Int J Adv Manuf Technol (2006) 30: 945–951

DOI 10.1007/s00170-005-0067-0

ORIGINA L ARTI CLE

J. Kulon . P. Broomhead . D. J. Mynors

Applying knowledge-based engineering to traditional

manufacturing design

Received: 1 September 2004 / Accepted: 3 November 2004 / Published online: 9 December 2005
# Springer-Verlag London Limited 2005

Abstract The paper outlines the development of a intellectual capital and create time to innovate and assess
knowledge-based engineering (KBE) system for hot forg- more advanced, complex types of tools to eliminate un-
ing design using state–of–the–art technology and the In- certainty and repetition from the design process. With an
ternet. The benefits of a KBE approach over a traditional ageing workforce it is vital that experience is not forgotten
design process are emphasized. The aim of the proposed or allowed to drift away through retirements etc., but is
KBE system is to integrate the hot forging design process captured and built upon within a company. More manu-
into a single framework for capturing knowledge and ex- facturers are now exploring new ways to harness the ad-
perience of design engineers. The KBE application guides vantages of new technologies and communications tools
the design engineer through different stages of the design such as Internet. Knowledge-based engineering (KBE)
process enabling the generation of forgeable geometry systems, which complement traditional CAD/CAM sys-
from a component profile taking into account machine, tems by adding the engineering knowledge that drives the
material and forging company specific data, and design product design process, are becoming an essential part of
considerations. strategy for improving effectiveness [1, 2].
This paper outlines the development of a KBE system
for hot forging design using state–of–the–art KBE tech-
1 Introduction nology and the Internet. Currently most of the activities and
costs related to planning forging processes depend strongly
Today’s economy presents a new set of challenges for the on human expertise, creativity and intuition as well as time-
manufacturing industry. In the UK and other developed consuming experimental, trial and error, work. The nu-
countries the manufacturing sector is under siege because merous variables required to describe forging processes
developing countries offer quality products at lower prices. make the knowledge capture and the encapsulation of the
Global competition has empowered customers who now decision-making and process planning rules difficult using
constantly demand new and better quality products. To conventional methods [3–7]. The aim of the proposed KBE
meet the challenge of foreign competition and survive in system is to integrate the whole design process into a single
the market, manufacturers in general and the metalforming computer model and to create a framework for efficient
industry in particular are now under tremendous pressure to capturing of a company’s most important asset—the knowl-
improve their efficiency in terms of product development edge and experience of its engineers. In the course of the
and resource utilisation. This means a dramatic reduction in design the KBE application guides the design engineer
the overhead cost of redundant engineering work, elim- through different stages of the hot forging design process
inating repetitive design tasks, reallocating technical re- enabling the generation of forgeable geometry from a com-
sources to new product development. To stay competitive ponent profile. It also supports computation of the forging
in today’s global economy a company must build on its load, complexity factor, component volume, flashland ra-
tios etc. A relational database is used to encapsulate the
design rules as well as their complex interdependencies on
material, production unit, and manufacturer capabilities.
As part of the system a cost-estimating procedure for hot-
J. Kulon . P. Broomhead . D. J. Mynors (*) forged parts is being developed, taking into account the
Design and Systems Engineering Department,
Brunel University, size of production volume, the production unit, the pro-
UB8 3PH Middlesex, UK duction time and the dimensions of the billet and the com-
e-mail: [email protected] ponent to determine the total forging cost.
946

