based Costing Profitability Analysis Human Capital Management

Personnel Management Payroll Learning Management

Time and Attendance Benefits Recruitment Management

(15)

ERP systems are characterized by complex interrelationships of business processes between the modules as shown in Table 2, which is why they cover cross-functional transactions between functions (Algeo and Barkmeyer, 1998). For this, transaction processing systems (TPS) with central relational databases constitute the backbone technology of ERP systems (Waschull et

al., 2019).

Transaction Processing Systems (TPS)

TPS provide transaction processing capabilities constituting the information backbone of manufacturing enterprises (Romero and Vernadat, 2016). They embody tight interdependencies between organizational functions (Ross and Vitale, 2000) and integrate them by allowing various modules to transfer data and information via a centralized database that is accessible by all modules and departmental systems (Chen, 2001). Transactions can be defined as records of resource modifications (e.g. materials, financials and labor) in or between companies whose transaction management can be seen as main undertaking of an ERP system. For instance, changes to principal resources (e.g. transportation and delivery of materials to next process) are recorded in central, relational databases and made available to users and other connected software modules and technologies (Algeo and Barkmeyer, 1998).

2.3 Supply Chain Management Systems

Supply chain management (SCM) comprises all activities undertaken by an organization from procurement of raw materials, transformation into finished goods and their delivery to the customer (Christopher, 2016). Within SCM, IT has always been important to support processes and manage transactions along the supply chain (Gunasekaran and Ngai, 2004). Some SCM modules are implemented as extension to ERP, hence providing transaction processing with limited planning and decision support capabilities (Chen, 2001). Due to shortcomings of ERP in production planning, scheduling, reporting and monitoring, one typically finds advanced planning systems (APS) and business intelligence (BI) systems besides ERP in the IT landscape (Stadtler, 2005; Hahn and Packowski, 2015). Operational SCM systems focus on functionalities to optimize planning and scheduling accuracy in production to have positive benefits for inventories, safety stocks and order quantities (Chen, 2001; Kadre, 2011). For this reason, SCM systems are also considered APS (Jonsson et al., 2007) which can be further substantiated as model-driven DSS that not only provide prescriptive analytics for strategic and operational supply chain planning (Power and Sharda, 2007), but also scenario analysis and simulation capabilities. On the other hand, BI systems are data-driven DSS that offer online analytical

processing (OLAP) where they can access and analyze large datasets through application of descriptive analytics for reporting and monitoring (Power and Sharda, 2007; Dickersbach,

2009). For this reason, the following to section briefly conceptualizes BI and APS.

Business Intelligence (Data-Driven DSS)

(16)

production before they emerge (Felsberger et al., 2016). They consist of database and model management systems to access, manage and manipulate data stored in external or internal databases via extract, transform and load (ETL) as well as to prepare and analyze complex data by applying mathematical and analytical models via OLAP (Romero and Vernadat, 2016). For the reasons mentioned above, this work considers BI to provide required analytics capabilities.

Advanced Planning Systems (Model-Driven DSS)

APS are considered model-driven DSS that follow a prescriptive approach through employment of sophisticated mathematical algorithms to help planners at various levels developing near-optimal plans and solutions (Chen, 2001; Romero and Vernadat, 2016) by providing interactive

planning capabilities (Waschull et al., 2019). APS often operate on top of ERP, extracting

transactional data from the centralized database to perform calculations and send finalized plans as a result back to ERP (Jonsson et al., 2007). APS attempt to bridge supply chain complexity and operational decisions based on integral planning, true optimization and hierarchical planning (Fleischmann and Meyr, 2003). Unlike ERP, APS consider finite resource capacities and other constraints to increase planning accuracy and reliability. The main features of APS relevant to production planning and scheduling are as follows (David et al., 2006, p. 707):

(1) System input: Characteristics, customer requirements, planned orders and production orders, routings and their alternatives, but also inventory information for calculations. (2) System output: APS determine job allocations and sequencing to resources with finite capacities, equipment timetables, choices of bill of materials and alternative routings. (3) Control and planning requirements: APS use constraints (e.g. material availability,

machine and labor capacity, safety stock levels, sequencing conditions, set-up times) to maximize profit and minimize costs.

(4) Plan and schedule generation: The techniques vary from single algorithms to sets of algorithms from which the user can choose. Metaheuristics such as genetic algorithms are mostly used in operational planning.

