International journal of computer integrated manufacturing , tập 24, số 2, 2011

Keywords: strategic partnering; supply chain coordination; information sharing; system dynamics 1.. Nowadays, companies are looking to apply enterprise resource planning ERP systems and

An approach to optimise an avatar trajectory in a virtual workplace

Mahmoud Shahrokhia*, Alain Bernardband Georges Fadelc

a

Engineering Department, University of Kurdistan, Sanandaj, Iran;bIRCCyN Laboratory, Ecole Centrale de Nantes, Nantes,

France;cCEDAR Group, Clemson University, Clemson, USA(Received 23 April 2010; final version received 1 October 2010)

A dynamic programming approach is developed to find the optimal operator walking trajectory in simulated riskyand encumbered industrial workplaces The risk for the trajectories is evaluated by calculating the area of the crosssection of the fuzzy danger zones in the operator’s path The workplace is discretised into rectangular cells, and theoptimal trajectory is approximated by a polygon, connecting the centres of ordered cells This method is adapted forboth determining the optimal operator paths and simulating the operator behaviour in the virtual engineeringplatforms The results are useful to develop emergency safety procedures, risk concept training and to evaluate safety

in the workplace

Keywords: simulation; risk analysis; virtual reality; path planning; fuzzy logic

1 Introduction

Path planning is the process of choosing a trajectory to

move from an initial position to a target point (Meyer

and Filliat 2003) One special case of this type of

problems concerns the optimisation of a human’s

trajectory in a flat workplace when simulating motion

in both computerised three dimensional and virtual

engineering platforms This problem arises during the

analysis of the performance and risk in industrial

settings, when an estimation of the work time and the

safety index for human operations are required It may

be during normal operation or after an accident or in

improvised situations, when new risks are raised and

new dangers appear In these critical moments, the

objective is finding the most efficient way to implement

preventive measures Also this problem may be

discussed, during repair and inspection and

mainte-nance operations, when barriers are removed and

operators are obligated to work with dangerous

machines and materials

In this paper, a dynamic programming approach is

proposed to find an optimal operator walking

trajec-tory in risky and encumbered industrial workplaces

The approach illustrates how a trajectory in a flat

workplace is planned for an operator from a starting

point to a destination point, around obstacles and

dangerous objects

The top view of the workplace, as a 2D search

space, is discretised into rectangular cells The

opera-tor’s trajectories are approximated by polygons,

constructed by connecting the centres of ordered cells.The objective is to find the trajectories with minimumcost-risk index (CR) as a sequence of straight pathsbetween the initial and the destination points

This method is adapted for finding the optimaloperator walking path in virtual engineering platformsand can be used to develop emergency safety proce-dures, risk concept training and evaluate the safety inthe workplace It can help to improve the layout andconfiguration of the manufacturing systems in thedesign phases, by evaluating different layouts andconfigurations and measure the effect of barriers on thehuman safety and performance The proposed ap-proach can also be used to simulate human perceptionand behaviour in the industrial workplace and haspotential to conduct artificial avatars in gamingsoftware, in a natural way

2 Related worksPath planning problems are mainly discussed in theaerospace, vehicle driving, robotics and computernetworking contexts This process includes searchingfor the shortest path or finding the trajectory withminimal displacement time, energy, cost and risk in anencumbered environment It is still one of the openproblems in the field of autonomous systems, whichinvolves meeting operational requirements and safety(Shanmugavel et al 2009) by searching in large or

*Corresponding author Email: Shahrokhi292@yahoo.com

Vol 24, No 2, February 2011, 95–105

ISSN 0951-192X print/ISSN 1362-3052 online

Ó 2011 Taylor & Francis

DOI: 10.1080/0951192X.2010.531290

infinite response space and avoiding local minima and

oscillatory movements

Many versions of the path-planning problem exist

An exhaustive classification of these problems and the

methods developed to solve them can be found in

Hwang (1992), Meyer and Filliat (2003), LaValle

(2006) Multi-agent modelling, dynamic programming,

artificial intelligence, graph-based searching (e.g

Dijk-stra’s algorithm) and analytical methods (e.g the

artificial potential field approach) are among the

most well known and used approaches They are

applied in discretised or continuous search spaces

Many of the human trajectory optimisation studies

concern the detail analysis of the movements and

postures of human body parts (see Campos and

Calado 2009), by considering the ergonomic, kinematic

and biomechanical parameters They attempt to model

and optimise the human performance and better

simulate his/her behaviours

Chedmail et al (2003) present an example of path

planning in an industrial context, by planning the

walking path for a manikin in a virtual workplace The

approach uses a multi-agent technique to optimise the

operator access and visibility, taking into account the

ergonomic constraints Using neural networks,

Bend-ahan and Gorce (2004) developed an adaptive model

to determine the motion path of the arm of an operator

to reach and grasp a prototype object, avoiding an

obstacle As a motion prediction model for obstacle

avoidance, Yang et al (2006) developed a geodesic

model to define the minimum distance curve between

two points on an obstacle surface Melchior et al

(2003) described the A* algorithm and the

fast-marching method which use a modified gradient

descent to obtain the minimum trajectory length

Kinematic models have been developed previously

to simulate the most natural human movement among

different possible solutions (e.g., analytical methods

and neural networks); however, many of them have

not yet been applied in design platforms In spite of the

fact that kinematic models for walking have been

developed and are successfully applied in design

applications, the designer is obliged to define a walking

trajectory to simulate a natural human operation

3 Methodology

This paper proposes a method to find the human

walking trajectory in flat workplaces that minimises

displacement risk and cost In contrast to the previous

path mapping methods, it presents dangers and barrier

effects by fuzzy sets In this way the supposed avatar

compares the total cost of the possible trajectories,

including his movement cost and the risk cost The

term ‘movement cost’ is used as a representative of

aggregated energy, fatigue and other undesirabledisplacement factors of the path, and the risk cost iscalculated by considering the intersection of the avatarpath with the defined fuzzy danger zones Also, theproposed approach introduces a backward check tofind the complex trajectories and several improvementrules, to apply the dynamic programming solution in amore effective way

Figure 1(a) exemplifies a computerised 3D model of

an industrial plant Three-dimensional and virtualplatforms are used during the design and analysis ofindustrial systems to evaluate various alternatives andpropose improvements Simulating human operation

in these platforms helps to better take into accountergonomic and safety considerations The problemunder study in this paper is finding the shortest safetrajectory from a start point to a destination point, in asimulated workplace In contrast to robots, humanbehaviour has a great inherent tolerance, and there-fore, the approximation of the person’s continuousresponse space trajectory by a discrete space isjustified

The notion of safety for a trajectory originatedfrom military requirements (Oustaloup and Linare`s1996) and is measured by the ability of the path toavoid threats and obstacles (Shanmugavel et al 2009)

In this paper, fuzzy danger zones, already proposed byShahrokhi and Bernard (2004, 2009), are used tosimulate danger in the 3D and the virtual workplace.This definition is derived from the concept of fuzzysets, which permits the gradual assessment of themembership of elements in a set In this way, thedistribution of danger in the workplace is expressed by

a spatial fuzzy set

A fuzzy danger zone can be used in the followingcases:

An accident that can occur due to the directcontact of a human with a danger source (e.g., asharp blade); however, the position of the dangersource is not exactly known and we use member-ship degree to represent our opinion about itsapproximate place

An accident that can occur due to the directcontact of a human with a danger source;however, the human movement is not preciseand may contact with it involuntarily We usemembership degree around the danger source toexplain our opinion about the danger thatmenaces a human in the neighbourhood of adanger source

An accident can occur due to the direct contact

of a human with a danger source; however, thedanger source is not fixed, and we use amembership degree to explain the proportion of

times that danger source is present in a point, as

an indicator for danger level of the point

An accident can occur due to the presence of a

human nearby a danger source (e.g., dangerous

sparks, slung articles or gases that are produced

around the manufacturing machines); however,

the danger reduces as the distance of the human

from the danger source increases We use a

membership degree to explain how the amplitude

of danger changes For example, radiation

amplitude is reduced proportionally to the

square of the distance from a radiation source

Then, that danger can be modelled by a fuzzy

danger zone with a parabolic membership

function

Our objective is to simulate the perception of a

human about danger in the workplace, and we

ask him to define dangers as a degree of

membership of points in danger zones

In conclusion, fuzzy sets can be used to present our

uncertain and approximate information or can be used

to explain the physical rules about the distribution of

danger space In any case, if there are several dangers

in a specific point, their dangerous effects may be

accumulated This is modelled by using the fuzzy union

operation because it is not strictly defined and is

adequately flexible to explain the various ways that the

effects of dangers may be accumulated

Similarly, the protection effects of barriers

aregeo-graphically modelled using the fuzzy memberships,

and the simultaneous effect of several barriers can be

calculated by using the suitable fuzzy union operation

The protection effect of barrier h against danger g is

modelled by fuzzy barriers zone ~B It is characterised

by function mBhg~ ðxÞ which illustrates the membershipdegree of point x in ~Bhg Danger will remain, if thebarriers are not sufficient Therefore, the remainingdanger for a specific point depends on the intersection

of the initial danger and insufficiency of union ofbarriers, in that point This makes formula (1), fuzzycomplement (Ø) for calculate insufficiency of the union

of the barriers:

~

Zrg¼ ~Zg

\ :[k h¼1

Kbarriers, on the danger zone ~Zg It is the percentage

of effectiveness of barrier h to reduce the dangerouseffect of danger g in point x According to DeMorgan’s laws (Keef and Guichard 2010) it can berewritten as:

of the barriers

Figure 2 illustrates a schematic 1D fuzzy dangerzone ð ~ZgÞ and the manner how barrier h reduces itsamplitude

Figure 1 Computerised 3D model of an industrial system

In addition, the accumulation of several danger

zones is calculated by using their bounded sum as

mZ~ r

ð5Þ

where ~Zr

T is the total remained danger zone and m~r

the membership degree of point x in fuzzy danger

zones ~Zg, after applying the barriers effects,

respec-tively Other kinds of s-norms can be used according

the accumulation effects of the dangers

A 2D fuzzy danger zone is characterised by a

membership function that assigns a membership in

[0,1] scale, to each point in its domain, indicating the

amplitude of danger, as follows:

mZ~:ðx; yÞ ! ½0; 1: ð6Þ

The red profiles, in Figure 1(b), illustrate the fuzzydanger zones around the dangerous facilities in theexemplified plant of Figure 1(a)

Analytical methods are not suitable to analyse thenon-linear, non-convex and non-continuous fuzzydanger zones that result from the accumulation ofseveral danger sources and barriers

In addition, the optimal trajectory in an cumbered workplace may be complicated [see Figure3(a)], and as illustrated in Figure 3(b), in the generalcase, a simple mathematical function that associates

en-a unique output ven-alue to een-ach input ven-alue en-as

f : x! y; ðx; y 2 RÞ cannot define these trajectories.Therefore, an analytical model should include analgorithm to combine several functions for modellingcomplicated trajectories by several different trajectorypartitions

Dynamic programming is an efficient approachwhich solves complicated path planning problems byreducing their computational complexity (Meyer andFilliat 2003, LaValle 2006) In this paper, a backwardcheck algorithm is developed and is used to examine allpossible solutions and find an optimal solution

In the example, presented in Figure 1, an operatorshould find the shortest trajectory to arrive to theelectric power transformer from his original position,

by avoiding the dangers characterised by the red fuzzydanger zones

Figure 4 presents the flow chart of the proposedalgorithm to solve this problem In the first step of thisalgorithm, the obstacles, danger zones and the startand destination points are specified To use thedynamic programming technique, the top view of theworkplace is modelled as a 2D Cartesian plan,including the obstacles, danger zones and the operatorstart and destination points (Figure 5(a)) The effect ofbarriers on dangers is modelled by eliminating fuzzydanger zones or reducing their membership amplitude.Figure 5(b) illustrates that the protective walls cut theFigure 2 Effect of a barrier on a fuzzy danger zone

Figure 3 A complex trajectory (a), and a complex trajectory in Cartesian space (b), through the obstacles and around dangerzones

fuzzy danger zones, signifying the elimination of the

dangerous effects in the protected zones

In the second step of the algorithm, an outline is

defined for a virtual universe (VU) as geographical

limits of the search space Several hypotheses are

developed and applied to define an ‘as small as possible

area’ for enveloping all reasonable trajectories By

using a simpler example, illustrated in Figure 6(a),

these hypotheses are explained as follows:

ðlS þ ð1  lÞxÞ 2 VU 8x 2 A; l 2 ½0; 1; ð7Þ

ðlD þ ð1  lÞxÞ 2 VU 8x 2 A; l 2 ½0; 1: ð8Þ

The maximum search space, defined as VU is thesmallest area that satisfies the following two conditionsfor any arbitrary point x in the set A:

