FORMAL METHOD FOR VALIDATION OF PRODUCT DESIGN

THROUGH KNOWLEDGE MODELLING

Stilian Stanev, Jivka Ovtcharova

Institute for Information Management in Engineering, University of Karlsruhe, Adenauerring 20a, Karlsruhe, Germany

Waldemar Walla

Group Research & Advanced Engineering, Daimler AG, Ulm, Germany

Keywords: Production-oriented product validation, Knowledge Modeling, Expert systems.

Abstract: The aim of production-oriented product validation methods is to ensure the consideration of production

requirements during the product design phase and to validate the product specification against these

requirements before launching further steps. This work focuses on a formal method for product design

specification by using techniques from the knowledge modelling and management. The solution developed

within the EU project MyCar, in collaboration between the IMI institute of Karlsruhe and the Research and

Advanced Technology Group of Daimler AG, is based on the example of the body shop in the automotive

industry. A solution based on the Production-oriented Product Validation Platform gathers the relevant

production requirements and makes it available in a structured form to assess the product design with

respect to its ability to be produced on the production equipment.

1 INTRODUCTION

In the recent years the number of product variants

that are offered to the customer in automobile

industry increased enormously. At the same time the

product variety and the individualisation of the

products increase. This leads to more and more

complex products that are often facing conflicting

requirements, e.g. costs, quality or production.

A possible solution of the mentioned challenges

is an increase of the manufacturing flexibility.

Flexibility can be understood as adaptation of the

production to changing demand in term of order

variation and multiplication of different models. Yet,

other arguments, specific to each company, enforce

this trend like among others: the necessity to

optimize the use of existing resources, the limited

space allocated to production equipments, and the

reduction of the product lifecycle leadtime. In that,

flexibility is not only a matter of adaptability but

also a strategy for the reduction of production costs

by optimally using or reusing existing equipments.

In order to make it possible to produce on the

same line several different products simultaneously,

an overall concept is necessary, which links the

product development with the production planning.

During the whole product development process the

product has to be validated, if it can be

manufactured on an existing production line. For

this, a methodology supported by a software tool is

being developed by the Institute of Information

Management in Engineering of the University

Karlsruhe and the Daimler Research Group.

In this context the challenges are primarily the

identification of suitable information models for the

IT-supported integration of all the relevant

information, processes and methods and the

relationships between them.

2 RELATED WORKS

During the whole product development process

design engineers have to consider a lot of different

product requirements which can be summarised by

the term “Design for X” (Weyand, et al., 2008).

“Design for Production” takes the requirements of

the production already during the design phase into

account, in order to reduce the production cost and

development time as well as to raise the manufacture

166

Stanev S., Ovtcharova J. and Walla W. (2009).

FORMAL METHOD FOR VALIDATION OF PRODUCT DESIGN THROUGH KNOWLEDGE MODELLING.

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development, pages 166-170

DOI: 10.5220/0002293501660170

 SciTePress

quality (Pahl, et al., 2007). For checking if all

requirements are fulfilled the products have to be

validated during the whole product development

process considering the certain aspects. In this

context a X-orientated product validation must be

defined (Müller, 2008).

To support different kinds of digital validation,

the context has to be described more in detail.

Svensson (Svensson, et al., 1999) defined an

Engineering Information Management System

(EIMS) which is divided into four views: process,

information, system and organisation. All of them

are linked with each other, as shown in Figure 1.

Process

SystemOrganisation

InformationProcess

SystemOrganisation

Information

Figure 1: Engineering Information Management System,

according to (Svensson, 1999).

On the basis of the Engineering Information

Management System Burr (Burr, et al., 2007)

derived the six views of the EIMS (see Figure 2).

environment

processes

methods

information

data

Information system

organisation

Figure 2: Engineering Information Management System,

according to (Burr, et al., 2007).

According to Müller (Müller, 2007) a framework

for x-oriented product validation can be described

with the engineering information management

system showed in figure 2. This framework contains

all relevant influencing factors that are interacting

during digital validations.

3 POPV PLATFORM

In order to meet the industrial challenges in this area

a platform for production-oriented product

validation is being developed within the framework

of the EU-funded project MyCar. Figure 3 illustrates

the overall framework.