2.4 Product Lifecycle Management Systems

(17)

process designs and production capabilities (Denkena et al., 2007). However, some organizations only employ PDM systems that manage geometric product data, drawings and documents from CAx applications (Sudarsan et al., 2005). Product and process engineering activities are not part of the manufacturing execution process and therefore out of scope for MES. Nonetheless, PDM is considered the main technology of PLM which is further explained in the following section.

Product Data Management (PDM) systems

Product data management (PDM) systems are positioned at the enterprise IS level with ERP and SCM that execute financial and logistical activities required for production (Moones et al., 2015) and give support to lifecycle activities (e.g. design, sign-off, data sharing, tracking engineering changes and control of product configurations) in complex PLM environments. Global competition to enhance productivity has become the driver in efficiently managing product data which can also include paper-based activities (Stark, 2015). Whereas ERP systems support continuously occurring and repetitive transactions in the product delivery process of manufacturing (e.g. arriving orders, jobs completed), PDM systems support rather non-transactional, recursive and iterative, intellectual activities of product development processes such as engineering drawings and specifications (Terzi et al., 2010). According to Stark (2015) and Kumar and Midha (2006), PDM systems must at least have following modules to provide full benefits to an organization (Kropsu-Vehkapera et al., 2009, p. 760):

• Information warehouse

• Information warehouse management: Tracing data related actions • Document Management

• Configuration management •_{Product structure management} •_{Product and workflow structure} • Workflow and process management • System administration management

2.5 Manufacturing Execution Systems

As shown in Figure 1, manufacturing execution systems (MES) were designed to bridge shop floor and business planning levels to provide a more granular and very detailed view on production data and to compensate for the incapability of ERP systems. Thus, MES control and support production processes and enable vertical integration of real-time manufacturing data generated on the shop-floor to a more aggregated data view at ERP level (Wannenwetsch and Nicolai, 2004; Louis and Alpar, 2007). The ultimate goal is to increase manufacturing transparency and create closed control loops in vertical and horizontal direction. This is why the following characteristics of an MES are essential (Kletti, 2007):

(1) Highly detailed data acquisition from production

(2) Reactive planning with relatively short planning horizons

(18)

The question of which functions an MES has to provide, led to several standards. The most prominent were proposed by ISA-954_{, Verein Deutscher Ingenieure}5_{(VDI) and MESA and}

assign nearly identical functions to MES (Iarovyi et al., 2016). The early definition by MESA International (1997) states that “Manufacturing Execution Systems (MES) deliver information that enables the optimization of production activities from order launch to finished goods. Using current and accurate data, MES guides, initiates, responds to, and reports on plant activities as they occur” (p. 1). MESA is an American association of various industrial software firms that focus on MES only. Their guidelines comprise eleven MES functionalities that have reached acceptance in the global manufacturing industry through a pragmatic and comprehensive approach (MESA International, 1997, pp. 5-6):

Table 3 MES functionalities as defined by MESA International (1997, pp. 5-6)

Function Description

Operations/Detailed Scheduling Sequencing and timing activities for optimized plant

performance based on finite capacities of the resource.

Resource Allocation and Status Guiding what people, machines, tools, and materials should do

and track what they are currently doing or have just done.

Dispatching Production Units Giving the command to send materials or orders to certain parts

of the plant to begin a process or step.

Document Control Managing and distributing information on products, processes,

designs, or orders, as well as gathering certification statements of work and conditions.

Product Tracking Monitoring the progress of units, batches, or lots of output to

create a full history of the product.

Performance Analysis Comparing measured results in the plant to goals and metrics set

by the corporation, customers, or regulatory bodies.

Labor Management Tracking and directing the use of operations personnel during a

shift based on qualifications, work patterns, and business needs.

Maintenance Management Planning and executing appropriate activities to keep equipment

and other capital assets in the plant performing to goal.

Process Management Directing the flow of work in the plant based on planned and

actual production activities.

Quality Management Recording, tracking, and analyzing product and process

characteristics against engineering ideals.