Set A is the set of all points representing obstacles anddanger zones plus the required pads for the operator tosurround them, and S and D are the starting anddestination points, respectively [Figure 6(a)] Figure6(b) illustrates the search space including the VUI andVUII for this example

VU is the union of two convex areas, called VUIand VUII that are separated by a line connecting thestart point to the destination point Figure 7 illustratesthe search space including the VUI and VUII for theFigure 4 Flow chart of proposed dynamic programming approach

example presented in Figure 6 In the absence of

obstacles and danger zones, a straight line is always

selected to connect the trajectories’ partitions The

virtual universe can be reduced if two straight

trajectories can be found from the start to the

destination points that cross in the search space before

contacting any other objects For example, in Figure

8(a) the two trajectories T1and T2are connected at the

point P1, before contacting any other object

There-fore, danger zones Z1and Z2and the obstacle O1are

excluded from the search space and the final VU

becomes as presented in Figure 8(b) The

obstacle areas and the zones with an unacceptable

risk level are excluded from the VU The unacceptable

risk zones are regions that are too dangerous for

human presence

Figure 9 illustrates the resulting VU, for theexample presented in Figure 6, after applying thesehypotheses The obstacle areas and the zones withunacceptable risk level are marked by a tiles orcheckerboards in Figure 9 In the next step of thealgorithm, a baseline is drawn for use as a reference forgirding the virtual universe It is a straight line, whichisolates an as large as possible empty space in front ofthe start point by separating it from all other objects.All trajectories in this empty space are straight,because there is no obstacle or danger zone in thisspace Therefore, the dynamic programming approach

is applied in the rest of the VU by considering thesquare cells only (Figure 9) The next step of thealgorithm is concerned with identifying the processingorder of the cells For simplicity the cells are drawn as

Figure 5 Top view of obstacles and danger zones in the exemplified workplace (a), and the effect of barriers on the fuzzy dangerzones (b)

Figure 6 An example of obstacles and danger zones (a), and the related surrounding set A (b)

unit squares Generally, the size of the cells determines

the accuracy of the results As these results identify a

human walking trajectory, a high resolution approach

is an overkill and too cumbersome to justify The cellsentirely hidden by the obstacles and unacceptable riskzones are eliminated To address the cells, integernumbers (i.e., 1,2, , n) are assigned to them, accord-ing to the processing order illustrated in Figure 9

In the fifth step of the algorithm, the From-To chart

is developed It is a table that includes risk-cost indices(RCs) for paths between neighbouring cells Morediscussion related to neighbouring cells is in step 9 ofthe algorithm

The sub-trajectory for motion from the centre ofcell i(xi, yi) to the centre of cell j(xj, yj) is approximated

by line l characterised by the following equation:

y¼yj yi

xj xi

ðx  xiÞ þ yi minðxi; xjÞ  x  maxðxi; xjÞ:

ð9ÞFigure 7 An example of obstacles and danger zones, and

the related largest virtual universe

Figure 8 Contact of two straight trajectories (a), and the final virtual universe outline (b)

Figure 9 Dividing the VU to the cells by using the baseline as a reference guide

The RC index for this displacement, (RCij), is

calculated by accumulating risk and cost on the line

displacement cost (e.g., energy, time, length) factors,

respectively The symbol m indicates the membership of

specific points in the danger zones As previously

explained, danger zones are constructed by considering

the distribution of hazards and effects of the mitigating

barriers on them (see Figure 5(b)) The risk of a

straight path is estimated by calculating the integral

presented in equation (10) This integral is equivalent

to the area under the fuzzy section of the danger zone

described earlier, traversed by the operator during the

trajectory from the cell i to cell j (Figure 10(a)) It is

easily evaluated using conventional geometric tools

provided in most CAD applications (Figure 10(b))

The state variable RCi represents the minimum

value of calculated RC indices for displacement from

the starting points to the cell i The RC index is

calculated and compared only for the possible

trajec-tories that do not cross any obstacle or interdiction

zone in the path As the paths from the starting point

to the cells of the first column are straight lines in the

empty zone, where there are no obstacles or dangers, at

step 6 of the algorithm, the cost indices of these

displacements are simply calculated as follows:

The dynamic programming process commences atthe 7th step of the algorithm by considering the secondcolumn of cells The stage variable is defined as thenumber of the current cell i, which increases throughthe progression of the program

Step 8 checks if there is an un-processed cell in thiscolumn and the response ‘yes’ to this question leads tostep 9 to calculate the optimal trajectory for the nextun-programmed cell The RC

i is evaluated by ing the cost indices for different alternate paths forarriving to cell i, by using the following formula:

l corresponds to the cell which is placed in the

lth row above and k th column at the right hand of cell

i, and i1 are all cells in the previous column Afterfinding the optimal trajectory to cell i, the optimal RCindex (RCi) and the number of the latest cell in theoptimal trajectory (N) before arriving to cell i is saved.Figure 10 Cross section of fuzzy danger zone along the operator path

To take into account the complex trajectories, a

backward check is done, which recalculates the RC

index of all neighbourhood cells (Ni) that have an RC

index superior to RCi Step 10 of the algorithm

describes this process The possible decrease of the RC

index of neighbourhood cells is performed After

considering the RC index for each cell, it is necessary

to do the backward check to study the possibility of

further propagation of improvement to its

correspond-ing neighbourhood cells By uscorrespond-ing the steps 11 and 12

of algorithm, this process is continued for all columns

Step 13 is concerned with aggregating and

present-ing the optimal trajectory from start to destination

points The optimal trajectory is a set of ordered cell

numbers, identified one by one, by tracking back the

optimal sub-trajectories already found

4 Example

Figure 12 presents a simple example, including danger

zones and obstacles in a workplace To provide

traceability, several simplifications are applied The

workplace is placed in a rectangular area, divided into

30 cells Start and destination points are placed in the

cells 1 and 25, respectively The neighbourhood cells

for cell i are limited to the cells in the under-process

column and its previous column The relevant

‘From-To’ chart also is illustrated in this figure, containing

the RC index for the possible paths between the centres

of the neighbourhood cells The impossible paths are

marked by * in this chart These paths are barred by

the physical barriers

In this model, the risk function is defined as the

accumulated risk through the motion trajectory, and

the productivity is evaluated by considering thetrajectory length The programming process is in theup-down and left-right direction, respectively

The calculated optimal trajectory is illustrated bythe dotted line in Figure 13 Cell numbers, the optimal

RC index for arriving to every cell and the previous cell

in the optimal trajectory are illustrated for all cells.This trajectory connects the start and destinationpoints by passing through the obstacles and risk zones.The calculated total RC index is 156 units Afterapplying the backward check in cell 14, the RC indexfor cell 8 is improved The improvement is propagatedfor cells 3 and 4, in the previous columns The sameprocess for cell 29 leads to an improvement of the RCindex for cells 21 and 20 This way, a complextrajectory with optimal RC cost is identified

5 DiscussionThe proposed method provides a global optimalsolution for planning an operator’s path, by consider-ing the fixed obstacles and danger zones in the 2Dworkplace Danger zones are differentiated fromobstacles: an operator cannot pass through anobstacle; however, a danger zone can be crossed ifone accepts the relevant consequences The risk for atrajectory does not consider the duration of thepresence of an operator in a cell, especially cells indanger zones, the optimisation method can be easilymodified to consider such cases if required

The approximation of the operator’s movement as

a polygon is acceptable when one considers the naturaloperator movement’s tolerance The slopes and thesteps in the workplace can be modelled byFigure 11 Clustering the exemplified workplace to the rectangular programming cells

Figure 13 The optimal operator trajectory for the simplified example.Figure 12 A simple example of a workplace and the relevant From-To chart.

differentiating movement costs in the square From-To

chart, according to the motion direction

It is not necessary that the size of the cells be equal

or that their shape be square; however, they should be

sufficiently small to define the human trajectory with

an acceptable accuracy Testing of the model in

hypothetical workplaces that include obstacle and

danger zones showed its ability to converge to the

global optimum with an acceptable calculation cost

More investigation is planned by developing a

computer application to prove the method’s

applic-ability in a real design environment

6 Conclusion

In this paper, using dynamic programming, an

approach is developed to determine the optimal

trajectory for an operator in a dangerous workplace

The workplace area is discretised into square cells, and

the operator’s trajectory is approximated by polygons,

connecting ordered cell centres The dangers and the

effect of the barriers are modelled using fuzzy spaces

The risk-cost indices are calculated for trajectories by

considering the displacement length and the area

provided by the intersection of the operator path and

fuzzy danger zones A backward check method is used

to enhance the response space of complex trajectories

This model is used to map 2D trajectories where

danger zones are illustrated in computerised 3D

platforms

References

Bendahan, P., 2004 A neural network architecture to learn

the arm reach motion planning in a static cluttered

environment In: SMC 2004-IEEE conference on systems,

man and cybernetics, 1, 666–671

Campos, F.M.M.O and Calado, J.M.F., 2009 Approaches

to human arm movement control–A review AnnualReviews in Control, 33, 69–77

Chedmail, P., Chablat, D., and Le Roy, C., 2003 Adistributed approach for access and visibility task with

a manikin and a robot in a virtual reality environment.IEEE Transactions on Industrial Electronics, 50 (4), 692–698

Guichard, D., 2010 An introduction to higher mathematics.Washington: Whitman College

Hwang, Y.K., 1992 Gross motion planning – a survey ACMComputing Surveys, 24 (3), 219–291

Keef, P., et al., 2006 Real-time optimal reach postureprediction in a new interactive environment Journal ofComputer Science and Technology, 21 (2), 189–198.LaValle, S.M., 2006 Planning algorithms Cambridge: Cam-bridge University Press Available form: http://plan-ning.cs.uiuc.edu/booka4.pdf [Accessed 7 April 2010].Melchior, P., et al., 2003 Consideration of obstacle dangerlevel in path planning using A* and Fast-Marchingoptimisation: comparative study Signal Processing, 83,2387–2396

Meyer, J.A and Fillat, D., 2003 Map-based navigation inmobile robots: II A review of map-learning and path-planning strategies Cognitive Systems Research, 4, 283–317

Oustaloup, A and Linare`s, H., 1996 The CRONE pathplanning Mathematics and Computers in Simulation, 41,209–217

Shahrokhi, M and Bernard, A., 2004 A fuzzy approach fordefinition of dangerous zone in industrial systems In:SMC 2004 – IEEE conference on systems, man andcybernetics, 7, 6318–6324

Shahrokhi, M and Bernard, A., 2009 A framework todevelop an analysis agent for evaluating human perfor-mance in manufacturing systems CIRP Journal ofManufacturing Science and Technology, 2, 55–60.Shanmugavel, M., et al., 2009 Co-operative path planning ofmultiple UAVs using Dubins paths with clothoid arcs.Control Engineering Practice DOI: 10.1016/j.conengprac.2009.02.010

A system dynamics approach for strategic partnering in supply networks

Mohammad Reza Khaji* and Rasoul ShafaeiDepartment of Industrial Engineering, K.N Toosi University of Technology, Tehran, Iran

(Received 22 January 2010; final version received 1 October 2010)Individual businesses no longer compete as autonomous entities but rather by joining a supply chain alliance due tothe highly competitive business situation Supply chain coordination is truly a transformational business strategythat has a profound effect on competitive success and strategic partnering This paper conceptually integrates supplychain coordination and strategic partnering In this paper, we describe a generic system dynamics simulation modelfor strategic partnering in supply networks Our model addresses the whole supply chain starting from the suppliers

to the final customers and including production and distribution actors It is generic and can adapt to variousnetwork structures Finally, some scenarios based on cost and benefits are designed and the results are analysedwhich can assist the decision makers in a supply network

Keywords: strategic partnering; supply chain coordination; information sharing; system dynamics

1 Introduction

Individual businesses no longer compete as autonomous

entities but rather by joining a supply chain alliance due

to the highly competitive business situation Therefore,

suppliers, manufacturers, logistic companies, and

re-tailers in the supply chain always forge stronger

alliances, vertically or horizontally, to compete against

other supply chains There are three immediate benefits:

securing critical technologies and knowledge, expanding

market entry and share, dispersing costs and risks

(Auster 1989)

A supply chain is a dynamic, stochastic and complex

system that might involve hundreds of participants It

can be defined as a network of suppliers, manufacturers,

distributors and retailers, who are collectively concerned

with the conversion of raw materials into goods which

can be delivered to the customer Companies associated

in the same network require efficient supply chain

integration in order to optimise their collective

perfor-mances In order to achieve this goal, supply chain

management (SCM) has become recognised as a key

business competency Moreover, numerous companies

have started to appreciate that SCM plays a major role in

building a sustainable competitive edge for their

products in highly competitive markets (Jones 1998)