Production-oriented Product

Validation Platform

Validation

Models

Validation

Software

Validation

Methods

External applications

Validation Process

Activity Activity

Validation model

Validation Methods

Inputs

Validation goals

 Validation of

Manufacturability

 Validation of the

assembly

sequence

 Continuous

validation of

production-driven

requirements

Objectives

Product

Development

Production

Planning

Production

Figure 3: Production-oriented Product Validation Concept.

Input for the platform is taken from the new

developed product, the process that executes the

product and the respective production facility. The

core of the framework is an integrated validation

model (knowledge base), which describes the

relevant product, process and production

information by means of knowledge engineering

techniques. Based on this information model the

validation method can be instantiated as a process

and executed in order to check if a product can be

produced on a production line.

The validation process itself can be performed

either based on the information specified in the

validation model or by the user of an external

application (Müller, 2007). This is usually the case if

some facts are missing in the knowledge base or if

there is a need of complex algorithmic calculations

in order to check the design. However in both cases

the user will be lead by the validation methods since

this information is modelled into the knowledge

base.

4 VALIDATION METHOD

The developed production-oriented product analysis

method aims at giving a comprehensible evaluation

of product specifications for manufacturing

purposes. Therefore, the method developed for

production-oriented analysis is applied to evaluate

product manufacturability. In order to apply the

method, information regarding the product, the

FORMAL METHOD FOR VALIDATION OF PRODUCT DESIGN THROUGH KNOWLEDGE MODELLING

167

resources, the manufacturing process and the

relations between these entities must be provided.

After this the respective information is being

modelled and evaluated based on methods from the

knowledge engineering followed by the last step of

the method where the identification of appropriate

solution takes place.

4.1 Basic Definitions

An appropriate formalism concerning product-,

process- and resource-specifications must be

defined, whereupon the product specification can be

validated by means of the algorithmic evaluation of

the method.

For this purpose the following formal definitions

are needed:

An attribute of a product, process or resource is

a geometrical, structural, physical, chemical,

mechanical or other characteristic that can be

represented formally through mathematical

formulas.

Let P be the set of attributes, which specifies a

product. Then P is finite and there exists a set of

attributes

PP ⊆

′

, which contains the production

oriented analysis attributes of a product. The

number of attributes

′

depends on the application

domain and may vary.

The most used product attributes for production-

oriented analysis using the example of body shop in

automotive industry are the product structure,

product geometry, number of parts, type of material,

type of joining elements.

A resource can be defined as follows:

Let R be the set of attributes, which specifies a

resource. Then R is finite and there exists a set of

attributes

RR ⊆

′

, which contains the production

oriented analysis attributes of a resource. The

number of attributes

′

depends on the application

domain and may vary.

The most used resource attributes concerning a

production-oriented analysis using the example of

body shop in automotive industry are resource

geometry, resource structure, footprint,

manufacturing technology.

The formal specification of a process is given

analogously to the product/resource definitions:

Let A be the set of attributes, which specifies a

process. Then A is finite and there exists a set of

attributes

AA ⊆

′

, which contains the production

oriented analysis attributes of a process. The

number of the attributes

′

depends on the

application domain and may vary.

The most used process attributes for production-

oriented analysis using the example of body shop in

automotive industry are process structure, process

sequence, flow of material and additional process

characteristics such as cycle time, operating cost,

etc.

Let

{

}

ARPV

′

be the set of relevant product,

process and resource attributes. For all

Vv∈

the

function

)(: vdomvdom 

defines the co-domain

for every attribute. A function

)(: vdomB Bv ∈→ ∪

over a set of attributes

VB ⊆

with

)()( bdomb ∈

for all

∈

that assigns each attribute to a

feasible value

)(b

, is called Assignment of B.

In order to define compounded attributes, which

are inter-related with other attributes, a calculation

must be defined.

Let

be the set of any partially defined

functions

VVPf →)(:

of finite arity over the

power set of

, which assigns every set of

attributes to an attribute by a definite and calculable

operator. If

∈

is such an assignment and if

cBf

)(

is true for

VB ⊆

and

Vc∈

, then

called calculation model for the attribute

c over

By means of this definition the product structure

(bill of material) attribute of a product component

can be defined as a binary relation

BOM

×⊆→

, for

which

cBBOMBbVBcb BOM

∈⊆

∃

⇔

→ )(,:

is true. Thus, a directed graph can be derived. The

relation

BOM→ directly induces the graph’s edges

and the nodes are the referenced attributes, which

represent structures of other components on their

part, as well.

