International Conference on Computer Systems and Technologies - CompSysTech’ 2005

A Three-tier Software Architecture for

Manufacturing Activity Control in ERP Concept

Stefan Dumbrava, Doru Panescu and Mihaela Costin

Abstract: Markets globalisation makes industrial companies to change radically, becoming more
flexible, customer oriented and agile in manufacturing. Consequently, the enterprise software applications
are becoming more and more complex, exploiting distributed object technology in multi-tier architectures.
The aim of the paper is to present the design specification for an enterprise application that integrates the
ERP concept in three-tier architecture, containing modularized, distributed subsystems in configurable and
maintainable software. Starting from the customer order, to planned order dispatch, the ERP main
components and the information flow within the system is presented. Then, the software architecture is
discussed, where the ERP components are implemented using Enterprise Java Beans (EJB), deployed in a
J2EE application server, for performing transactions with business environment and the database.
Constructing the visual model in Unified Modeling Language (UML), the application is designed and
analyzed. The use case diagram of the system is presented.
Key words: Enterprise Resource Planning, Material requirements Planning, Distributed object
technology, Unified Modeling Language.

INTRODUCTION
“Whereas at one time the decisive factor of production was the land,
and later capital…..today the decisive factor is increasingly man himself, that is his knowledge”.
Pope John Paul II - 1991 Centesimus Annus

Due to increasing inter-organizational production and control, a new management

dimension occurs that focuses on both manufacturing activities and the supply chain. Its
demands are the cost control, needs to analyze costs/revenues on a product or customer
basis, flexibility to respond to changing business requirements, more informed
management decision making and changes in the way of doing business, [5], [10]. The
enterprise information systems and the business software applications have continuously
followed the new managerial concepts, evolving from systems without decision support
towards complex applications, including assistance in the decision taken process. ERP
system is an overall company concept that has capabilities for quality management, field
service, maintenance management, distribution, marketing, and supplier management.
Software solutions implementing this concept may address the enterprise needs, making
the process view of an enterprise to meet the organizational goals by integrating all
functions of an enterprise. In the same time, it may focus on the external supply chain
enabling the firm to deal directly with suppliers and to assess the availability of their
resources. Consequently, with the support of software technology, ERP becomes a
distributed application in an Internet environment, developed and able to communicate
directly with suppliers and customers, sharing e-markets in an e-business environment,
[6]. For these purposes enterprise applications typically require the ability to handle a large
number of users and large amount of data, flexibility to expand quickly, security services in
order to prevent unauthorized access, complex transaction processing and database
access, [7]. The advances in computer communication and databases together with the
emergence of new software technologies, based on scalable, interoperable and
reconfigurable information architectures enabled business applications to be designed,
build and implemented more rapidly, cheaper and with fewer resources. The multi-tiered
application model for enterprise application is a client-server architecture in which the user
interfaces, functional process logic, data storage, meets these requirements, [1]. The
three-tier architecture is distributed in three locations: client machine, server machine and
the database machine, the architecture extending the two-tier client/server applications to
three-tier applications.
1
- - IIIB.3-1
- -
 International Conference on Computer Systems and Technologies - CompSysTech’ 2005
THE ERP SYSTEM ARCHITECTURE
The basic blocks and the information flow within the system are represented in Fig. 1.

Order Sales Order Processing Accounts

(Product configuration, order entry, sales management) Invoicing Receivable
entry

Sales Material Bill of Materials

Forecast File Master General
Production Requirements
Planning Inventory Ledger
Schedule
Production Management
capacity File
Capacity
Requirements Order release Accounts
Process planning Scheduled System Payable
and production Planning ERP receipts
routing file
Vendor Communication Purchasing
(Electronic Data Interchange, System Shop Floor-Control
shipping notice, e-commerce)