On the one hand, partnering between firms is an

increasingly common way for firms to find and maintain

competitive advantage (Mentzer et al 1999) A

partner-ship occurs through extensive social, economic,

servi-cing and technical ties over time, but it requires mutual

commitment, trust and common goals as well ascommunication and cooperation (Mentzer et al 2000).Thus, selecting an appropriate partner is a critical andstrategic decision-making process Evaluation attributesinclude both quantitative indices, such as annualproductivity and financial stability, as well as qualitativeindices, such as trademark reputation and communica-tion openness (Lin and Chen 2004) So a deep insightshould be given into the nature, motivation andimplementation of the strategic partnering as a value-added component in the forefront of SCM (Liu et al.2006)

On the other hand, supply chains are generallycomplex since numerous activities usually spread overmultiple functions or organisations and sometimes overlong time spans Therefore, it is necessary to overlay acoordination system, which may include an explicitdefinition of processes, responsibilities and structuresaligned with overall objective of the whole supply chain

to bring multiple functions and organisations together.The continuous evolving dynamic structure of thesupply chain poses many interesting challenges foreffective system coordination Supply chain memberscannot compete as independent members The productused by the final customer passes through a number ofentities contributed in the value addition of the productbefore consumption To improve the overall perfor-mance of supply chain, its members may behave as a part

of a unified system and coordinate with each other(Arshinder et al 2008)

*Corresponding author Email: mrkhaji@gmail.com

Vol 24, No 2, February 2011, 106–125

ISSN 0951-192X print/ISSN 1362-3052 online

Ó 2011 Taylor & Francis

DOI: 10.1080/0951192X.2010.531288

There are many driving forces for coordination in

supply chains For example, the innovative nature of

products, the length of the life cycle and the duration

of retail trends in the industries, the longer and more

complex supply chains and the general movement to

offshore production are only some of the associations

that move supply chains into that direction Global

markets and more competition are likely to move

supply chains towards a more universal participation

where final retailers and upstream suppliers will be

more willing to coordinate in an effort to cut costs

(Fliedner 2003, Polychronakis and Syntetos 2007)

According to Udin et al (2006) collaborative SCM can

be defined as a condition in which all parties in the

supply chain are dynamically working together

to-wards objectives by sharing information, knowledge,

risk and profits, and possibly involves consideration of

how other partners operate and make decisions

As outsourcing has increased, the scale of supply

chains has become larger, and each member in the

supply chain needs more information to improve his/

her efficiency and effectiveness (Parrish et al 2004)

Nowadays, companies are looking to apply enterprise

resource planning (ERP) systems and business

intelli-gence systems to supply chain management, where

implementing ERP systems and sharing information

among trade partners is an important issue to be

concerned with (Hodge 2002) Therefore, studying

information sharing in supply chains is important in

order to satisfy the needs of the members of the chain

and the customer However, sharing information

among members of a chain is a less researched area

within the general supply chain management literature

Sharing information among members of a supply chain

can reduce not only the Bullwhip Effect but also the

costs of the whole chain (Park et al 2003)

Consequently an effective strategic partnering

within supply network context should be based on

the concept of coordination and information sharing

In this paper, the customers’ desired priorities for

selecting retailers is considered as information to be

shared throughout the chain The process of selecting

upstream partners in a supply chain is based on

various attributes which have been investigated by

many scholars (Shui-ying and Rong-qiu 2001, Biehl

2005, Ha and Hong 2005) Sharing priority weights of

mentioned attributes will lead to formation of

co-ordination among upstream partners This

coordina-tion is directed towards maximising profitability of all

participants while gaining customers satisfaction

Hence, the main objective of this paper is to

develop a framework for: (1) enhancing information

sharing within supply chain members as a coordination

technique, (2) improving decision making process by

focusing on customer attributes in production

planning, and finally (3) strategic partnering withinsupply network participants by focusing on coordina-tion issues

The remainder of the paper is organised as follows.The results of a literature review on related subjectsincluding information shared in a supply chain, systemdynamics simulation in a supply chain, governance ofvalue chains, fuzzy analytical network process (FANP)method and balanced scorecard (BSC) are presented inthe next section In section 3, the proposed methodol-ogy is presented In the next section, the proposedmethodology is applied in supply network model, andthis is followed by a numerical example in section 5.Section 6 concludes the paper with a discussion of theimplications of this study, research directions andconcluding remarks

2 Literature review2.1 Shared information in a supply chainThe supply chain members coordinate by sharinginformation regarding level of demand, orders, in-ventory, etc Information sharing between downstreamand upstream partners in a supply chain is considered

to be a major indicator of the use of SCM Informationsharing is used, in effect, to integrate the entire valuechain into one longer chain (Shapiro et al 1993,Bhattacharya et al 1995, Rayport and Sviokla 1995,Towill 1997, Jain et al 2009) Timely information oradvanced commitments from downstream customershelp in reducing the inventory costs by offering pricediscounts This information can also be a substitute forlead time and inventory (Reddy and Rajendran 2005).The value of information sharing increases as theservice level at the supplier, supplier-holding costs,demand variability and offset time increase, and as thelength of the order cycle decreases (Bourland et al

1996, Chen et al 2000) There are some comparativestudies in which no information sharing policy iscompared with full information sharing policy In-formation sharing policy results in inventory reduc-tions and cost savings (Yu et al 2001) In other words,sharing information in a supply chain is important toreduce not only the Bullwhip effect but also the cost ofentire chain (Gavirneni et al 1999)

Most of the models assume that a supply chainpartner has complete information (including cost,demand, lead time, etc.) about the other partner.This is considered to be a major limitation of thesemodels In a decentralised supply chain, there is hardly

a situation where complete information is availablewith the parties Coordination under limited informa-tion sharing is an important issue of concern to bestudied for the decentralised supply chain (Sarmaha

et al 2006)

In terms of the information content classification,

Chopra and Meindl (2001) classified supply chain

information into supplier information, manufacturer

information, distribution and retailer information and

demand information Handfield and Nichols (1999)

classified supply chain information into 10 categories

Customer information includes customer forecast,

sales history, point of sale and promotional plan

Supplier information includes product line, product

lead time, capacity and production plan Inventory

information includes inventory level and inventory

cost Bensaou (1997) measured IT use by the

informa-tion exchanged in electronic forms in the following six

areas: purchasing, production control, quality,

engi-neering, transportation and payment Chen and Chen

(1997) found that the JIT environment required the

exchange of information between supplier and

manu-facturer in the following items: schedules, schedule

changes, design data, engineering changes, quality or

delivery issues, cost, etc Lummus and Vokurka (1999)

described the requirements of sharing information

among supply chain partners The information

in-cludes supplier information (e.g finished goods

inventory, MPS, and delivery information), consumer

information (e.g promotion plan and demand

fore-cast), retailer information (e.g inventory and POS)

and distributor information (e.g delivery schedule)

Chen (2002) provided a comprehensive literature

review about information sharing in supply chains

Several studies discuss the upstream passing of various

types of information: costs (Chen 2001), lead-times

(Chen and Yu 2001a) and uncommitted production

capacity (Chen and Yu 2001b) Other studies involve

passing information downstream In Chen (2002),

upstream information refers to the information

ex-changed between upstream members of supply chains,

while downstream information refers to the

informa-tion exchanged between downstream members of

supply chains Fulkerson (2000) suggested that sharing

bill of materials with distributors could help implement

postponement strategy in the supply chain A

post-ponement strategy delays the final assembly of

products until customer demand is known Lee and

Whang (2000) studied the PC industry, in which

manufacturers share information (e.g capacity,

de-mand forecasts, production plan, promotion plan,

POS, customer’s forecast, sales data) with their

suppliers They suggested five types of information to

share: inventory level, sales data, order status for

tracking, sales forecast and production/delivery

information

Huang et al (2003) classified information in the

supply chain into six categories: product, process,

planning, inventory, order and resource The

classifi-cation of information is shown in Table 1 Upstream

partner selection in a supply chain is based on the fourmost important factors – price, quality, service leveland lead time In this paper, the scope of informationshared throughout the chain has been restricted tothese factors as upstream partner selection criteria.These are actually the four most important criteria insimilar studies (Dickson 1966, Weber et al 1993,Talluri 2002)

Most organisations mainly concern the partnerselection decision because the cost of procuring isparamount to their profits In industrial companies,the cost of raw materials and component partspurchased from external partners typically rangedbetween 50% and 90% of the total production cost(Burton 1988) The other factor that leads to customersatisfaction is quality Consumers have heterogeneouswillingness to pay for quality Based on a distinctengineering principle, for a given production technol-ogy, the unit production cost tends to rise more rapidly

as quality increases (Chen 2006) As changing mer preferences requires a broader and faster demand

custo-of products and service, the companies urge for a moresystematic and transparent approach to purchasingdecision-making, especially regarding the area ofpartner selection (Carter et al 1998)

The efficient information flow between partners isidentified as the key to improve the time, quality,service and cost factors Meeting the customerobjectives satisfactorily depends on coordination ofinformation that helps produce highest quality, lowcost and minimum time to service (Titus and Bro¨chner2005)

2.2 System dynamics simulation in a supply chainSimulation is often performed to quantify results in thefield of supply chain management to support decisionssuch as designing supply chains (strategic level) andsetting control policies (operational level) (Kleijnen2005) Dynamic simulations are necessary to analyse

Table 1 Classification of production information model(PIM) (Huang et al 2003)

Planning Demand forecast, order schedule,

forecasting model, time fenceInventory Inventory level, holding cost, backlog cost,

service levelOrder Demand, demand variance, order batch size,

order due date, demand correlationResource Capacity, capacity variance, supply

the supply chain because it is interactive and contains

hierarchical feedback loops The merits of dynamic

simulations are that they can combine these feedback

loops with static simulations There are two major

approaches to simulate supply chain dynamics One

way is to simulate movement of a supply chain by

focusing on the dynamic features from the system

perspective The other approach is to demonstrate the

mechanisms related to the information distortion

(Chan and Lee 2005)

There are four types of simulation methodologies for

SCM: risk analysis via spreadsheet simulation, system

dynamics, discrete-event dynamic systems simulation

and business games (Kleijnen and Smits 2003) Systems

dynamics is a typical method to analyse dynamic

systems from the viewpoint of the whole system It was

called industrial dynamics at the beginning because it

focused on the shifting nature and behaviour of the

companies over the passage of time System dynamics is

a field of study which was founded by Jay Forrester at

the Massachusetts Institute of Technology in the 1950s

(Forrester 1961) Defining the concept of a system is

important in understanding the basic viewpoint of

system dynamics A system is a collection of parts which

interact in such a way that the whole network has

properties which are not evident from the parts

themselves (Coyle 1996) System dynamics is an

important systemic approach to problem solving that

views a system in a holistic manner rather than analysing

its elementary elements as in most conventional analytic

methods A core assumption of system dynamics is that

the world can be regarded as a set of linked systems

whose boundaries depend, in part at least, on the

viewpoint of the observer or analyst (Pidd 2003) System

dynamics is a simulation technique capable of explicitly

modelling the feedback loops of decision rules and

evaluating the dynamics of complex processes and

systems (Shin et al 2010) The field of system dynamics

has been thriving over recent decades Since the

inception of system dynamics, it has attracted great

attention from academia and found a wide range of

applications in practice by large companies, consulting

agencies, innovative universities, business schools and

government organisations If employed properly,

sys-tem dynamics modelling brings the following benefits

(Miser and Quade 1985):

Policies and actions that are effective and

efficient

Explicit considerations of assumptions,

uncer-tainties, costs, consequences, spillovers, etc

A logical framework for considering and setting

policy goals

Improved understanding of the issues and hence

better insights on the part of the decision-makers

New options, new goals, and new horizons thatexpand people’s perceptions of what might offerthem the chance of improving systemperformance

Generally, conceptualisation, formulation and mulation are the main outcomes achieved in a systemdynamics modelling approach (Richardson 1996).They are realised in two phases of system dynamicssimulation modelling: qualitative and quantitativesystem dynamics modelling Table 2 lists the mainobjectives of the two phases in system dynamicssimulation modelling (Wolstenholme 1993)

si-Qualitative modelling uses causal loop diagrams(CLD) to depict cause and effect relationships betweenvariables within the system boundary Then the CLDsare converted into a quantitative model using logicalrelationships and mathematical equations, and aresimulated using computer software applications todesign experiments by changing parameter values,system structures and strategies options (Wolsten-holme and Coyle 1983, Senge 1992)