4.2 Modeling and Evaluation

Based on techniques from the knowledge

engineering the modelling and evaluation is

described in this paragraph.

Based on a mathematical model, the new method

should perform algorithmized evaluations of the

product specification regarding manufacturability.

For this purpose an algebraic structure over the set

of relevant attributes is required, which allows the

algorithmized performance of the method.

Let

G be the set of any calculable logical terms

of finite arity over

VB ⊆

. Then

Gg ∈

is referred

to as a condition or predict regarding

. The set of

all conditions along with the set of calculation

models build algebra of terms over a set of

attributes. If in a condition g every free attribute is

assigned to a value by

, then the condition results

in a calculable term, which can be evaluated directly

{

}

1,0: →Bg

. This logical condition is correct if

the term evaluates to 1.

In order to evaluate the product

manufacturability on a particular production

resource, it is sufficient to prove the correctness of

KEOD 2009 - International Conference on Knowledge Engineering and Ontology Development

168

the conditions over the attribute set of the respective

production process. Such conditions are

characterized as process assertions as shown in the

table below.

Table 1: Process assertions as logical requirements.

Process assertion Description

Manufacturing sequence

is correct

Is the entire manufacturing

sequence correct?

Manufacturing method

is applicable

Is the manufacturing

method applicable?

Flow of material is

consistent

Is the process’ flow of

material consistent?

Due to the complex relations and

interdependencies between relevant information

regarding the products, resources and processes

these emerge between the assertions as well. Using

the example of concrete assertions in Table 1 the

consistence of flow of material, the applicability of

joining method and the compatibility of

manufacturing sequence represent the premise for

the assertion “Manufacturing sequence is correct”.

Concerning this form of knowledge representation,

the most appropriate logical representation of

knowledge is the one given by rules and facts

(Gottlob et al., 1990). The rules are characterized by

postpre GG ThenIF

, whereby the logical premises or

preconditions are expressed by

preG

and the logical

assertions are expressed by

postG

(Puppe, 1991).

Thus, if the premises are correct, then the assertion

is correct as well. In turn, the preconditions consist

of the correlation of the rules and facts set, which are

necessary for the production oriented analysis of a

component, are characterized as validation model as

shown in the table below.

Table 2: Exemplary rules for production-oriented analysis.

Rule 1 IF Conditions to Rule 1

THEN Manufacturing sequence is

compatible

Rule 2 IF Conditions to Rule 2

THEN Manufacturing method is applicable

Rule 3 Conditions to Rule 3

Flow of material is consistent

Rule

Final

IF Manufacturing sequence is

compatible

Manufacturing method is applicable

Flow of material is consistent

THEN Manufacturing sequence is correct

For the evaluation of such rules the backward

chaining inference method is best-suited, since it

aims at proving a given assertion. If the proof cannot

be executed automatically (e.g. due to some rule

which is not included in the knowledge base) this

fact must to be queried by the user (Winston, 1993).

This feature is particularly appropriate with respect

to the integration of external tools (e.g. tolerance

analysis), which implement assertion methods of

high mathematical effort.

4.3 Identification of Solutions

By use of the presented method for production

oriented analysis it can be analyzed, whether the

product can be manufactured on a resource. If not,

the sections of the production resource are identified

in this step, which are not able to manufacture the

particular product component, due to technological

and process-related restrictions. First, the emerging

failure within specification is exactly analyzed.

Therefore, the attributes, which caused the failure

and the particular components, must be identified.

Let

VVf

⊆→

be a binary relation over the

set of attributes

V regarding a calculation model

and

cbfBbVBcb f =∈⊆

∃

⇔

→ )(,:

are

true. Then given an attribute

Vb∈ by means of the

transitive closure

→

of f→ it is possible to

find all attributes

c, which are influenced directly

or indirectly by a change of

A requirement evaluating in a wrong way,

indicates to an incorrect specification of a particular

component. The corresponding attributes are

included in the condition directly or via a calculation

model and can be derived by means of the transitive

closure

⊆→

VVf ×⊆→

Afterwards, all the conditions must be derived,

which are influenced by change of altered attributes

and thus are not potentially satisfiable. Let

∈

be a defective condition over

′

⊆

and

VVg ⊆

be the set of attributes, on which

g is directly or

indirectly dependent. Then the set of conditions

depending on attributes, which determine

, can be

derived as follows:

}:)(,{ bahdefbVaGhgG

→∈∈∃∈∪=

(1)

Thus, affected components with respect to

product and resource can be identified, which

feature a defective specification. Identifying a

solution is carried out by browsing a solution

catalogue, which includes alternative resource

configurations. After having indentified the

defective component by the means of the described

approach, action alternatives can be defined

FORMAL METHOD FOR VALIDATION OF PRODUCT DESIGN THROUGH KNOWLEDGE MODELLING

169

manually by appropriate engineers based on their

implied knowledge. Afterwards, another iteration

loop of production-oriented analysis must be

executed in order to ensure technologically

manufacturability of the particular component. This

iterative process is then executed for each action

alternatives. In doing so, multiple iterations can be

performed until a proper solution is available.

Therewith, possible resource configurations are the

result of the proposed method, which ensures the

manufacturability of a product component at an

existing production system.

4.4 Application Scenario

The following paragraph describes an application

scenario of the introduced validation method. The

main application is to support the product

development process when developing a new car

that should be manufactured on an existing

production line. In the context of this challenge the

validation method can be applied for continuous

validation of the product specification thorough the

different phases of the product lifecycle.

The core of the framework is the validation

method that is executed in the frame of a validation

procedure. In case that the validation procedure is

positive, the new developed product can be

manufactured on the existing production line. In

case the result is negative, an engineer needs to

change either the product design or the production

resources. At this the validation tool suggests the

possible action alternatives.

To react accordingly on changing requirements

from the production planning, the product

specification needs to be validated iteratively

through the whole product development process.

With the progress of the product development

process the product design gets more mature and

more information can be regarded for the validation.

In the early stages of the development process the

materials are specified. Already at this point of the

development process the engineer should check if

the product materials can be joined using the

technology on the production line.

Later in the development process the product

geometry, joints, functions are defined. These are

the base for further validations (i.e. accessibility

checks). At this point the validation can be done

either by using internal validation methods or using

external ones.

5 CONCLUSIONS

Due to an increasing customer-individualisation and

the growing complexity of the production systems

the integration of new products on existing

production facilities is getting more important. This

work presents an integrated method for validation if

a product component can be manufactured on a

given production line. This is performed by

modeling and evaluating the respective product,

resource and process informaion based on

knowledge engineering techniques. The application

of this method in the industry ensures time reduction

between introducing a new product to the market

and also creates an important improvement of the

decision security for all responsible personnel.

ACKNOWLEDGEMENTS

The authors wish to acknowledge their appreciation

to the European Commission that partially funded

the project and to all the MyCar project partners for

their valuable cooperation and contribution.

REFERENCES

Weyand, L., Bley, H., Kaiser, J.-M., Meißner, R., Bär, T.

2008. Production Oriented Design-Check in Car

Manufacturing – Not Only a Question of Powerful

Validation Tools. In: Intelligent Computation in

Manufacturing Engineering, 6th Conference, Neapel.

Pahl G, Beitz W, Feldhusen J, Grote K-H (2007):

Konstruktionslehre Grundlagen erfolgreicher

Entwicklung. Methoden und Anwendungen, Springer

Verlag. Berlin Heidelberg New York, 7

Edition.

Müller, M., 2007. Reifegradbasierte Optimierung von

Entwicklungsprozessen – am besonderen Beispiel der

produktionsbezogenen Produktabsicherung. In: Bley

H, Weber C (eds.) Schriftenreihe Produktionstechnik.

Saarland University, Saarland.

Svensson, D., Malmström, J., Pikosz, P., Malmqvist, J.: A

Framework for Modelling and Analysis of

Engineering Information Systems. Proceedings of

ASME DETC ‘99, 1999. Las Vegas, USA.

Burr, H., Müller, M., Vielhaber, M., 2007. EIMS – A

framework for engineering process analysis. In:

Proceedings of the International Conference on

Engineering Design, ICED ’07, Paris.

Winston, P. H., 1993. Artificial intelligence, Addison-

Wesley Reading. 1

Edition.

Puppe, F., 1991. Einführung in Expertensysteme, Springer

Verlag. Berlin Heidelberg, 1

Edition.

Gottlob, G., Frühwirt, T., Horn, W., 1990. Experten-

systeme, Springer Verlag. Wien, 1

Edition.

KEOD 2009 - International Conference on Knowledge Engineering and Ontology Development

170