Fig. 1 Information flow in an ERP architecture

The main goal of the architecture in Fig. 1 is to disaggregate the production plane, to
schedule and issue production orders for the final products together all their components,
while ensuring the feasibility of production plans and the general ledger, [6]. The input in
the system is the customer order that after it is processed in the sales and marketing
department enters the Master Production schedule (MPS). The MPS is a modelling
technique that allows for demand-driven production plans, making statements on what,
when and how much is to be produced. Based on data input (Manual forecast, system
forecast and actual customer order) the system generates the total demand for each time
periods, using the following rule: for the lead-time (emphasized in yellow) the actual
customer order is considered; outside, the manual forecast has priority to system forecast,
[5]. The initial inventory stock is also correlated to generate the MPS records: the projected
available balance and the available to promise quantities. The MRP modelling technique
uses time-phased gross requirements, scheduled receipts, inventory information and other
system parameters, for elaborating the net requirements, planned order receipts and
planned order release schedule, Fig. 2.

a) b)
Fig. 2 MPS (a) and MRP (b) matrices for a generic item with Lot for Lot Sizing Technique

Gross requirement is total actual demand for that item at that time. Scheduled receipt
indicates the expected units that will be completed at that time. Planned order release
indicates the suggested amount of units to order from the Shop-floor system. With the
2
- - IIIB.3-2
- -
 International Conference on Computer Systems and Technologies - CompSysTech’ 2005
knowledge of time-phased gross requirements, scheduled receipts and other MPS
parameters, MRP can be filled out. In order to adapt to suppliers conditions, several
techniques for lot sizing are used. Explosion of requirements in MRP programs flows down
through component levels following the linkage specified in the Bill of Materials (BOM)
structure. The planned order release of a parent item multiplied by the quantity of the child
item that goes into the making of one parent item, become the gross requirement of the
component required at the next level. The information is organised in tree type structures,
each tree being used to describe the structure of one final product. The BOM structure is
exploded many times during MRP calculation, requiring fast data retrieval. The format
used to structure its content and to represent the hierarchical structures is the Extensible
Markup Language (XML). The attributes considered for each node are name, quantity, and
lead-time. As a sample, the Fig. 3 shows hierarchy structure of BOM for a mixed type
production of final products F0 and F1. Each of them consists of assemblies and parts.
F0 F1

A0(1) P1(2) A2(2)

P0(2) P1(1) A0(1) P0(3)

P0(2) P1(1)
<BOM>
<product name="F0">
<subassembly name="A0" quantity="2" lead_time="2">
<part name="P0" quantity="2" lead_time="1"></part>
<part name="P1" quantity="1" lead_time="1"></part>
</subassembly>
<part name="A1" quantity="2" lead_time="3"></part>
</product>
<product name="F1">
<subassembly name="A2" quantity="2" lead_time="2">
<subassembly name="A0" quantity="1" lead_time="1">
<part name="P1" quantity="2" lead_time="1"></part>
<part name="P2" quantity="1" lead_time="1"></part>
</subassembly>
<part name="P0" quantity="3" lead_time="1"></part>
</subassembly>
</product>
</BOM>
Fig. 3 BOM Hierarchy Structure and the corresponding XML file

The file is plain text format and Java language is used to access the BOM file,
parsed with Document Object Model (DOM) specification, no matter what platform it
resides on.

THE THREE-TIER ARCHITECTURE OF THE ERP APPLICATION

The ERP application is designed as a web-based and server-side application that is
partitioned in terms of application logic into three-tiers, Fig. 4. Each layer has a different
responsibility in the overall deployment, and within each layer, there can be one or more
components. The layer partitioning is as follows. Presentation Tier contains components
dealing with user interfaces and user interaction. The presentation layer of the web-based
deployment could use HTML pages, Java Server Pages, and/or Java Applets. Middle Tier
is composed of the Web Tier – JSP and Business Tier. The Web-Tier is the web server
part that includes the web container and other protocols by the J2EE specifications. In the
web container, servlets, JSP pages, filters, and web event listeners execute and may
3
- - IIIB.3-3
- -
 International Conference on Computer Systems and Technologies - CompSysTech’ 2005
respond to HTTP requests from web clients. They may also be used to generate XML or
other format data that is consumed by other application components.
Presentation Tier MiddleTier
DataTier