Data Collection and Acquisition Monitoring, gathering, and organizing data about the processes,

materials, and operations from people, machines, or controls. Furthermore, those functionalities can be individually adjusted to the needs of an organization, which is why MES were originally designed to be configurable on a modular basis in order to meet the specific organizational goals. However, it sometimes happens that MES supports functionalities that are similar to those of other systems (mainly ERP), but focus slightly more

(19)

on production performance. If this is the case, the said overlap results in redundancy within the respective system (Saenz de Ugarte et al., 2009).

2.5.1. Scope of Different Enterprise Layers

The functional enterprise hierarchies differ in terms of data granularity and planning horizons as shown in Figure 3. The tactical level at the top provides sets of enterprise wide functions and an integrated database (i.e. relational database of ERP system). The level of detail is rather low, providing an aggregated view on data collected from automation systems which usually provide real-time data at high level of detail. Planning horizons greatly differ between layers where ERP is used for corporate long-term planning (i.e. days to months), whereas automation systems operate in milliseconds to hours. MES consolidate shop floor data and make it available to ERP. Simultaneously, MES translate the long-term plans from ERP into operational plans and orders.

Figure 3 Data granularity and planning horizons (Louis and Alpar, 2007; Witsch and Vogel-Heuser, 2012) 2.5.2. Converged Technologies and AFs

In contrast to classic technologically homogeneous IS (e.g. ERP), that are supported by a single underlying technology, MES were identified to be technologically heterogeneous (Waschull et al., 2019). That is, they constitute an interfaced functionality converging various AFs enabled by different technologies which allows MES to take on central integrating roles. Again, these functionalities (technologies) are transaction processing (TPS) that provide the information backbone of ERP through relational databases, interactive planning (APS) that employ mathematical algorithms to create near-optimal plans and analytics (BI) to analyze large transactional data sets. Additionally, Waschull et al. (2019) identified further AFs required by MES: document management (PDM) and process monitoring and control (PMC). The former is enabled by PDM systems to maintain complex engineering and document data, whereas the latter describes automation systems that are deployed on the shop floor (level 2) to real-time control and monitor production parameters through measuring and manipulating certain variables. The components usually associated with PMC can be separated into hardware and software. Hardware comprises sensors and programmable logic controllers (PLCs) to generate data from production processes, whereas software components are usually known as e.g., SCADA or DCS that provide the infrastructure to gather, dispense and store such data

Tactical Level (4)

Operational Level (3)

Shop-Floor Level (2,1,0)

Months, weeks, days

(20)

Waschull et al. (2019) recently found that MES focus on the creation of digital twins through collection and reporting of data. They receive and extract data from other IS and databases, enabled by an interfaced functionality that converges technologies (e.g. TPS) that are usually considered as the backbone of other systems. In creating the digital twin for instance, MES overlap with functions of product innovation systems (i.e. PLM), precisely the management of product lifecycle data (i.e. PDM) and information stemming from engineering related systems (Waschull et al., 2018; Waschull et al., 2019). Other research as well underpins the concept of an interfaced AF. For instance, Kelley and Moss (2007) found that MES act as operational data stores (ODS) to consolidate transactional data from different sources which implies transaction

processing capabilities of MES similar to ERP. Besides their deployment to collect real-time

transactional data of products or components, MES incorporate DSS logics and functions allowing instant adjustments to schedules and dispatching plans (Koch et al., 2010) which implies interactive planning capabilities. This is consistent with MES also being positioned as a DSS covering three areas of manufacturing, namely storage and analysis of historical data from production, real-time control of open and closed loop production and lastly, forecasting and planning of future production runs (Naedele et al., 2015; Arica and Powell, 2017) which requires transaction processing, analytics and interactive planning capabilities. Kletti (2007) for instance states that ‘detailed scheduling’ is the APS function connecting MES and ERP. The same observation was made by Koch et al. (2010) who states that MES belong to the decision support infrastructure due to their capability of providing immediate and consistent feedback for manufacturing reporting and analyses. Also ERP systems possess functions that are usually perceived as functions of DSS (Holsapple and Sena, 2003). Another perspective is that MES can be turned into data warehouses by either including additional data repositories6_{or built-in}

features that provide data storage and analysis (Koch et al., 2010) which is supported by Saenz de Ugarte et al. (2009) who imply requirements for analytics and interactive planning within MES, as they must analyze shop floor data in real-time to be reactive and make immediate adjustment to schedules.