2 Traditional organisation of hot forging design process requiring the highest level of expertise. According to some
estimates approximately 80% of the final product cost is
As illustrated in Fig. 1 the traditional organisation of hot decided at the design stage. Finally after all the necessary
forging design and process planning contains several stages tools are created, the design engineer generates drawings
starting with the component design and concluding with and reports that document the forging design. Currently in
the cost estimation and the process parameter selection the forging industry the design process is still mainly based
and optimisation. The component design stage involves the on very inefficient iterative ‘trial and error’ procedures.
transformation of the component geometry to obtain forge- After taking successive measurements of the forgings,
able shape. The design tasks at this stage involve the de- gradual corrections to the die geometries are made, which
termination of a parting line, specification of appropriate eventually lead to the production of so called ‘good parts’.
radii, fillets and corners, internal and external drafts, design Since many repetitive engineering tasks are not automated,
of holes, ribs, and webs. The empirical design guidelines, they need to be done by hand, which creates a bottleneck
which are used, often depend on the material, production that slows down the design cycle. As shown in Fig. 1 this
unit and manufacture’s experience. The creation of a forge- design process is really a loop, which is repeated until a
able shape is followed by the determination of the process satisfactory design is found, therefore the effect of the
sequence and designing all the tools needed during the bottleneck is multiplied by the number of times the loop is
production process, including those used for die profiling repeated.
and trimming. The most critical and expensive tools are the In order to speed up the design and optimise the process
forging dies, of which, for a particular forging, there may be sequence more often software tools such as CAD packages
several preform dies. The main objectives of the design of [8] and numerical methods like the finite element method
preform dies are to ensure a good distribution of material in (FEM) are used. Many studies have demonstrated that nu-
order to fill the finishing dies without any defects, minimise merical analysis can help the forging industry to define the
metal losses and minimise die wear, i.e. minimise metal best process sequence, reducing time and costs involved
flow in the last die so as to extend its life. The perform [9–11]. Typically, CAD software is used to display, manip-
design stage is generally considered to be the most creative, ulate, and transform the component geometry, which

Fig. 1 A traditional, ‘manual’ organisation of forging design process

 947

subsequently is exported to FEA packages to simulate the designs are created or engineering analyses are under-
forging process. Although traditional CAD systems allow taken” [12]. Unlike the traditional CAD systems, which
engineers to process designs faster than when designs were mostly contain geometric information for a single design,
done completely by hand, they go only part of the way KBE does not just create a single design; it models the
toward providing a completely flexible tool for adapting to whole process of generating the design. A KBE system
changes in a product design and provide no means of stores knowledge about a product in a comprehensive
capturing the engineering expertise behind the design. product model composed of engineering rules that describe
While the geometric information can be handled using how products are designed, analysed and manufactured.
CAD systems other design knowledge, including cata- The rules can be design rules, standard engineering rules,
logues, databases, and engineering standards, must be man- or experiential ‘rules of thumb’, which reflect years of
aged separately without integrated tools. In addition, if the design experience. They can be expressed in the form of
input specifications change at any time, the entire process physical equations, graphical and tabular relationships de-
must be repeated. Even with all of the engineering team’s scribing the attributes of the physical product such as ge-
experience, the process of developing a single design is still ometry, functional constraints material type etc. [13]. Once
laborious, repetitive, and subject to error. The following engineering knowledge about the product is collected and
sections will show how this situation can be significantly stored as a product model, design engineers can generate
improved using KBE environment. and evaluate new designs quickly and easily by changing
the input specifications for the product model, or modify
designs by extending or changing the product model. This
3 KBE approach frees the engineer from repetitive time-intensive and often
tedious design tasks such as repetitive calculations and al-
3.1 KBE basic principles and benefits lows more time for creative design work. Another strategic
benefit of KBE is that the product model also includes
The conventional definition of KBE is that it is “the process various outputs such as reports of engineering results, data
of combining engineering knowledge, methodologies, of engineering analysis, 3D geometric models as well
rules and best practices with process knowledge and best as non-geometric output such as bill of materials, costs
practice to create product models that describe how product and mass estimations and manufacturing instructions [14].