JSP Enterprise
Web Page Servlet Bean
Browser

Web Container
Enterprise
Bean

EJB Container
Application
Client Database
Container
WebTier - JSP Business Tier

Fig. 4 The three-tier architecture of an enterprise application

The model-view-controller (MVC) paradigm was developed to map the input,

processing, and output tasks to the graphical user interaction model. The presentation
model, view, and controller exist in the Presentation services tier. The Model 2 architecture
introduces a controller servlet between the browser and the JSP pages or servlet content
being delivered. The controller centralizes the logic for dispatching requests to the next
view based on the request URL, input parameters, and application state. The controller
also handles view selection, which decouples JSP pages and servlets from one another.
The Model 2 controller servlet provides a single point of control for security and logging,
and often encapsulates incoming data into a form usable by the back-end MVC model.
The Business Tier contains Enterprise Java Beans (EJB) components that encapsulate
business logic into server side and work together to solve the business problems, [4], [8].
The ERP application is implemented as a set of business-logic-controlling EJB
components that have been configured in application-specific ways inside an EJB
container such as an application server. Clients are written to communicate with the EJB
components and handle the results. EJB covers two fundamental models for synchronous
communication in building enterprise applications: Session Beans– objects that represent
a transient conversation with a client and Entity Beans - data in a database, along with the
methods to act on that data. Session Bean can be Stateless (each call to the bean’s
methods is independent of the following call) or Stateful (a client has exclusive use of the
bean’s methods and maintains state between method calls). Another way to communicate
with EJB components is using Message-Driven Beans for sending receiving asynchronous
messages The EJB Container provider is part of the infrastructure as the runtime system
for one or multiple enterprise beans. It links different beans and EJB server. Data Tier is
represented by the database system that storages the business data and state. It is
accessible from web components, enterprise beans, and application client components
through the JDBC API. The advantage to partition the application into these logical layers
is to isolate each tier. Thus, it should be possible to change the Presentation Tier while
minimizing impacts on the business logic in the Application Tier or Data Tier.

VISUAL MODELLING IN UNIFIED MODELING LANGUAGE (UML)

Visual modelling is a communication tool that uses standard graphical notations for
specifying, visualizing, constructing and documenting the software and/or business
4
- - IIIB.3-4
- -
 International Conference on Computer Systems and Technologies - CompSysTech’ 2005
models, independently of the implementation language, [2]. UML is such a platform-
independent graphical standard used to define the enterprise architecture, the strategic
reuse, system capabilities and enterprise application integration. The model of the system
comprises several views that characterise it; use case and class diagrams (which define
the functionality and the logical view of the system), state chart and activity diagrams
(which define the behaviour of the system), sequence and collaboration diagrams (which
define the interaction within the system), component and deployment diagrams (which
define the implementation issues). It also has the following main capabilities: identify and
design business objects and then map them to software components, partition services
across a multi-tier model and/or architecture, design a distributed object architecture, code
generation directly from the model, use reverse engineering to create models from existing
components and applications, use round-trip engineering facilities in order to keep the
designs synchronized with code, [3]. The conceptual design of the application is done
using this visual modelling tool. The use case diagram for the ERP application is
represented in Fig.5.