The use of system dynamics modelling in supplychain has been very limited but recently, given thecomplexity in supply chains, has gained increasingpopularity The dynamic nature of supply chainsystems and their behaviour depends on the uncertain-ties of customers’ demand, different suppliers, differentlogistics routes, or alternative inventory methods, etc

In fact uncertainty rules the supply chain Therefore, it

is natural to apply system dynamics simulation.Perhaps the most well-known supply chain beingsimulated using system dynamics concepts is the BeerGame, dating back to early 1960s (Ashayeria and

Table 2 Qualitative and quantitative SD modelling(Wolstenholme 1993)

Qualitative system dynamics modelling

1 Creating and examining feedback loop structure ofsystems

2 Providing a qualitative assessment of the relationshipbetween system processes, information, organisationalboundaries and strategy

3 Estimating system behaviour and postulatingalternative strategy to improve behaviourQuantitative system dynamics modelling

1 Examining the quantitative behaviour of all systemvariables over time

2 Examining the validity and sensitivity of systembehaviour to changes in information structure,strategies and delays/uncertainties

3 Designing alternative system structure and controlstrategies based on intuitive ideas and controllingtheory algorithms in terms of non-optimisingrobust policy design

4 Optimising the behaviour of specific system variables

Lemmes 2006) The work was then complemented by

John Burbidge, who deployed his five golden rules to

avoid bankruptcy, yet it was not until 1997 that

Forrester’s work on system dynamics and Burbidge’s

concept of multi-phasing of the information flow were

officially merged into a set of best practices of

communication and material flow in the supply chain

(Holweg and Bicheno 2002) Angerhofer and

Angel-ides (2000) gave an overview of research work in these

areas until 2000, followed by a discussion of research

issues which has evolved, and presented a taxonomy of

research and development in system dynamics

model-ling in supply chain management (Table 3)

Bhushi and Javalagi (2004) reviewed industrial

dynamics study and applications of system dynamics

to various facets of SCM in 2004 According to their

work application of system dynamics modelling to

supply chain management covers the following facets:

(1) International supply chain management

(2) Inventory management

(3) Participate business modelling

(4) Supply chain design

(5) Demand amplification

(6) Information visibility

(7) Decision making in stock management

(8) Supply chain reengineering

(9) Integrated system dynamics approach

Georgiadis et al (2005) examined capacity

plan-ning policies for a food supply chain with transient

flows due to market parameters/constraints by use of

system dynamics Finally they demonstrate the

applic-ability of the developed methodology on a

multi-echelon network of a major Greek fast food chain

Ozbayrak et al (2007) in their paper proposed a

modelling framework which was used to simulate the

operation of a supply chain network of moderate

complexity The proposed model comprised four

echelons and was built around a central medium-sized

manufacturing company operating as a typical

make-to-order system The developed model was built using

a systems dynamics approach The modelling effortfocused on measuring the supply chain systemperformance in terms of key metrics such as inventory,WIP levels, backlogged orders and customer satisfac-tion at all four echelons Jia et al (2007) concentrated

on the common process of information sharinganalysis for supply chain systems They treated it as

a problem by a multi-dimension view to informationsharing They defined three dimensions for informa-tion sharing: the content of the information, the level

of the sharing and the time selection of sharing Rabelo

et al (2007) presented the approach that integrates theanalytic hierarchy process technique, system dyna-mics and discrete-event simulation to model the serviceand manufacturing activities of the global supplychain of a multinational construction equipmentcorporation

2.3 Governance of value chains

If a theory of global value chain governance is useful topolicymakers, it should be parsimonious It has tosimplify and abstract from an extremely heterogeneousbody of evidence, identifying the variables that play amajor role in determining patterns of value chaingovernance while holding others at bay, at leastinitially (Gereffi et al 2005) The definition of

‘governance’ is first introduced by Gereffi (1994),defined as ‘authority and power relationship thatdetermine how financial, material, and human re-sources are allocated and follow within a chain’ In the1990s Gereffi and others developed a new framework,called global commodity chains, to tie the concept ofthe value-added chain directly to the global organisa-tion of industries A set of strategic parameters can behighlighted as characterising governance types: ‘what’

or ‘how’ a product/service should be produced as well

as ‘when’, ‘how much’ and even ‘at what price’ (Sunand Zhang 2009)

The most recent valuable governance model hasbeen published by Gereffi et al (2005) They have gonedeeper into the analysis of factors affecting five kinds

Table 3 A taxonomy of research and development on system dynamics modelling in supply chain management (Angerhoferand Angelides 2000)

Category research areas

Research:

modelling fortheory building

Practice:

modelling forproblem solving

Putting research into practice:improving the modelling approach

Techniques and methods applied: (a) causal loop diagramming; (b) continuous simulation; (c) OR techniques; (d) discrete simulation.

of transactional linkages between lead firms and

subordinate firms: market, modular, relational, captive

and hierarchy They also identified the idea of three

key determinates of value chains relationships: the

complexity of information and knowledge transfer

required to sustain a particular transaction, especially

with respect to product and process specifications; the

extent to which this information knowledge can be

codified and, therefore, transmitted efficiently and

without transaction-specific investment between the

parties to the transaction; the capabilities of actual and

potential suppliers in relation to the requirements of

the transaction These five basic types of value chain

governance are (Gereffi et al 2005):

(1) Markets Market linkages do not have to be

completely transitory, as is typical of spot

markets; they can persist over time, with

repetitive transactions The essential point is

that the costs of switching to new partners are

low for both parties

(2) Modular value chains Typically, suppliers in

modular value chains make products to a

customer’s specifications, which may be more

or less detailed However, when providing

‘turn-key services’ suppliers take full

responsi-bility for competencies surrounding process

technology, use generic machinery that limits

transaction-specific investments and make

ca-pital outlays for components and materials on

behalf of customers

(3) Relational value chains In these networks there

are complex interactions between buyers and

sellers, which often create mutual dependence

and high levels of asset specificity This may be

managed through reputation, or family and

ethnic ties Many authors have highlighted the

role of spatial proximity in supporting

rela-tional value chain linkages, but trust and

reputation might well function in spatially

dispersed networks where relationships are

built-up over time or are based on dispersed

family and social groups

(4) Captive value chains In these networks, small

suppliers are transactionally dependent on

much larger buyers Suppliers face significant

switching costs and are, therefore, ‘captive’

Such networks are frequently characterised by

a high degree of monitoring and controlling by

lead firms

(5) Hierarchy This governance form is

charac-terised by vertical integration The dominant

form of governance is managerial control,

flowing from managers to subordinates, or

from headquarters to subsidiaries and affiliates

2.4 Fuzzy analytical network process (FANP)method

The FANP is a generalisation of the FAHP as a widelyused multi-criteria decision-making tool by replacinghierarchies with network More recently, a moregeneral form of FAHP approach, which incorporatesfeedback and interdependent relationships amongdecision criteria and alternatives, has been proposed

as a more accurate approach for modelling complexdecision environments While FAHP is a well-knowntechnique that decomposes a problem into severallevels in such a way that they form a hierarchy, FANPenables interrelationships among the decision levelsand criteria to be taken into consideration in a moregeneral form Thus, the FANP can be used as aneffective tool in those cases where the interactionsamong the elements of a system form a networkstructure (Saaty 1996)

Since nature of decision making usually includesuncertainty, it is sufficient to apply fuzzy concepts inproblems which human has a role in them (Zadeh1965) Further to the fuzzy set theory introduced byZadeh, it has been applied in various contexts(Zimmermann 1994)

Fuzzy ANP is investigated by several researches.Kahraman et al (2006) used fuzzy FANP for QFDplanning process in which the coefficients of theobjective function are obtained from a fuzzy ANPapproach Lin and Hsu (2008) considered performancemeasurement systems by fuzzy ANP The essential point

is the existence of inner dependence between objectivesand criteria For this purpose, the super matrix method

is applied as follows (Kahraman et al 2006):

As shown in Figure 1, if W1represents the weightvector of the objectives in respect to goal, W2 is amatrix that denotes the impact of the objectives oneach of the criteria W3and W4are the matrices thatrepresent the inner dependence of the objectives and

Figure 1 Configuration of the problem (Saaty 1996)

the inner dependence of the criteria, respectively, then

the super matrix of the problem is as follows:

3

In this phase experts conduct pair wise comparisons

Since there is uncertainty in decisions, they asked to

express their opinions with linguistics data The

Chang’s extent analysis method is used to obtain

weights (Chang 1996)

If ~M1¼ (l1, m1, u1) and ~M2¼ (l2, m2, u2)

repre-senting two triangle fuzzy numbers (Figure 2), where d

is the ordinate of the highest intersection point d

between two membership functions The value of ~Mk

relates to row k and is calculated as follows:

Xm j¼1

Mij

ð2Þ

~

Mijis the element in row i and column j

The degree of possibility of ~M2 ~M1is defined as

The followings are the steps of the algorithm:Step 1 Determining importance degrees ofobjectives by assuming that there is no depen-dence among objectives: Calculation of W1Step 2 Determining the importance degrees ofcriteria with respect to each objective by assumingthat there is no dependence among the criteria:Calculation of W2

Step 3 Determining the inner dependency matrix

of the objectives with respect to each objective:Calculation of W3

Step 4 Determining the inner dependency matrix

of the criteria with respect to each criterion:Calculation of W4

Step 5 Determining the interdependent priorities

of the objectives: Calculation of WA¼ W36 W1Step 6 Determining the interdependent priorities

of the criteria: Calculation of WB ¼ W46 W2Step 7 Determining the overall priorities of thecriteria: Calculation of WANP¼ WB6 WA

2.5 Balanced scorecard in a supply chainRecently, an increasing number of the literature focus

on the adaptation of BSC to fit the needs of SCM(Brewer and Speh 2000, Bullinger et al 2002).Balanced scorecard (BSC) receives broad attentionnot only in scientific literature but also in practicalapplications In addition to financial criteria, the BSCcomprises a customer perspective, a learning andgrowth perspective as well as an internal businessperspective These perspectives can integrate a set ofattributes that provides a deeper insight for decisionmaking (Stadtler and Kilger 2005) Every attributeselected for a scorecard should be part of a link of

Figure 2 Comparison of two triangular fuzzy numbers

cause-and-effect relationships, ending in financial

objectives that represent a strategic theme for the

business The attributes are designed to pull

organisa-tions towards the overall vision This methodology is

consistent with the approach of supply chain

manage-ment by helping organisations to overcome traditional

functional barriers and ultimately lead to improved

decision making and problem solving (Waters 2007)

In this paper, the concept of balanced scorecard (BSC)

perspectives is applied which links financial and

non-financial, tangible and intangible, inward and outward

factors as customers’ objectives for prioritising the

attributes that affect selection of upstream partners in

a supply chain

There are three types of relations among the

factors: first, direct relations including subordinate

relations, feedback relations and dominating relations;

secondly, indirect relations, in which the subordinate

relations are ambiguous and the mutual influences

between each two are transferred by another index;

third, self feedback or self-associated relations These

three relations embrace all the ways in which BSC

indexes interact (Yu and Wang 2007) Kaplan and

Norton (1993, 2004) articulated four perspectives that

can guide companies as they translate strategy into

actionable terms:

Financial perspective The revenues, profit margins,

and expenses are very important to an organisation

seeking to achieve its goals A common mistake with

organisations is that they normally do not link the

financial goals with the non-financial strategic

objec-tives of the company The financial perspective gives

respect to the relationship between stated financial

goals and other goals that feed the machine to create

the result

Customer perspective The customer perspective is

viewed as the set of objectives the organisation must

achieve to gain customer acquisition, acceptance and

perpetuation Objectives are an outgrowth of

assump-tions made about the customers and their attitudes, the

markets they represent and the value they perceive in arelationship with the organisation

Internal perspective The internal perspective minds us that the background works, driven byobjectives and goals, must be in place to ensure thatthe customer and financial objectives are achieved.Internal processes, cultures and procedures in alldepartments and business units support the valueproposition to the target market segments

re-Learning and growth perspective This perspective isthe basis for all other perspectives and serves to remindthe practitioner that the basis for all other results in theinternal, customer and financial perspectives are found

in the learning and growth of the people Learningdictates how people absorb new ideas, improve theirskills and turn them into action