Fig. 2 KBE model of hot forging process

948

When the design changes, so do the outputs. Information is but also the solutions of a family of previously solved
always up–to–date and reflects the current state of the cases. The redesigned preform may be again validated by
design, automatically. A KBE system is also able to in- FEA simulation. Finally, the KBE generates automatically
tegrate other engineering programs such as FEA software in the document set for the forged part, e.g. the drawings,
a wider organised system working as a kind of pre-pro- reports etc. The design engineer can easily modify the
cessor for the analysis applications. design either by changing the answers to some of the key
design questions or by modifying the input specifications.
In this scenario, the bottleneck of the traditional design
3.2 Proposed KBE model of hot forging process, the repetitive tasks that had to be done by an
engineer, is eliminated. All such tasks are managed by the
The diagram shown in Fig. 2 illustrates a proposed KBE product model. The design loop can be repeated efficiently
model of the hot forging design process. The centre of the as many times as necessary without incurring the expensive
design process is the product model, which manages all the trial–and–error procedure and experimental work. With the
routine engineering tasks. The role of the engineer who product model, generating a new design can be done much
interacts with the model is to provide the input specifica- faster in a matter of hours by a single design engineer, as
tions such as component geometry and make the important opposed to several days or weeks with traditional methods.
design decisions. Databases of materials, standards books,
production units’ specifications, design rules encoded in
the system by the forging engineer etc., are all integrated 4 KBE technology
into the product model through the relational database.
First the KBE application automatically constructs forging 4.1 KBE server architecture
geometry from component geometry using design speci-
fications. Once a design has been generated from the pro- Traditionally, all computer aided design tools were de-
duct model, the shape can be exported to a FEA for a signed as interactive applications for a single user. How-
forging simulation allowing the forging designer to test ever, with the fast development of the Internet and database
several process parameters in order to improve the forging technologies, it is advantageous to build an integrated sys-
process. According to the simulation results, intermediate tem based on the web in order to better implement knowl-
forgings (preforms) may appear to be necessary. At any edge share and to distribute designs [15]. The web-based
stage the geometric information can be transferred back to environment offers several advantages in comparison to
a CAD tool for further detailing/modifications. A repos- more traditional software platforms, such as: an open ar-
itory of the previous designs can be used to retrieve the chitecture, uniform information model, and object-oriented
relevant cases and exploit analogies and similarities with structure. All of these can help to eliminate the time and
previously solved cases. The designer of preform shapes space restriction of engineering design and realise maxi-
can use not only the information from the simulation results mum knowledge share. The web environment also pro-

Fig. 3 KBE server architecture

 949

vides the user interface, file transfer and can display graph- lows a non-procedural paradigm for application develop-
ics representing the geometry through any standard web ment and execution, with multiple threads of execution and
browser. the ability to load new code and redefine objects dynam-
The proposed web-based KBE system architecture is ically. The server side framework is implemented using
shown in Fig. 3. The left side of the diagram in Fig. 3 GWL as an integrated web server-based user interface that
indicates user interface, the right side corresponds to the manages data and controls other software tools on the
processes run and the steps carried out on the KBE ‘server’ server. The main application tasks carried out on the server
machine. The core of this KBE application is the GDL/ are: accessing the database of rules, design features and
GWL (General–purpose Declarative Language/Generative materials, facilitating file transfer between the user and
Web Language) developed by Genworks International CAD package or a FEA software and performing the KBE
[16]. GDL/GWL is a knowledge base programming en- computations. GDL/GWL integrates the application logic
vironment, which implements the features of caching and and databases through a direct MySQL interface and han-
dependency tracking as well as demand–driven evaluation. dles geometry, through its library of 3D geometric objects.
Further, because GDL is based on a full object-oriented It is able to export and import component surfaces directly
language model, a superset of ANSI Common Lisp, it fol- embedded in IGES format. The upcoming integration to

Fig. 4 Integrated database of materials and design rules

950

NURBS surface and solid modelling libraries will add new 4.2 Integrated database of materials and design rules
powerful surface and solids capabilities to GDL’s built–in
3D wireframe facilities [16]. The schematic diagram of the integrated database system is
shown in Fig. 4. The database of the KBE application is

Fig. 5 User interface

 951

designed as an independent separate part of the whole presses, screw presses, hammers etc.) and component
system containing the design rules, production units’ geometry (axisymmetric, plain strain and 3D). The work is
specifications and material properties. ongoing aiming at further implementation of design rules,
Knowledge (design rules) has been obtained from hand- transfer of simulation results as well as the storage and
books, plasticity theories, relevant references and empirical retrieval of a large number of case studies.
know–how of field experts in hot forging companies. The
database also takes into account the dependency of the Acknowledgements The authors would like to thank their
design rules on materials, production units (such as a industrial collaborators on this project and especially Mike Twelves
mechanical press, screw press, hammer etc.) and compo- of Corus Automotive, Neville Turner of George Dyke Ltd. and John
nent geometry (such as axisymmetric, plain strain, 3D etc.). Tildesley of W.H. Tildesley Ltd. as well as the United Kingdom’s
Engineering and Physical Sciences Research Council: Grant: GR/
Rules are arranged in the form of tables or equations. N21611.