Table 4 Summary of application functionalities (AFs) relevant in MES

AFs Description Reference

Transaction

Processing MES consolidates transactional data similar to ERP; Storage of historical data Kelley and Moss (2007), Naedele et al. (2015)

Interactive Planning

MES have DSS logics for shop floor planning; Real-time shop floor control; MES must analyze shop floor data to immediately re-plan and re-schedule

Koch et al. (2010), Arica and Powell (2017), Naedele

et al. (2015), Saenz de Ugarte et al. (2009), Kletti (2007)

Analytics

Analysis of past data, prognostic of the future; Manufacturing reporting and analysis; ERP can be perceived as DSS; MES must analyze shop floor data to immediately re-plan schedules

Naedele et al. (2015), Koch

et al. (2010); Holsapple and

Sena (2003), Saenz de Ugarte et al. (2009)

Document Management Overlaps with PDM in creating digital twin Waschull et al. (2018) Process Monitoring

and Control MES interfaces shop floor systems for real-time data collection Waschull et al. (2019)

6_{Data repository refers to large database infrastructures that comprise multiple databases to collect, manage and}

(21)

2.5.3. Different Configurations of MES Functionalities

Before the introduction of standard shop floor IT support to the market, companies developed their own IS through individual programs. Originally, two approaches have emerged to support shop floor processes. The first was to customize and extend ERP functions or to add new functions to the shop floor process. The second was to implement an MES and integrate shop floor IT with the overall enterprise IT architecture (Schmidt, 2004). The first approach together with insufficient standards for different production environments, has led to IS that have historically grown to non-standardized proprietary systems (Lu et al., 2016). Because of resulting insufficiency, the second approach was widely adopted in the industry. That is, MES are implemented on the shop floor to real-time connect production and ERP based on different standards (e.g. ISA-95 and MESA). However, Schmidt et al. (2011) argue that current standards “are deeply rooted in the chemical industry” (p. 304) and “requirements on MES differ significantly depending on the industry and the type of production” (p. 305) which for instance hinders automotive companies from effectively applying these standards. This is because current standards attempt to simultaneously cover MES requirements for all industries and production environments which is why they lack required depth and level of detail for integration and interoperability in specific sectors. The following example shall provide clarity how MES function configurations can differ between dissimilar production environments.

Figure 4 Functional assignment in component manufacturing and assembly plants. Adapted from Schmidt et al. (2011, p. 308)

and notations modified according to MESA International (1997)

The above figure shows that configurations of MES functions differ significantly between two different automotive plants. The left shows the configuration of a component manufacturing plant, whereas the configuration of an assembly plant is depicted on the right. Core MES functionalities are those only executed and covered by MES. Partly MES covered indicates an interaction between the different layers involved. It becomes apparent that MES covers more functions in the component plant and that e.g., the function ‘dispatching production units’ is only covered by ERP. For this, the authors have identified a list of configuration parameters which eventually poses different requirements on the assignment of MES functions. The three most important are as follows (Schmidt et al., 2011, p. 308):

ERP Sho p-Fl oor ME S Sho p-Fl oor ME S ERP Labor

Management ManagementQuality Document Control

Operations / Detailed Scheduling

Product

Tracking ManagementProcess Performance Analysis Resource Allocation and Status Data Collection and Acquisition Maintenance Management Labor Management Document Control Quality Management Operations / Detailed Scheduling Product Tracking Process

Management Performance Analysis

Resource Allocation and Status Data Collection and Acquisition Maintenance Management

Core MES Functionalities Partly MES covered

Component Manufacturing Plant Assembly Plant

(22)

(1) Manufacturing process type: Different process types pose different requirements on the functions needed at MES. Within this parameter, the authors distinguish between process flow and task complexity. While the component manufacturing plant was classified as batch production, the assembly plant was categorized as mass production. (2) Number of production process variants: This depends on the manufacturing process

type as e.g., component plants are characterized by more complex production processes due to more complex and diverse products.

(3) Production volume: Determines the number of items produced i.e., low volume individual production or high volume series production. The production volume in assembly plants are usually higher because of less complex, more standardized and faster processes.

(4) Green versus brown field: This was mentioned as minor parameter which was not further explained. MES functions can be assigned more easily in new production environments than in plans where IT architectures have grown historically with systems covering different, but partly overlapping scope of functionalities.