Chiang (2005) proposed a dynamic decision proach for long-term vendor selection based on AHPand BSC Ravi et al (2005) combined analytic networkprocess and balanced scorecard for conducting reverselogistics operations for EOL computers Leung et al.(2006) applied the analytic hierarchy process andanalytic network process to facilitate the implementa-tion of the balanced scorecard Leem et al (2007)proposed modelling the metrics for measuring theperformance on logistics centres by BSC and ANP inthe Korean context Xue-zhen (2007) proposed adynamic model based on AHP and BSC for long-term strategic vendor selection problems

ap-3 Proposed methodologyAccording to the extensive literature (Weber et al

1991, Choi and Hartley 1996, Boer et al 1998, Ni et al.2007), it can be concluded that some properties areworth considering for upstream partner selection in asupply chain First, the criteria may be considered inquantitative as well as qualitative dimensions (Weber

et al 1991, 1998, Choi and Hartley 1996, Verma andPullman 1998, Dowlatshahi 2000) In general, theseobjectives among these criteria are conflicted Astrategic approach towards supplier selection mayfurther emphasise the need to consider multiple criteria(Ellram 1992, Donaldson 1994, Swift 1995) Second,several decision-makers are very often involved in thedecision process for upstream partner selection (Boer

et al 1998) Third, decision-making is often influenced

by uncertainty in practice An increasing number ofdecisions can be characterised as dynamic andunstructured Situations are changing rapidly or areuncertain and decision variables are difficult orimpossible to quantify (Cook 1992)

A partner selection framework is necessary to bedesigned to assist the decision-making process inselecting efficient and compatible partners (Huanga

Table 4 Linguistic scales for importance

Linguistic scale

for importance

Triangularfuzzy scale

Triangular fuzzyreciprocal scale

Equally important (EI) (1/2, 1, 3/2) (2/3, 1, 2)

Weakly more

important (WMI)

(1, 3/2, 2) (1/2, 2/3, 1)Strongly more

important (SMI)

(3/2, 2, 5/2) (2/5, 1/2, 2/3)Very strongly more

important (VSMI)

(2, 5/2, 3) (1/3, 2/5, 1/2)Absolutely more

important (AMI)

(5/2, 3, 7/2) (2/7, 1/3, 2/5)

et al 2004) So there is a need for a systematic

approach to elicit customers’ preferences based on

their strategic perspectives Supply chain members

often select their upstream partners based on their

strategic priorities Strategic priorities of customers for

selecting retailers can be categorised in four

perspec-tives, namely financial, customer, internal process, and

learning and growth, as defined in the balanced

scorecard Balanced scorecard is a carefully selected

set of measures derived from an organisation’s strategy

(Niven 2002) Perspectives of balanced scorecard and

attributes for selecting upstream partners in a supply

chain are interrelated (Moser 2007)

Therefore, in the addressed supply chain network,

one will face a multiple criteria decision making

problem with BSC perspectives as criteria for

obtain-ing priority weights of upstream partner selection

attributes

Unlike many traditional multiple criteria

decision-making methods that are based on the independent

assumption, the analytic network process (ANP) which

incorporates interdependence relationships between

perspectives and attributes is a new approach for

multi-criteria decision making The analytical network

process (ANP) provides an effective tool for solving

complex decision-making problems Due to its

con-sideration of interdependence between the elements of

the decision problems, the ANP method establishes a

better understanding of the complex relationships

between the elements in decision making, and at the

same time improves the reliability of decision making

(Jharkharia and Shankar 2007) Saaty and Vargas

(2006) suggested that ANP can be used in many

disciplines such as political, economic, social,

techno-logical, etc Thus, an effective model has been

developed based on BSC and ANP to help customers

in supply chains to evaluate the priority weights of

attributes for selecting upstream partners

On the other hand, managing a supply chain is very

difficult, since various sources of uncertainty and

complex interrelationships between various entities

exist in the supply chain In general, a supply chain is

defined as follows (Mabert and Venkataramanan 1998):

A supply chain is the network of facilities and activities

that performs the functions of product development,

procurement of material from vendors, the movement

of materials between facilities, the manufacturing

products, the distribution of finished goods to

custo-mers, and after-market support for sustainment

Based on this definition, such a network in a system

contains a high degree of imprecision This is mainly

due to its real-world character and its imprecise

interfaces among its factors, where uncertainties in

activities from raw material procurement to the end

user make the SC imprecise Thus, it is summarisedthat fuzzy set theory is a suitable tool to come up withsuch a complicated system (Zarandi et al 2002).Therefore, in this paper fuzzy analytical networkprocess (FANP) method is used in order to increase thereliability of customers’ priorities for selecting upstreampartners in a supply chain since in many cases decisionmakers can be uncertain about their own level ofpreference, due to incomplete information or knowl-edge, complexity and uncertainty within the decisionenvironment, or a lack of appropriate measurementunits and scale In addition, the preference model of thehuman decision maker is uncertain, and it is relativelydifficult for the decision maker to provide exactnumerical values for the comparison ratios Duran andAguilo (2008) argued that by adopting fuzzy numbersdecision makers would be able to achieve a betterflexibility in estimating the overall importance ofattributes in developing real alternatives to assessproblems with greater confidence Consequently, sincefuzzy set theory can give a much better representation ofthe linguistic data (Cheng et al 1999), this research used

an FANP base to calculate customers’ priorities forselecting an upstream partner in a supply chain.There is extensive literature available on theanalysis of supply chains (Mertins et al 2005, Pirard

et al 2008) Two of the most common ways ofanalysing a supply chain are through simulation andanalytical modelling On the analytical front, effortshave been made to integrate two or more activities andsolve them together The major activities of a supplychain include sourcing of raw materials, manufactur-ing the product and distributing the finished product(Folie and Tiffin 1976) In the analytical method, as theproblem size increases, obtaining solutions becomesmore difficult Moreover, even for reasonable sizedproblems, it is not easy to consider all aspects of theproblem in analytical solutions, especially the uncer-tainty This is where the simulation approach ispreferable It is easier to imitate the real life problem

in a simulation model (Swaminathan et al 1998).Simulation provides a flexible and useful mechanismfor capturing uncertainty and modelling dynamicfeedback interactions which are the characteristics of

a complex system (Mahanti and Antony 2005)

In this paper, in order to simulate the supply chainnetwork, system dynamics modelling, as a powerfulsimulation approach, was used System dynamics is apowerful simulation approach to problem solving thatviews a system in a holistic manner rather thananalysing its elementary elements as in most conven-tional analytic methods System dynamics is a rigorousmethod for qualitative description, exploration andanalysis of complex systems in terms of their processes,information, organisational boundaries and strategies

It also facilitates quantitative simulation modelling

and analysis for the design of system structure and

behaviour (Wolstenholme 1990) Simulation models

developed by system dynamics methodology are

important tools for strategy development and policy

planning as an expert support tool (Yim et al 2004) In

addition system dynamics model can be used for

simulating information visibility in supply chains The

main advantages of system dynamics models, as an

integral part of decision support systems, are the

possibility of dynamic analysis of the considered

problem under different scenarios (Larsen et al 1997)

In summary, in the proposed model some major

features such as information sharing, multiple layers,

uncertainties, dynamic environment and financial

flows are considered This study has two types of

contribution to the literature, first in some of these

features new ideas have been taken into account and

secondly a holistic approach is used to integrate these

features together In Table 5 the proposed model is

compared with some other models found in the

published literature

According to the above descriptions, in the present

study, the phases of the proposed methodology for

strategic partnering in supply networks (Figure 3) are

as follows:

(1) Supply network configuration The supply

net-work configuration deals with structural design

and configuration of the relationships between

an enterprise and its suppliers and customers

So in this phase the number of network layers is

determined and in each layer main players are

identified

(2) Acquisition of customers’ priorities Customer

priorities are regarded as privileged information

and will be shared throughout the network This

information is acquired by use of FANP method

considering BSC perspectives as criteria

(3) Model customisation Customisation deals with

all the characteristics which are needed to

adjust the model to a real world situation

These characteristics include production

attributes, storage capacities, transportationsystem, costs and expected revenues associatedwith different partners in the network

(4) Software implementation of the model In thisphase for the purpose of illustrating the modelproficiency and validity, and obtaining theresults while a sufficient period of time haspassed, the chosen model will be implemented

in simulation software

(5) Simulation and analysis After implementing themodel in the software, simulation will be run anumber of times under different conditions.The simulation results obtained from differentconditions should be compared, discussed andconfronted

(6) Effective relationships determination In thosesimulation results for which the system reachesthe steady state, the best solution will beregarded as optimum and will be the basis fordeterminations of effective relationships in thenetwork Therefore, in the steady-state period

of the optimal solution, there are differentvolumes of transaction between the networkpartners If the steady state is assumed to besufficiently close to the real optimum solution

of the network, the relationships can beweighted according to their volume of transac-tion in the steady-state period In this mannerthe system will reach the steady state from thebeginning of the run

(7) Relationship change management and tation After determining the effective relation-ships, the type of some relationships should bechanged from market based to relational type.This phase varies from case to case, and thereshould also be some practical considerationswhen implementing these changes

implemen-4 Application of the proposed methodology in a supplynetwork model

Supply network has a multi-level structure so that eachlevel is influenced by various entities’ decisions in the

Table 5 The comparison of proposed model with other models found in published literature

Model

Informationsharing Multiple layers Uncertainties

Dynamicenvironment

Financialflows

*Features that have been taken into account in the model.

network Here, the proposed network consists of four

layers: customer, retailer, manufacturer and supplier

(Figure 4) The aim of this paper is to develop a

method to manage a multi-layer supply chain

archi-tecture in a dynamic, collaborative and optimised

manner In fact, every layer in the system described

above contains a series of actors with management

functions and decisions to protect themselves against

changes in demand, supply and manufacture

Each of the players in the supply network is

modelled as an actor who makes independent decisions

based on information gathered from the upstream and

downstream levels Each actor in system dynamics

modelling is represented by two level variables:

inventory and the backlog, which are integrated

through some rate variables All incoming orders are

aggregated into backlog, which is a level variable, and

by filling the orders of products, backlogs will

decrease This mechanism is illustrated in Figure 5

In addition, there exists more than one member in

each layer All members in each echelon are

perform-ing the same activities, rivallperform-ing each other, tryperform-ing to

increase demand for their products Consequently each

member in this network should select one, two or more

counterparts from its upstream trading partners

Briefly the buyers in a market fundamentally are faced

with the supplier selection Moreover, the presence of

multiple suppliers will require the buyer to set up a

competitive mechanism for capacity allocation among

the selected suppliers Thus, an evaluation of each

potential supplier, who responds to a call for aproposal from a customer according to rules andcriteria which are impartial and common to all, can becarried out Hence, procurement generally involvesmany criteria other than price For example, productquality, payment terms and delivery conditions arealso commonly treated as negotiable criteria In themodel among various criteria investigated by scholars,four attributes, namely price (P), lead time (LT),quality (Q) and service level (SL), are assumed Theseattributes will cover approximately all the needs andpriorities of customers

For the purpose of supplier selection, a variable iscreated, using some variables related to the perfor-mance of the supply network actors and somevariables regarding customers’ priorities This variable

is called the desirability ratio which can be calculatedfor all retailers related to customers’ perspectives.Figure 6 depicts the details of desirability ratio of thefirst retailer for the first customer

The desirability ratio is calculated through thefollowing formula which contains indices and weights:

where these indices and weights are calculated andnormalised as follows:

Service level (SL) Service level is used in supplychain management and in inventory management tomeasure the performance of inventory systems In thismodel, SL Rj is defined as the ratio of retailer’sbacklogs to its total incoming orders This ratio variesbetween zero and one, and when it is near zero, itreveals there isn’t much backlog, so service level ishigh; on the other hand, if the ratio is near one itmeans that a high percentage of orders are not fulfilled

Figure 3 Phases of the proposed methodology

Figure 4 Supply network model

on time, and so the service level is low This variable is

Lead time (LT) Lead time is the period of time

between the initiation of any process of production

and the completion of that process A more

conven-tional definition of lead time in the supply chain

management realms is the time from the moment the

supplier receives an order to the moment it is shipped

In this model lead time of retailers is calculated

through summation of lead times of upstream partners

and their share of retailer’s demand The value of

calculated lead time should be compared to desired

lead time proposed by customers For this purpose

lead time of the retailer is divided by lead time of the

customer; if this ratio is near one, it means lead times

are approximately equal; if the ratio is greater than

one, it reveals that lead time of retailer is not desirable

for customer; otherwise when the ratio is smaller than

one, it means the retailer will deliver the goods earlier

than the time proposed by customer This variable is

is calculated through summation of its cost andpreferable profit Cost is calculated through summa-tion of transportation cost and price of upstreampartners with respect to demand share of upstreampartners This variable is normalised as follows:

how it compares to competitors in the marketplace.