Fig. 5 The Use case diagram of the ERP application

According to the architecture, five kinds of group users will use the system:
Customer, Administrator, Supplier, Shop-floor system, and the database. They are the
actors in the model, categorized by means of generalization relationships. Use cases can
be related by association and generalization relationships and by two stereotyped
relationships called <<include>> and <<extend>>. Actors and use cases are linked by a
special kind of association called <<communicate>> relationship. Use case for
Administrator is to manage the accounts and to supervise the system. Checked and
updated by the system, Customer group may upload orders that are processed before
issuing the invoice. It includes the firm record of the demand and the update of the
receivable account. The customer group may check the order status or modify the order.
Based on actual demand and the sales forecast the MPS is generated, taking into account
the available production capacity. The schedule for final products it is correlated with
inventory information, scheduled receipts and the exploded BOM to generate the MRP.
Using the make/buy decision associated to the part, the planned orders are directed to the
Shop-floor system or to purchasing system by Order Release System. Supplier group may
receive the purchase orders from MRP system by using web browser. Once the customers
add or modify orders, MRP system may update data automatically to rapidly respond to
change of customer requirement. MPS, MRP and Capacity Requirements Planning are
designed as Session Beans that invokes other Entity Beans for logic performing. BOM,
5
- - IIIB.3-5
- -
 International Conference on Computer Systems and Technologies - CompSysTech’ 2005
Customer Order, Process Plan and Inventory Data are designed as Entity Bean,
representing persistent data. All these Enterprise Java Beans reside in Application Server
and supply services like security, transaction management, and scalability. The complete
implementation of the application using EJB comprises: writing remote and home
interfaces, which define the methods of creating, finding and interacting with an EJB,
writing the main EJB class, writing the deployment descriptor component and compiling
the EJB classes and interfaces. The UML is extensively used through the entire
procedure.

CONCLUSIONS AND FUTURE WORK

In this paper, the specifications for an enterprise application designed to integrate an
ERP system using distributed object technology are presented. The research has been
conducted within the framework of the project entitled “The development of an Executive
Information System Portal for Manufacturing Activity Control in the ERP Concept”,
developed with the National Counsel for Scientific Research in the Higher Education. This
is part of the first year activity, in which the objectives have been to design the architecture
of the ERP application, the information flow and activity simulation. The architecture is
designed in Enterprise Java Beans specifications, as a modern solution for developing
multitier enterprise services. J2EE applications can be rapidly deployed and easily
enhanced, ensuring a flexible reaction to the business environment feedback.

REFERENCES
[1] Blaha M., W.Premerlani. Object-Oriented Modeling and Design for Database
Applications, Prentice Hall, 1997.
[2] Cheesman, J., J. Daniels. UML Components: A Simple Process for Specifying
Component-Based Software, Addison-Wesley, 2000.
[3] Quatrani, T. Visual Modeling with Rational Rose and UML - Addison-Wesley,
1998.
[4] Monson-Haefel, R. Enterprise Java Beans, Sebastopol, CA: O’Reilly &
Associates, Inc., 2000.
[5] Proud, J. P. Master Scheduling: A Practical Guide to Competitive Manufacturing,
John Wiley & Sons Inc., 1999.
[6] Rehg, J. A., H. W. Kraeber. Computer-Integrated Manufacturing, 2nd edition,
Prentice Hall, 2001.
[7] Senn, J.A. Analysis and Design of Information Systems, McGraw-Hill, 1989.
[8] Singh, I., B. Stearns, M. Johnson, and the Enterprise Team. Designing Enterprise
Applications with the J2EETM Platform, 2nd edition, Addison-Wesley, 2002.
[9] Sipper, D., B. Bulfin. Production: planning, control, and integration - New York:
The McGraw-Hill Companies, Inc., 1997.
[10] Vollman, T. E, W. L. Berry, D.C. Whybark. Manufacturing Planning and Control
Systems, 4th edition, McGraw-Hill, 1999.

ABOUT THE AUTHORS

Assoc. Prof. Stefan Dumbrava, PhD and Prof. Doru Panescu, PhD, Department of
Automatic Control and Industrial Informatics, Faculty of Automatic Control and Computer
Engineering, “Gh. Asachi” Technical University of Iasi, Blvd. Mangeron 53 A, Iasi, 700500,
Romania, Phone: +40 232 278683, Е-mail: [email protected], [email protected].

Scientific researcher Mihaela Costin, PhD, Computer Science Institute, Romanian

Academy, Iasi Branch, Blvd. Carol I 22A, Iasi, 700505, Romania, Phone: +40 232 241708,
Е-mail: [email protected].

6
- - IIIB.3-6
- -