The influence of these configuration parameters can be explained using the example of the ‘detailed scheduling’ function of MES. The production of automotive components is more diverse and complex than the car assembly process. In order to quantify process complexity, a maximum of three proprietary applications are used to cover core MES functions in assembly plants whereas an average of 70 proprietary systems were required for MES in component manufacturing plants. This is because standardization is more difficult to achieve in component manufacturing which traces back to more product variants, more complex products that require specialized tasks, eventually leading to the need for specialized applications in most cases. This inherent complexity, but also higher production volumes (i.e. much shorter tact times) leads to “a more disruptive manufacturing process necessitating rapid response in the production planning and control process” (Schmidt et al., 2011, p. 308). As a consequence, planning horizons for ‘detailed planning’ functions become rather short, if not even real-time. Therefore, execution is done by MES rather than ERP systems. On the contrary, ERP systems are more suitable for ‘detailed scheduling’ in assembly plants as planning horizons are rather long-term. For this reason, the function shifts from MES in component manufacturing to ERP in assembly plants (Schmidt et al., 2009). Support is given by Kletti (2007) who states that APS functions (e.g. detailed planning), “depending on the type of production, tend to be closer to ERP or to MES” (p. 29).

(23)

2.6 Manufacturing Systems and Characteristics

Depending on the type of product, “manufacturing environments vary greatly with respect to their process structure, that is, the manner in which material moves through the plant” (Hopp and Spearman, 2011, p. 9). As mentioned in paragraph 2.5.3, manufacturing process types and their individual characteristics are amongst the most influencing MES configuration parameters. Slack et al. (2010) classify five different manufacturing process types. These can range from very high volume in continuous processes to very low volume in project processes. On the other side, they can produce a high product variety in project and very low variety in continuous processes. Usually those two dimensions go together as low production volume is normally subject to higher variety of products manufactured, whereas higher volume manufacturing generates a lower variety of end products. In addition, higher volumes usually go hand in hand with smoother process flows. That is, newly introduced products in the start-up phase are typically subject to small volumes and production design optimization for which flexible jobbing processes are well suited. As products mature and volumes experience growth, ramping-up for more efficient (disconnected) batch production could be justified until high volumes allow for even greater standardization, i.e., automated and continuous flows (Hopp and Spearman, 2011). This development is abstracted in Figure 5, traversing from project to continuous processes (see Appendix A - Part A4: Description of Manufacturing Process Types).

Figure 5 Manufacturing process types imply different characteristics (Slack et al., 2010, p. 92; Hopp and Spearman, 2011) Diverse / Complex Repeated / Divided Project Processes Jobbing Processes Batch Processes Mass Processes Continuous Processes High Low Intermittent Continuous Low High

Manufacturing Process Types

Process Task Process Flow

Volume Variety

(24)

3 Methodology

This section describes the adopted research methodology and steps taken to safeguard the quality of this research. Essentially, this research investigates the impact of manufacturing characteristics and explorative factors on the configuration of MES functionalities. Furthermore, implications for application functionalities (AFs) will be determined to highlight requirements for different environments. This is abstracted in the conceptual model shown in Figure 6. The goal is to explore how different configuration are affected to eventually increase system performance. However, determination of the effect on performance is out of scope.

Figure 6 Conceptual Model

Firstly, the research design and the applied method is explained, followed by case selection criteria. Subsequently, measures to address common issues in case study research regarding generalizability, reliability and validity are provided.

3.1 Research Design

A multiple-case study with semi-structured interviews was conducted to empirically explore the allocation of MES function at three case companies. The research approach was

inductive as suggested by Robson (2002) within a qualitative research framework. This research

is conducted from a critical realism perspective, as the reality is considered a construct with real world boundaries, for which it is the researcher’s duty to ensure truthful interpretation of respondent data (Easton, 2010). Case studies are well recognized methods in information systems (IS) research and will be increasingly used in future research because IT architectures become progressively complex, dynamic and uncertain. This requires researches to conduct their studies in natural settings where control and manipulation is not feasible (Tsang, 2014). In addition, case studies are especially useful to explore a phenomenon in its natural settings and simultaneously develop relevant concepts, constructs and classifications. In particular,

exploratory case-studies allow to answer research questions of “how”, “which” and “what” by

gaining a relatively high understanding of the complexity and nature of a phenomenon. Secondly, case research can be used to research “the same issue in a variety of contexts in the same or different firms” (Karlsson, 2016, p. 167). As this research inherently requires to compare MES configurations for different manufacturing process types, a multiple-case study was chosen as method to answer the following RQs:

(1) “How do different manufacturing characteristics influence the assignment of MES functions to MES and other IS?”