Manufacturers are spending more and more money on

quality control to provide a quality product and avoid

customer returns Quality of Rjaddresses the quality of

products which will be delivered to customers and is

obtained through aggregation of upstream partners’

quality and their share of retailer’s demand This

variable is normalised as follows:

In addition service level Ci, lead time weight Ci,

price Ci, and quality Ciare the priority weights of the

first customer for selecting its upstream partners These

weights are obtained based on BSC perspectives as

objectives In order to obtain priority weights of

upstream partner selection attributes, FANP method

based on Table 4 is applied in a three-step procedure as

follows:

Step 1 Acquiring the decision makers’ assessments

of comparing BSC perspectives At first, by

assuming that there is no dependence among

perspectives, the importance degrees of

perspectives (W1) are determined Then, thedecision makers are asked to define the relationnetwork among the BSC perspectives and based

on the network, the importance degrees of BSCperspectives with respect to each perspective (W3)are determined

Step 2 Acquiring the decision makers’ assessments

of comparing upstream partner selection attributes

In this step, the importance degree of influentialupstream partner selection attributes (W2) isdetermined Then, by defining the relation net-work among selection attributes, the importancedegrees of upstream partner selection attributeswith respect to each selection attribute (W4) isdetermined

Step 3 Calculating and analysing interdependentpriorities According to FANP method, theinterdependent priorities of the BSC perspectives(WA), the interdependent priorities of influentialcriteria in each BSC perspective (WB) and overallweights of upstream partner selection attributes(WANP) are calculated

When the desirability ratio of all customers andretailers is obtained, market share of retailers fromcustomers’ demand, which is called purchase ratio(PR), can be calculated as follows:

PR Ci Rj ¼ D:R: Ci RjP  Demand Ci

After calculating all ratios, the ratios associatedwith each retailer are summed up, and the result of thissummation for each retailer is considered as itsdemand from upstream partners Now the selectionprocess should be repeated for retailers and also formanufacturers For this purpose, required informationregarding customers’ priorities should be collected andshared throughout the chain among its partners Thiskind of information sharing will be an important toolfor ensuring customer satisfaction Consequentlyretailers’ priorities are obtained through aggregation

of customers’ priorities and their purchase ratios Withthis regard, by calculating retailers’ demand andpriority weights, the selection process can be donefor retailers again

In this paper, manufacturers are assumed to be able

to consider retailers priorities in their productionprocess Manufacturers have different options forproduction planning regarding the retailers’ priorities

In the proposed model these options are calledproduction policies and are obtained through a rulebased approach Manufacturers can consider threepolicies with different quality, price and lead time.These policies are as follows:

Figure 6 Components of desirability ratio

First policy In this policy, low price has high

priority compared with other criteria of the selection

process It means customers are ready to buy their

products at a cheaper price, even if it sacrifices quality

and lead times So in this production policy price and

quality are lower, and lead times are longer compared

with other policies

Second policy In this policy, high quality has high

priority compared with other criteria of selection

process It means customers are ready to buy their

products in higher quality, even if it sacrifices prices

and lead times So in this production policy price and

quality are higher, and lead times are longer compared

with other policies

Third policy In this policy, short lead time has high

priority compared with other criteria of selection

process It means customers are ready to receive their

products sooner, even if it sacrifices prices and quality

So in this production policy price and quality are

lower, and lead times are shorter compared with other

policies

Finally through a rule based approach,

manufac-turers will propose a production policy to each retailer

which has the most compatibility with their customers’

priorities

Material and information flows in the supply

network model have been delineated and are visible

now But there is more to supply network management

than just material and information flows and financial

flows are also important Consequently, beside

materi-al and information flows, financimateri-al flows are considered

by defining some rate and level variables throughout

the network (Figure 7)

As shown in Figure 7, for each customer there is alevel variable which aggregate customer credits The ratevariable income is calculated through sells to down-stream partners; the rate variable cost is calculatedthrough paying to upstream partners Also some otherlevel variables are defined to show debts of each of thenetwork actors to its upstream partners When there is

an order to an upstream partner amount of debts willincrease and debts will be paid when credit exists

In addition each upstream actor prioritises itsbuyer and fulfills his/her orders with regard to theirrank Customer prioritisation, a process to identify acompany’s most valuable customers, is commonplace

in most industries but there isn’t much concern aboutthis in supply chain literature In this model the basisfor prioritising the downstream partners is their ordersize, credit, and loyalty According to the obtainedrank of each customer, its orders will be fulfilled Thepriority of each partner according to its upstreampartner is obtained according to the following formula:

priority S1 M1 ¼ order ratioS1 M1  loyalty ratioS1 M1  debt ratio S1 M1;

ð12Þ

where the order ratio is the weight of each customer’sorder to overall demand of the upstream partner in thecurrent period, and it is calculated as follows:

as the ratio of total orders of each costumer to itsupstream partner from the beginning of time Theloyalty ratio is shown in the following formula:loyalty ratio S1 M1

debt ratio S1 M1

¼ debt M2 S1 þ debt M1 S1

ð15ÞFigure 7 Financial flows in the supply network

Finally each customer’s order will be fulfilled

according to the following formula:

order fulfill S1 M1¼ priority S1 M1

priority S1 M1þ priority S1 M2

 rate out S1;

ð16Þwhere the ratios are obtained as discussed above, and

rate out is the total number of products produced in

each period which is ready to be moved to the

customers

5 Simulation and discussion

As mentioned, the developed model gives the strategic

decision makers a framework strategic partnering

which improves the process of decision making and

the total performance of the network In the proposed

model both quantitative and qualitative information

are shared within supply chain members Here, one

example is delineated for each dimension of

informa-tion, i.e total cost and customer satisfaction for

quantitative and qualitative dimensions, respectively

To illustrate the model proficiency and validity, a

prototype system is implemented using Vensim

soft-ware In this study, two scenarios are designed In the

first scenario, the proposed approach is considered

The second scenario is based on the fixed interval order

system which is a classical inventory control model and

the selection process is done according to earliest due

date (EDD) method These two scenarios are

investi-gated in two different environments, namely, stable

environment in which demand variation is not high

and in an unstable environment in which demand

variation is high Consequently, four scenarios, which

are summarised in Table 6, are constructed

Before starting the simulation, the information on

customers’ priorities should be collected For this

purpose a numerical example was used to demonstrate

the procedure In order to obtain priority weights of

upstream partner selection attributes, FANP method

based on Table 7 was applied in three steps and the

following results were obtained The performance of

both scenarios is measured using total cost in terms of

costs associated with inventory and backlog orders.Figures 8 and 9 depict the time series plot of the totalcost

The results presented in Figure 8 reveals that thefirst scenario outperforms the second one, and duringthe time passed the difference between the scenarios isstriking Since the second scenario considers the safetystocks, a lower total cost has been obtained, except atsome time in the beginning of the simulation Hence, itcan be interpreted that some time units are required forthe first scenario to adjust the network and afterwards

a remarkable decrease in the costs can be observed.Therefore, the proposed scenario will be preferable inlong-term runs In addition, comparison of the last twoscenarios reveals the same results However, a closerlook at Figure 9 demonstrates additional informationabout these scenarios The proposed scenario seems to

be more sensitive to the demand variation than thescenarios based on the fixed interval order system.Other aspects of information sharing may be applic-able to reduce effects of demand variation in thenetwork total costs

In addition, the performance of each scenario wasinvestigated in terms of customer satisfaction It wascalculated as the average of desirability ratios ofcustomers Figures 10 and 11 depict the time seriesplot of the customer satisfaction

For the purpose of comparison, means andvariances of different scenarios should be comparedwith each other A statistical hypothesis test is amethod of making statistical decisions using experi-mental data and can be used for comparing means andvariances of different populations For the first andsecond scenarios two tests with a confidence level of99% were performed, assuming that in the nullhypothesis means and variances are equal and in thealternative hypothesis they are not equal The resultsindicate that in terms of both the mean and variancevalue of customer satisfaction index, the secondscenario outperforms the first scenario Same resultswere achieved using the same tests for third and fourthscenarios

The first scenario, in which the system reaches thesteady state, will be regarded as optimum and will bethe basis for determinations of effective relationships inthe network Before running the model all

Table 6 Constructed scenarios

relationships are considered to be market based whichwas introduced in the literature as a type of value chaingovernance, so there isn’t any difference betweenvarious relationships in the model In the initialnetwork, market linkages do not have to be completelytransitory, as is typical of spot markets (Figure 12).Therefore, in the steady-state period of the optimalsolution, there are different volumes of transactionbetween the network partners This indicates that therelationship between some partners differs from others

in terms of the volume and frequency of transactions.These differences have caused the network to reach theoptimum steady state after long simulation runs Inthis scenario the results for some values of parametersare summarised in Figure 13

Table 7 Weights of upstream partner selection attributes

Figure 11 Customer satisfaction

Figure 12 Initial network

Figure 13 Optimised networkFigure 10 Customer satisfaction

Figure 8 Total costs

Figure 9 Total costs

If the steady state is assumed to be sufficiently close

to the real optimum solution of the network, the

relationships can be weighted according to Figure 13

These weights are associated with the volume and

frequency of transactions between each player and its

upstream partners For example, the relationship

between the second supplier and first manufacturer

has a relatively large weight, so this relationship should

be converted from market based to relational model

which is built-up over a long time Consequently, in the

optimised network each player should transact with its

upstream partners according to weights shown in

Figure 13 In this manner, by changing all of

highlighted relationships to relational model the

net-work will reach the steady state from the beginning of

the run

6 Conclusion

Supply chain management is concerned with the

coordination of material, information and financial

flows within and across often legally separated

organisa-tional units With the recent advances in information

technology, real time data exchange has become feasible

and affordable As a result, an equally (if not more)

important issue for supply chain coordination is to

incorporate information into a coordination policy

Collaborating supply chain across organisational

boundaries may be one of the most difficult aspects of

supply chain management Many firms are simply

unaware of the fundamental dynamics of supply chains,

but even those firms that are enlightened enough to

understand these dynamics are often unable to realise

inter-organisational coordination Coordination can be

a very stable platform for strategic partnering

In this paper, the strategic partnering problem has

been studied in a supply network model The proposed

model addresses the whole supply chain starting from

the suppliers to the final customers and including

production and distribution actors It is generic and

can adapt to various network structures The main

objective of this paper is to develop a framework for:

(1) enhancing information sharing within supply chain

members as a coordination technique, (2) improving

the decision-making process by focusing on customer

attributes in production planning, and finally (3)

strategic partnering within supply network participants

by focusing on coordination issues The information

sharing within supply chain members is considered in

two dimensions The first dimension is the quantitative

information on demand, inventory and backlog, which

will decrease bullwhip effect, and improve efficiency of

production planning and inventory control The

second dimension is the qualitative information on

customer preference, which can be used in decision

making and planning that will lead to customersatisfaction In this model the decision-making process

is based on four prominent aspects of customersatisfaction, namely price, lead time, quality andservice level This framework can be used by strategicdecision makers who need comprehensive models toguide them in efficient decision making that increasesthe profitability of the entire chain It means thisframework can help actors of the network to strength-

en their strategic relations with upstream and stream partners Also, this framework provides themwith the information about the most efficient policiesthey can use These findings warrant that furtherresearch is needed for investigating more facets ofcoordination in the supply chain The performance ofthis model may be improved by defining some aspects

down-of information sharing within tiers down-of the network Inaddition, using game theory models may enhance theapplicability of the proposed model with real cases byconsidering competition between different playersacting in the same layer

ReferencesAngerhofer, B.J and Angelides, M.C., 2000 Systemdynamics modelling in supply chain management:research review In: Proceedings of simulation conference,Vol 1, Winter 2000, 342–351

Arshinder, K.A and Deshmukh, S.G., 2008 Supply chaincoordination: Perspectives, empirical studies and re-search directions International Journal of ProductionEconomics, 115 (2), 316–335

Ashayeria, J and Lemmes, L., 2006 Economic value added

of supply chain demand planning: A system dynamicssimulation Robotics and Computer-Integrated Manufac-turing, 22, 550–556

Auster, E.R., 1989 International corporate linkage: dynamicforms in changing environments Columbia Journal ofWorld Business, 22 (2), 3–6

Bensaou, M., 1997 Interorganizational cooperation: the role

of information technology, an empirical comparison ofU.S and Japanese supplier relations Information Sys-tems Research, 8 (2), 107–124

Bhattacharya, A.K., Coleman, J.L., and Brace, G., 1995 positioning the supplier: an SME perspective ProductionPlanning and Control, 6 (3), 218–226

Re-Bhushi, U.M and Javalagi, C.M., 2004 System dynamicsapplication to supply chain management: a review In:

2004 IEEE International Conference on EngineeringManagement,Vol 3, 1244–1248

Biehl, M., 2005 Selecting internal and external supply chainfunctionality: the case of ERP systems versus electronicmarketplaces Journal of Enterprise Information Manage-ment, 18 (4), 441–457

Bourland, K.E., Powell, S.G., and Pyke, D.F., 1996.Exploting timely demand infomration to reduce inven-tories European Journal of Operational Research, 92,239–253

Brewer, P.C and Speh, T.W., 2000 Adapting the balancedscorecard to supply chain management Supply ChainManagement Review, 5 (2), 48–56

Bullinger, H.-J., Ku¨hner, M., and van Hoof, A., 2002.