4.3 User interface and visualisation References

The browser-based user interface for accessing the design 1. Mynors DJ, Griffiths BJ, Twelves M, Etheridge D, Moore J,
knowledge is shown in Fig. 5. The tree menu on the left Griffiths T, Piddington C (2004) Final report of the tooling and
allows the selection of several expandable submenus, experiential design initiative (TEDI). (ICT Scoping Study
which in turn update the main viewing window to guide the 0116). March 2004. Obtainable from either the UK Department
of Trade and Industry or the Department of Design and Systems
user interactively through the design process. Engineering, Brunel University, Uxbridge, Middlesex, UB8
The same can be achieved by using the block diagram, 3PH, UK
shown in the main window in Fig. 5. Various boxes re- 2. Cooper S, Fan IS, Li G (1998) Achieving competitive advantage
present different design decisions and guidelines, which are through knowledge based engineering—a best practice guide.
Department of Trade and Industry
being implemented in the system, such as material se- 3. Tisza M (1995) Expert systems for metal forming. J Mater
lection, determination of component geometry, parting Process Technol 53:423–432
line, specification of appropriate radii, internal and external 4. Kim DY, Park JJ (2000) Development of an expert system for
drafts, design of holes, ribs, webs etc. Most of the web the process design of axisymmetric hot steel forging. J Mater
Process Technol 101:223–230
pages consist of information entered directly using stan- 5. Zhang X, Peng Y, Ruan X (2004) A web-based cold forging
dard HTML form objects. Visualisation of the component process generation system. J Mater Process Technol 145:1–6
geometry on the client’s machine is achieved using the 6. Cheok BT, Foong KY (1994) Some aspects of a knowledge-
Virtual Reality Markup Language (VRML) [17]. The based approach for automating progressive die design. Comput
VRML is currently the best way to create interactive, 3D Ind 24(1):81–96
7. Caporalli A, Gileno LA, Button ST (1998) Expert system for
environments on the WWW. VRML is generally viewed hot forging design. J Mater Process Technol 80–81:131–135
within a web browser like Netscape Navigator or Microsoft 8. Shin Y, Han S, Bae D (2000) Integration of heterogeneous
Internet Explorer. These browsers do not support VRML CAD databases using STEP and the internet. Decis Support
natively, therefore a plug-in such as Cosmo Player is Syst 28:365–379
9. Mynors DJ, Bramley AN (1997) The consequences of
required to view VRML content. incorporating computer simulation within the die design
procedure for closed die forging. In: Proceedings of the
International Conference on Design and Production of Dies and
5 Summary and conclusions Molds, Istanbul, Turkey, pp 277–282
10. Mynors DJ, Snape RG, Bramley AN (1996) A proposal for
implementing complex simulation techniques into the forging
This paper presented the preliminary stage of the develop- industry. In: Proceedings of the 12th NCMR Conference,
ment of a KBE system for hot forging design using state– University of Bath, September 1996
of–the–art KBE technology and the Internet. Instead of 11. Bramley AN (1987) Computer aided forging design. Ann CIRP
trying to fully automate the forging design the KBE leads 36:135–138
12. Javed Y (2004) Knowledge based engineering. MSc Project
designers through the design process and enables appro- Dissertation, Systems Eng. Dept, Brunel University, to be
priate manufacturing rules as well as company capabilities submitted in 2004
to be complied with. The key feature of KBE application is 13. Chapman CB, Pinfold M (1998) Design engineering—a need to
that it integrates the whole design process within one rethink the solution using knowledge based engineering. Knowl
Based Syst 11(5–6):257–267
computer model. The relational database is used to en- 14. Knowledge Technologies International (2000) ICAD (WWW
capsulate the design rules as well as their complex inter- Document) http://www/ktiworld.com
dependencies on material, production unit, and manufacturer 15. Zeng J, Chen W, Ding Q (2003) A web-based CAD system.
capabilities. Design rules have been obtained from hand- J Mater Process Technol 139:229–232
16. Genworks International: http://www.franz.com/success/custumer_
books, plasticity theories, relevant references and empirical apps/knowledge_mgmt/genworks.1html
know–how of field experts in hot forging companies. The 17. Wagner R, Castanotto G, Goldberg K (1997) FixtureNet:
KBE also takes into account the dependency of the design interactive computer aided design via the WWW. Int J Hum
rules on materials, production units (such as mechanical Comput Stud 46:773–788