(2) “What are additional (explorative) factors and how do they influence the assignment of MES functions to MES and other IS?”

(25)

(3) “How do different configurations of MES functions influence the requirement for application functionalities within MES and other IS?”

In order to answer the RQs, the research was conducted along following process path:

Figure 7 Research process path adapted from Karlsson (2016)

The process path for this research shown in Figure 7 illustrates the course of action during the case study research. Accordingly, after retrieving relevant concepts and contemporary issues in the field from literature, the research questions were derived. Subsequently, a case study and interview protocol was generated to guide the data collection. Semi-structured interviews were used as main data collection method. In addition, an expert interview was conducted to validate the assignment of AFs to the individual (sub functions) activities of the operationalized MES functions. After conducting the interviews, they were transcribed, coded and analyzed.

3.2 Unit of Analysis and Case Selection

The unit of analysis in a case research is the entity that wants to be studied and analyzed (Yin, 2014). The unit of analysis is the manufacturing characteristics and explorative factors and their influence on the distribution of MES functionalities and implications for AFs at each case company. Precisely, the unit of analysis consist of MES functions and resulting requirements for AFs based on the distribution to MES and the other IS relevant for SM.

Schmidt et al. (2011) have identified manufacturing characteristics in the automotive sector that influence the distributions of MES functions. The most important resemble the individual characteristics of the different process types as described in paragraph 2.6. Due to the assumption that distributions significantly vary between the remaining manufacturing process types, this research inherently requires companies/production systems representing those, except for project process type. The project type is considered irrelevant because the time horizons for planning and production are considered out of scope for MES. Ideally, between four and six cases are required to adequately ensure theoretical replication (Yin, 1998). One case company represents two process types due to different departments. Thus, three cases were selected to cover each type of process. Each company’s manufacturing process type was determined by measuring the different manufacturing characteristics as presented in Table 5. The characteristics refer to the production system which is (to be) controlled by an MES system or application. Thus, the respective process complexity (1 = very repeated to 5 = very complex), process flow (1 = very intermittent to 5 = very continuous), relative production volume (1 =

Company Records and Documents Field Notes Interview Transcripts Observations THEOR ETI CAL

Case Study Protocol / Interview Guide Theory / Research Question EMPIRICA L Induc tive / Expl ora tory P roc es s InductiveCoding Transcribing Identification of Links and Patterns Between-case Analysis Semi-structured Interviews

(26)

standardization (1 = low degree to 5 = high degree) were indicated by the interviewees (see appendix F, G, H: Manufacturing Characteristics for detailed analysis).

Table 5 Manufacturing parameters of case companies A-C

A B1 B2 C Process Task (x1) 3.5 1 2 4 Process Flow (x2) 1 5 3 2 Volume (y1) 1.5 3 5 3 Variety (y2) 3 3 1 5 Standardization (y3) 1 5 5 2

Based on the values presented, the manufacturing process type is determined by merging the average of both vertically and horizontally distributed values into aggregated values to eventually position the case on the process type map (see Part C1: Determination of Manufacturing Process Type). The final classification map is shown in Figure 8.

Figure 8 Classification of cases

All manufacturing process types are covered by at least one case company or production system. This is important, as the case selection criteria states to cover four manufacturing process types besides having an MES system implemented or designed.

3.3 Data Collection and Analysis

The data was collected using semi-structured interviews. Semi-structured interviews are in-depth interviews in which individual participants or even groups are asked to respond to preset and open-ended questions (Corbin et al., 2014). This kind of interview is conducted once, based on an interview protocol containing questions and topics that the researcher wants to explore. Interview guides support systematic and comprehensive exploration and ensure that interviews do not drift off the initially planned topics and concepts (DiCicco‐Bloom and Crabtree, 2006). Semi-structured interviews fall in between structured (i.e. pre-determined questions) and unstructured interviews (i.e. conversation with a particular purpose) allowing for targeted questions and exploration (Blandford, 2013). Thus, manufacturing characteristic and functional

(27)

allocations were determined in a structured way, whereas exploratory open-ended questions identified additional factors that influence the allocation of MES functions.