Analyzing supply chain performance using a balanced

measurement method International Journal of

Produc-tion Research, 40 (15), 3533–3543

Burton, T.T., 1988 JIT repetitive sourcing strategies: ‘‘Tying

the knot’’ with your suppliers Production and Inventory

Management Journal, Fourth quarter, 38–41

Carter, P.L., et al., 1998 The future of purchasing and

supply: a five- and ten- year forecast CAPS Research

Report AT Kearney, Arizona State University

Chan, C.K and Lee, H.W.J., 2005 Successful strategies in

supply chain management New York: Idea Group

Chang, D.Y., 1996 Applications of the extent analysis

method on fuzzy AHP European Journal of Operational

Research, 95, 649–655

Chen, F., 2001 Auctioning supply contracts Working paper

New York, NY: Graduate School of Business, Columbia

University

Chen, F., 2002 Information sharing and supply chain

coordination Working paper New York, NY: Graduate

School of Business, Columbia University

Chen, F and Yu, B., 2001a Quantifying the value of

lead-time information in a single location inventory system

Working paper New York, NY: Graduate School of

Business, Columbia University

Chen, F and Yu, B., 2001b A supply chain with asymmetric

capacity information Working paper New York, NY:

Graduate School of Business, Columbia University

Chen, F., et al., 2000 Quantifying the bullwhip effect in a

simple supply chain: the impact of forecasting, lead-time,

and information Management Science, 46 (3), 436–443

Chen, M., 2006 Coordinating demand fulfillment with

supply across a global supply chain [Dissertation of

Doctor of Philosophy degree.] College Park, MD:

University of Maryland

Chen, S and Chen, R., 1997 Manufacturer-supplier

relationship in a JIT environment Production and

Inventory Management Journal, 38 (1), 58–62

Chen, Q.X., Chen, X., and Lee, W.B., 2007 Qualitative

search algorithms for partner selection and task

alloca-tion in the formulaalloca-tion of virtual enterprise Internaalloca-tional

Journal of Computer Integrated Manufacturing, 20 (2/3),

115–126

Cheng, C.H., Yang, K.L., and Wang, C.L., 1999 Evaluation

attack helicopters by AHP based on linguistic variable

weights European Journal of Operational Research, 116,

423–435

Chiang, Z., 2005 A dynamic decision approach for

long-term vendor selection based on AHP and BSC In:

Proceedings of International Conference on Intelligent

Computing, ICIC 2005,Hefei, China, Part II, 257–265

Choi, T.Y and Hartley, J.L., 1996 An exploration of

supplier selection practices across the supply chain

Journal of Operations Management, 14, 333–343

Chopra, S and Meindl, P., 2001 Supply chain management:

strategy planning and operation Upper Saddle River, NJ:

Prentice Hall

Cook, R.L., 1992 Expert systems in purchasing applications

and development International Journal of Purchasing and

Management, 18, 20–27

Coyle, R.G., 1996 System dynamics modelling: a practical

approach London: Chapman and Hall

de Boer, L., van der Wegen, L., and Telgen, J., 1998

Outranking methods in support of supplier selection

European Journal of Purchasing and Supply Management,

4, 109–118

Dickson, G., 1966 An analysis of vendor selection systemsand decisions Journal of Purchasing, 2 (1), 5–17.Donaldson, B., 1994 Supplier selection criteria on the servicedimension European Journal of Purchasing and SupplyManagement, 1, 209–217

Dowlatshahi, S., 2000 Design–buyer–supplier interface:Theory versus practice International Journal of Produc-tion Economics, 63, 111–130

Duran, O and Aguilo, J., 2008 Computer-aided tool selection based on a fuzzy-AHP approach ExpertSystems with Applications, 34 (3), 1787–1794

Duran, O and Aguilo, J., 2007 Computer-aided tool selection based on a Fuzzy-AHP approach ExpertSystems with Applications, 34 (3), 1787–1794

machine-Ellram, L.M., 1992 The supplier selection decision instrategic partnerships International Journal of Purchas-ing and Management, 18, 8–14

ElMaraghy, H.A and Mahmoudi, N., 2009 Concurrentdesign of product modules structure and global supplychain configuration International Journal of ComputerIntegrated Manufacturing, 22 (6), 483–493

Fliedner, G., 2003 CPFR: an emerging supply chain tool.Industrial Management and Data Systems, 103, 14–21.Folie, M and Tiffin, J., 1976 Solution of a multi-productmanufacturing and distribution problem ManagementScience, 23, 286–296

Forrester, J.W., 1961 Industrial dynamics Cambridge, MA:MIT Press

Fulkerson, W., 2000 Information-based manufacturing inthe informational age The International Journal ofFlexible Manufacturing Systems, 12 (2/3), 31

Gavirneni, S., Kapuscin, R., and Tayur, S., 1999 Value ofinformation in capacitated supply chains ManagementScience, 46 (1), 16–24

Georgiadis, P., Vlachos, D., and Iakovou, E., 2005 A systemdynamics modelling framework for the strategic supplychain management of food chains Journal of FoodEngineering, 70, 351–364

Gereffi, G., Humphrey, J., and Sturgeon, T., 2005 Thegovernance of global value chains Review of Interna-tional Political Economy, 12 (1), 78–104

Gereffi, G and Korzeniewicz, M., 1994 Commodity chainsand global capitalism Westport, CT: Praeger

Ha, S.H and Hong, G.H., 2005 Selecting supply partnersfor e-collaboration in supply chains, challenges ofexpanding Internet: e-commerce, e-business, and e-government In: Proceedings of 5th IFIP Conference e-Commerce, e-Business, and e-Government (I3E’2005),October 28–30, 2005, Poznan, Poland

Hajji, A., et al., 2010 Production control and replenishmentstrategy with multiple suppliers European Journal ofOperational Research Doi:10.1016/j.ejor.2010.08.010.Handfield, R and Nichols, E., 1999 Introduction tosupply chain management Englewood Cliffs, NJ:Prentice Hall

Hodge, G., 2002 Enterprise resource planning in textiles.Journal of Textile and Apparel, Technology and Manage-ment, 2 (3), 1–8

Holweg, M and Bicheno, J., 2002 Supply chain simulation –

a tool for education, enhancement and endeavour.International Journal of Production Economics, 78, 163–175

Huang, G.Q., Lau, J.S.K., and Mak, K.L., 2003 Theimpacts of sharing production information on supplychain dynamics: a review of the literature InternationalJournal of Production Research, 41 (7), 1483–1517

Huang, X.G., et al., 2005 Critical-path-analysis-based

dynamic component supplier optimization International

Journal of Computer Integrated Manufacturing, 18 (8),

702–709

Huang, X.G., Wong, Y.S., and Wang, J.G., 2004 A

two-stage manufacturing partner selection framework for

virtual enterprises International Journal of Computer

Integrated Manufacturing, 17 (4), 294–304

Jain, V., Wadhwa, S., and Deshmukh, S.G., 2009 Enhancing

flexibility in supply chains: modelling random demands

and non-stationary supply information International

Journal of Computer Integrated Manufacturing, 22 (8),

812–822

Jharkharian, S and Shankar, R., 2007 Selection of logistics

service provider: an analytical network process (ANP)

approach The International Journal of Management

Science, 35, 274–289

Jia, Q., Guo, W., and Li, B., 2007 Study on the effect of

information sharing strategy to complex supply chain

system based on multi-dimension view by simulation In:

2007 International conference on wireless communications,

networking and mobile computing 4847–4850

Jones, C., 1998 Moving beyond ERP: making the missing

link Logistics Focus, 6 (7), 2–7

Kahraman, C., Ertay, T., and Buyukozkan, G., 2006 A

fuzzy optimization model for QFD planning process

using analytic network approach European Journal of

Operational Research, 171, 390–411

Kaplan, R and Norton, D., 1993 Putting the balanced

score-card to work Harvard Business Review, 71 (5), 134–147

Kaplan, R and Norton, D., 2004 Strategy maps: converting

intangible assets to tangible outcomes Boston, MA:

Harvard Business School Press

Kleijnen, J.P.C and Smits, M.T., 2003 Performance metrics

in supply chain management Journal of the Operational

Research Society, 54 (5), 507–514

Kleijnen, J.P.C., 2005 Supply chain simulation tools and

techniques: a survey International Journal of Simulation

and Process Modelling, 1 (1/2), 82–89

Larsen, N., et al., 1997 Temperature histories in liquid and

solid polar stratospheric cloud formation Journal of

Geophysics Research, 23 (102), 505–517

Lee, H and Whang, S., 2000 Information sharing in a

supply chain International Journal of Manufacturing

Technology and Management, 1 (1), 79–93

Leem, B., et al., 2007 Modelling the metrics for measuring

the performance on logistics centers by BSC and ANP in

Korean context In: Proceedings of Portland international

conference on management of engineering and technology,

2007 (PICMET 2007),Portland, OR, USA

Leung, L.C., Lam, K.C., and Cao, D., 2006 Implementing

the balanced scorecard using the analytic hierarchy

process and the analytic network process Journal of the

Operational Research Society, 57, 682–691

Lin, C.-W.R and Chen, H.Y.S., 2004 A fuzzy strategic

alliance selection framework for supply chain partnering

under limited evaluation resources Computers in

Indus-try, 55 (2), 159–179

Lin, L.-Z and Hsu, S.-H., 2008 The qualitative and

quantitative models for performance measurement

sys-tems: the agile service development Quality and

Quan-tity, 42 (4), 445–476

Liu, D., et al., 2006 A networked engineering portal to

support distributed supply chain partnership

Interna-tional Journal of Computer Integrated Manufacturing, 19

(2), 91–103

Mabert, V.A and Venkataramanan, M.A., 1998 Specialresearch focus on supply chain linkages: challenges fordesign and management in the 21-st century DecisionSciences, 29 (3), 537–552

Mahanti, R and Antony, J., 2005 Confluence of six sigma,simulation and software development Managerial Audit-ing Journal, 20 (7), 739–762

Mentzer, J.T., Min, S., and Zacharia, Z.G., 2000 The nature

of interfirm partnering in supply chain management.Journal of Retailing, 76 (4), 549–568

Mentzer, J.T., et al., 1999 What is supply chain management?Unpublished Working Paper Knoxville, TN: University

of Tennessee

Mertins, K., Rabe, M., and Ja¨kel, F.W., 2005 Distributedmodelling and simulation of supply chains InternationalJournal of Computer Integrated Manufacturing, 18 (5),342–349

Miragliotta, G and Staudacher, A.P., 2006 The grated use of information and stocks in the manage-ment of uncertain lumpy demand International Journal

inte-of Computer Integrated Manufacturing, 19 (3), 220–233

Miser, H.J and Quade, E.S., 1985 Handbook of tems analysis – overview of uses, procedures, applica-tion, and practice New York: Elsevier SciencePublishing

sys-Moser, R., 2007 Strategic purchasing and supply ment – a strategy-based selection of suppliers Wiesbaden,Germany: Springer

manage-Ni, M., Xu, X., and Deng, S., 2007 Extended QFD anddata-mining-based methods for supplier selection in masscustomization International Journal of Computer Inte-grated Manufacturing, 20 (2/3), 280–291

Niven, P.R., 2002 Balanced scorecard step-by-step: ing performance and maintaining results New York: JohnWiley & Sons

maximiz-Ozbayrak, M., Papadopoulou, T.C., and Akgun, M., 2007.System dynamics modelling of a manufacturing supplychain system Simulation Modelling Practice and Theory,

15, 1338–1355

Park, K.H., Kim, K.B., and Jung, B.J., 2003 Analysis of theperformance of supply chain partnership based oninformation sharing and lead-time distribution In: 2003spring conference of korean operations research andmanagement science society, Seoul, 342–345

Parrish, E.D., Cassill, N.L., and Oxenham, W., 2004.Opportunities in the international textile and apparelmarketplace niche markets JFMM, 8 (1), 41–57.Pidd, M., 2003 Tools for thinking – modelling in managementscience Chichester, UK: John Wiley & Sons

Pirard, F., Iassinovski, S., and Riane, F., 2008 A genericscalable simulation model for strategic supply chainmanagement with emphasis on production activities.International Journal of Computer Integrated Manufac-turing, 21 (4), 455–467

Polychronakis, Y.E and Syntetos, A.A., 2007 ‘Soft’ suppliermanagement related issues: An empirical investigation.International Journal of Production Economics, 106, 431–449