Participant Criteria

Semi-structured interviews were conducted with senior managers from respective industry 4.0 and innovation departments. The criteria for respondents was as follows:

- They should have experience with information systems in general and in-depth knowledge about the configuration of MES functionalities within the overall IT architecture at site

- To classify the manufacturing environment, they had to have enough knowledge to answer questions regarding characteristics (i.e. volume, variety, process complexity and -flow) of MES controlled production systems.

After the interviews were conducted and recorded, the tape or records were transcribed to ensure recall and enable follow-ups in case of missing data. In order to increase accuracy of the documentation, interview transcripts and summaries were sent to interviewees for revision (Karlsson, 2016). The software ATLAS.ti was used to analyze the qualitative data from interview transcripts. Using a software also increases the efficiency and reduces biases of information processing (Karlsson, 2016).

3.4 Case Descriptions

This section provides information about each company’s industry, products and current projects regarding definition, implementation and configuration of MES. It also provides information about experience and responsibilities of interviewees in order to increase reliability of data provided. A total of three interviews were conducted of which each took between 90 and 150 minutes. According to the information provided, all case companies have either an MES system implemented or MES requirements were clearly defined.

Table 6 Case descriptions

Case Description

A

The company operates within the global aerospace and defense industry and supplies aero structure composites to airplane manufacturers. Currently, the company carries out an innovation program towards industry 4.0 with already defined functional requirements for MES. However, the case plant does not yet have an MES system/ application up running, but uses the Ometa framework for system integration (see Part A5: Ometa Framework for more information).

The interviewee is the program leader for the company’s 4.0 initiatives and has 35 years of experience with information systems and landscapes and five years in the area of MES.

B

(28)

The interviewee has been with the company for 30 years, 15 years as production and operations leader and the last five years within industry 4.0 optimizations. As an end-user, he has 20 years of experience with MES and two years in-depth experience as department leader for innovations.

C

The company operates within the global aerospace and defense industry and produces wire harnesses for airplane manufacturers. The site has a newly implemented standard out-of-the-box MES application which is configurable via drag and drop processes.

The interviewee is the global IT and innovation manager of the business unit and has more than 30 years of IT experience in which he worked as an IT consultant for twelve years prior to his current position. He has been with the company since 2014 working on topics around MES, digitalization and industry 4.0.

3.5 Validity and Reliability

Case study research is often criticized to be biased by subjective judgements and to fail at developing sufficiently operational sets of measures. Two steps were defined to increase construct validity: (1) definition of specific concepts as the unit of analysis and (2) operationalization of MES functions to identify the functional configurations (Yin, 2014). In this regard, the unit of analysis and related concepts but also the operationalization of MES functions were defined in the literature review. Internal validity is no concern of explorative studies but is increased through mixed-methods data collection techniques (Benbasat, 1984; Karlsson, 2016) However, semi-structured interviews and documents were used to triangulate results and increase internal validity of this research. A widely spread concern of small case studies is the scientific rigor as they lack generalizability and validity (Tsang, 2014). Yin (1998) states that within multiple-case researches, the concept of theoretical replication can strengthen and broaden generalizability. It is built on the idea that selected cases cover different theoretical conditions. This however, requires assumptions that different outcomes can be expected from these varying conditions or contexts. That is, different configurations of MES functionalities and resulting AF requirements were expected from different manufacturing process types. The reliability of studies represents the degree to which a research can be repeated so that other researchers arrive at the same results and conclusion within the same settings (Karlsson, 2016). Therefore, a case study protocol (see Appendix B: Case Study Protocol) was set up to outline the steps taken including an interview protocol to safeguard the research during data collection (Yin, 2014).

3.6 Ethics

(29)

4 Within-Case Analyses

In this section, the empirical data is analyzed within each case to serve as basis for the between-case analysis. Accordingly for each within-between-case analysis, the manufacturing characteristics are explained first, followed by a description of explorative factors that influence the distribution of MES functionalities. Subsequently, the functional distribution is presented along with an analysis and implied application functionality (AF) requirements for each MES functionality. The functional distribution was generated from data collected through the operationalized MES functions and sub functions (see ‘Operationalized MES Functionalities’ in Appendix B: Case Study Protocol). Implications for AFs were determined by assigning at least one AF to each sub function and subsequent approval by an information system expert. Lastly, a small discussion of within-case results will be provided for the functional distribution and implications of AFs. The focus is on functionalities that were not allocated to MES in any case (see appendix F, G, H: Functional Distribution based on Interview) before adjustment, that is, deviate from standard and differ between cases. Accordingly, all functions apart from ‘data collection and acquisition’, ‘performance analysis’ and ‘quality management’ are analyzed in this section. The remaining three functions are discussed in the between-case analysis of the following chapter.

4.1 Results of Case A

This paragraph provides the results from case company A according to the above structure. 4.1.1. Manufacturing Characteristics

The manufacturing characteristics were investigated to determine the manufacturing process type of case companies and to provide some context of the production environment. All in support of both the within- and between-case analysis. Table 7 lists the different characteristics, their extent and a brief explanation of the context.

Table 7 Manufacturing characteristics at company A (see Appendix F: Data Summary Case A for original data)

Manufacturing

Characteristic Explanation

Process Task (Complexity)

The process complexity was found to be high as activities require lots of manual work and still some craftsmanship for which automation is difficult to achieve. Automated processes are complex too because of different heating curves being applied to production process.

High

Process Flow (Continuity)

The process flow was identified to have a very low continuity because of products and components being transported between

different workshops and areas after finishing a process. Very Low

Volume The volume is relatively low because only a few complex and _{time intensive shipsets are delivered to customers per year.} Low

Variety The variety was found to be medium as there are not that many different products. However, each customer has specific

requirements which results in a higher thus medium variety. Medium

Standardization Standardization was identified to be very low because each customer has specific requirements which results in different

(30)

4.1.2. Explorative Factors

This section provides the explorative factors that influence the distribution or configuration of MES functions. It is worth mentioning that the MES is non-standardized and uses existing systems that were integrated with an integration framework. The parameters were derived and consolidated from the transcript (see Appendix D:

Explorative Factors).

Disruptiveness and Flexibility Needs

The interviewee mentioned that shop floor disruptiveness and uncertainty would result in constant planning and re-scheduling within MES applications. Because standard MES applications cannot cope with frequent disruptions (especially external factors) and due to flexibility needs, e.g. excel and visual management (i.e. HMI) solutions are used for short-term sequencing at MES. This is because MES applications lack real-time dash-boarding and have insufficient possibilities for bi-directional human-machine interaction (HMI).

Green versus Brown Field

This factor mainly concerns the cost and benefit of legacy integration or an MES. Instead of an MES application, the company focuses on functions of ISA-95, definition of relevant components and the integration of existing legacy systems (brown field). In this regard, it was mentioned that an MES infrastructure is much easier in a green field where systems and functionalities can be designed properly from the beginning. Additionally, the concept of standard MES applications and their integration works perfectly in the green field, but is too rigid and costly for integration in the brown field. Therefore, the company sticks to integrating existing systems and functionalities.

Certification Requirements

The industry requires to closely manage qualifications and certifications of their operators because the customer requires to only have certified operators working on their products. Therefore, it is crucial to

manage employee information in ERP which can be used by MES for real-time job assignment. For quality assurance, it also requires to closely track the history of production data (i.e. process and operator related data) to be able to immediately respond to requests of relevant quality data.

Relation to Manufacturing Characteristics

As explained in paragraph 2.6, the manufacturing characteristics usually go hand in hand together which is why explorative factors cannot be related to a single characteristic only. The interviewee mentioned disruptiveness as allocation factor. It relates to complexity of ‘process tasks’ and product ‘variety’ (i.e. process specifications differ between customers which makes the process more complex) for which the company gives short-term decision making (i.e. “MES is claiming in principle

that they take care of the scheduling, the dispatching part of it. For sure, there are situation that MES can perfectly take care of it, but I think in […] situation where you have a lot of disruption, unpredictable disruptions, it is hard to get benefits out of it.”

“That is nice when

you start in a green field, but it is not nice when you start in legacy. […] And green field is very rare. And legacy is common.”

“Working in aerospace and defense requires a

Exploring the Use of MES Functionalities in Smart Manufacturing