Rabelo, L., et al., 2007 Value chain analysis using hybridsimulation and AHP International Journal of ProductionEconomics, 105, 536–547

Ravi, V., Shankar, R., and Tiwari, M.K., 2005 Analyzingalternatives in reverse logistics for end-of-life computers:anp and balanced scorecard approach Computers andIndustrial Engineering, 48, 327–356

Rayport, J.F and Sviokla, J.J., 1995 Exploiting the virtual

value chain Harvard Business Review, 73 (6), 75–85

Reddy, A.M and Rajendran, C., 2005 A simulation study of

dynamic order-up-to policies in a supply chain with

non-stationary customer demand and information sharing

International Journal of Advanced Manufacturing

Tech-nology, 25 (9/10), 1029–1045

Richardson, G.P., 1996 Problems for the future of system

dynamics System Dynamics Review, 12 (2), 141–157

Saaty, T.L and Vargas, L.G., 2006 Decision making with the

analytic network process: economics, political, social and

technological applications with benefits, opportunities,

costs and risks New York: Springer Science & Business

Saaty, T.L., 1996 Decision making with dependence and

feedback: the analytic network process Pittsburgh, PA:

RWS

Sarmaha, S.P., Acharyaa, D., and Goyal, S.K., 2006 Buyer

vendor coordination models in supply chain

manage-ment European Journal of Operational Research, 175, 1–

15

Senge, P.W., 1992 The fifth discipline: the art and practice of

the learning organization Sydney: Random House

Shapiro, J.F., Singhal, V.M., and Wagner, S.N., 1993

Optimizing the value chain Interfaces, 23 (2), 102–117

Shin, K.H., et al., 2010 Performance trajectory-based

optimised supply chain dynamics International Journal

of Computer Integrated Manufacturing, 23 (1), 87–100

Shui-ying, Z and Rong-qiu, C., 2001 A decision model for

selecting participants in supply chain Journal of

Shang-hai University (English Edition), 5 (4), 341–344

Stadtler, H and Kilger, C., 2005 Supply chain management

and advanced planning: concepts, models, software and

case studies Berlin: Springer

Swaminathan, J.M., Smith, S.F., and Sadeh, N.M., 1998

Modelling supply chain dynamics: a multiagent

ap-proach Decision Sciences, 29 (3), 607–631

Sun, Z and Zhang, G., 2009 Beyond typologies of global

value chain governance: the accumulation of

technologi-cal capabilities International Journal of Business and

Management, 4 (1), 32–36

Swift, C.O., 1995 Preference for single sourcing and supplier

selection criteria Journal of Business Research, 32, 105–

111

Talluri, S., 2002 A buyer-seller game model for selection and

negotiation of purchasing bids European Journal of

Operational Research, 143, 171–180

Titus, S and Bro¨chner, J., 2005 Managing information flow

in construction supply chains Construction Innovation:

Information, Process, Management, 5, 71–82

Towill, D.R., 1997 The seamless supply chain–the predator’sstrategic advantage International Journal of TechnologyManagement, 13 (1), 37–56

Udin, Z.M., Khan, M.K., and Zairi, M., 2006 A tive supply chain management framework, Part 1:planning stage Business Process Management Journal,

Weber, C.A., Current, J.R., and Benton, W.C., 1991 Vendorselection criteria and methods European Journal ofOperational Research, 50, 2–18

Weber, C.A., Current, J.R., and Desai, A., 1998 cooperative negotiation strategies for vendor selection.European Journal of Operational Research, 108, 208–223.Wolstenholme, E.F., 1990 System enquiry Chichester, UK:John Wiley and Sons

Non-Wolstenholme, E.F., 1993 A case study in community careusing systems thinking Journal of the OperationalResearch Society, 44 (9), 925–934

Wolstenholme, E.F and Coyle, R.G., 1983 The ment of system dynamics as a methodology for systemdescription and qualitative analysis Journal of theOperational Research Society, 34 (1), 569–581

develop-Xue-zhen, Z.H.U., 2007 A dynamic model for vendorselection China-USA Business Review, 6 (2), 75–80.Yim, N.H., et al., 2004 Knowledge based decision making

on higher level strategic concerns: system dynamicsapproach Expert Systems with Application, 27, 143–158

Yu, X and Wang, F., 2007 Method of performanceevaluation based on ANP and BSC Asian Social Science,

3 (11), 43–47

Yu, Z., Yan, H., and Cheng, E., 2001 Benefits of informationsharing within supply chain partnerships IndustrialManagement and Data System, 101 (3), 114–119

Zadeh, L.A., 1965 Fuzzy sets Information and Control, 8,338–353

Zarandi, M.H.F., Tu¨rksen, I.B., and Saghiri, S., 2002.Supply chain: crisp and fuzzy aspects InternationalJournal of Applied Mathematics and Computer Sciences,

12 (3), 423–435

Zimmermann, H.-J., 1994 Fuzzy set theory and its tions London: Kluwer Academic

applica-The development of a new STEP-NC code generator (GEN-MILL)Yusri Yusof*, Noor Diana Kassim and Nurul Zakiah Zamri TanFaculty of Mechanical and Manufacturing Engineering, Universiti Tun Hussein Onn Malaysia (UTHM),

Batu Pahat, Johor, Malaysia(Received 11 February 2010; final version received 1 October 2010)The early NC machines and the current CNC utilise the same standard programming of G&M codes formalised asthe ISO 6893 standard to machine parts with complex shape in a precise manner However, the G&M code containsimplicit technical information and does not meet the requirements of modern NC technology as it uses low levelcodes to describe tool movements and instructions Literature shows that the current CNC is difficult to manage,modify and verify NC data easily ISO 14649 referred to as STEP-NC is the result of an international effort toachieve full interoperability and bi-directional information exchange throughout the manufacturing network STEP-

NC has been designed in such a way that information remains in its context and without reducing to primitiveinstructions such as G&M code This research will explore ISO 6983 (G&M code) together with ISO 14649 (STEP-NC) and utilise STEP-NC in overcoming the problems of G&M code and at the same time take advantage of currentadvances in computing and controllers A new STEP-NC code generator (GEN-MILL) which focuses on the millingprocess that is able to generate STEP-NC codes based on Example 1 in ISO 14649 Part 11 was developed Theadvantages of the proposed STEP-NC code generator (GEN-MILL) program was verified and evaluated bycomparing the output generated by the software with ISO 14649 Physical File for Example 1 The STEP-NC codegenerator (GEN-MILL) can understand and generate STEP-NC codes through a STEP-NC compliant interface andwill give industries and academicians more in-depth understanding and confidence to switch from G&M code toSTEP-NC

Keywords: STEP-NC; G&M code; ISO 14649; ISO 6983; CNC; NC

1 Introduction

Numerical Control (NC) is the term used to describe

the control of machine movements and various other

functions by instructions expressed as a series of

numbers and initiated via an electronic control system

When NC machines were developed, the purpose of

the NC machine was to machine parts with complex

shape in a precise manner Computerised Numerical

Control (CNC) is the term used when the control

system includes a computer Numerical control is

applied to a wide range of manufacturing processes

such as metal cutting, woodworking, welding, flame

cutting and sheet metal forming Numerical control is

economical for mass, batch and in many cases

single-item production Many factors contribute to this

economic viability, the most important of these being

(1) high productivity rates, (2) uniformity of the

product, (3) reduced component rejection, (4) reduced

tooling costs, (5) less operator involvement, and (6)

complex shapes machined easily

It is also the case that fewer employees were

required as conventional machines are replaced by

modern technology, but those that remain will of be

high-calibre technicians with considerable knowledge

of metal-cutting methods, cutting speeds and feeds,work-holding and tool-setting techniques and who arefamiliar with the control systems and programming fornumerical control Most CNC machines are pro-grammed in the ISO 6983 G and M code language.Programs are typically generated by computer-aidedmanufacturing (CAM) systems that use computeraided design (CAD) information However, ISO 6983limits program portability for three reasons First, thelanguage focuses on programming the tool centre pathwith respect to machine axes, rather than the machin-ing process with respect to the part Second, thestandard defines the syntax of program statements, but

in most cases leaves the semantics ambiguous Third,vendors usually supplement the language with exten-sions that are not covered in the limited scope of ISO

6983 (International Standard Organization 2002).The ISO 6983 standard focuses on programmingthe path of the cutter centre location (CL) with respect

to the machine axes, rather than the machining taskswith respect to the part Thus, ISO 6983 defines thesyntax of program statements, but in most cases leaves

*Corresponding author Email: yusri@uthm.edu.my

Vol 24, No 2, February 2011, 126–134

ISSN 0951-192X print/ISSN 1362-3052 online

Ó 2011 Taylor & Francis

DOI: 10.1080/0951192X.2010.531289

the semantics ambiguous, together with low-level

limited control over program execution These

pro-grams, when processed in a CAM system by a

machine-specific post-processor, become

machine-de-pendent (Xu and Newman 2006) The

above-men-tioned ISO 6983 is widely used as a standard of data

interface for a numerical control (NC) apparatus in a

manufacturing process However, the ISO 6983 is a

low-level international standard that just defines an

axis movement command and switching commands,

which cause problems to users and manufacturers as

more are required in the manufacturing floor Several

CNC manufacturers attempt to solve this problem by

adding standards high-level command code of their

own to the existing ISO 2983 standards However,

since manufacturers have their own standard,

compat-ibility of the part programs that differs is reduced In

addition, since the shop floor programming in

manu-facturing is using the basic G&M code as per what was

in the ISO 6983, various production information such

as machining features, machining process, machining

technology, cutting tools, machining knowledge and

feature information may not be included in the part

program or worst become lost

With the rapid advancement of information

tech-nology associated with NC techtech-nology, the

manufac-turing environment has changed significantly since the

last decade However, the low-level standard, G&M

codes, have been used for over 50 years as the interface

between CAM and CNC, and are now considered as

an obstacle for global, collaborative and intelligent

manufacturing STEP-NC has been developed for the

exchange of information between CAD/CAM systems

and NC controls (Maeder et al 2002), where a new

model of data transfer between CAD/CAM systems

and CNC machines, known as STEP-NC, is being

developed worldwide to replace G&M codes

STEP-NC (ISO14649) is a STEP-compliant data interface fornumerical control, a new standard that has beendeveloped and intended to replace the G-code.STEP-NC data model provides a hierarchical informa-tion data model for CNCs to support variousmanufacturing technologies such as milling or turningmachines (Brunnermeier and Martin 1999) The STEP-

NC interface, which is based on an object-orienteddata model, has been developed in several recentresearch projects in which many industrial companiesand universities have participated

Figure 1 shows how the design data is nicated to manufacturing in current practice Designcreates the specification for a product as a 3D model.Detailing decides the manufacturing requirements forthe product by making a drawing Path planninggenerates tools paths Manufacturing controls produc-tion The job of design is performed by using aComputer Aided Design (CAD) system, detailing isperformed by using a drawing Computer Aided Designand Drafting (CADD) system, path planning isperformed by using a Computer Aided Manufacturing(CAM) system, and manufacturing is controlled using

in response to changes in the available tooling, thecontrol cannot optimise the machining process for thecapabilities of the selected machine, and the operatorcannot rely on software in the control to check thesafety of the set-up and the program (Steptools 2009)

Figure 1 Current CAD/CAPP/CAM/CNC data flow

2 STEP-NC related work

A solution to the drawbacks of ISO 6983, a new

standard called ISO 14649, has been introduced The

ISO 14649 defines a method for incorporating a variety

of production information in a new type part program

that is different from the previous one using the G&M

code Accordingly, a system adopting the ISO 14649,

or rather known as STEP-NC, should and would

exhibit a highly improved compatibility with upcoming

data, technology and other systems Besides that, the

part program for use in the STEP-NC would contain a

variety of production information such as machining

features, machining process, machining method,

cut-ting tools, machining technology and geometry

in-formation, which was not available in ISO 6983

A major benefit of ISO 14649 is its use of the

existing data models from ISO 10303 As ISO 14649

provides a comprehensive model of the manufacturing

process, it can also be used as the basis for a bi- and

multidirectional data exchange between all other

information technology systems (International

Stan-dard Organization 2002)

In the new method of STEP-NC, enterprises can

continue to use their existing systems for CAD, CADD

and CAM, but the end result is sent to the CNC as a

STEP-NC file instead of a G&M code file Figure 2

shows the modified pipeline where only a small changeneeds to be done particularly on the interface, but theadvantages are significant

2.1 Comparisons between G&M code and STEP-NCBased on various literature reviews published, theadvantages of STEP-NC as compared to the currentG&M code can be summarised Table 1 gives asummary of how STEP-NC steps up as remedies forthe shortcomings of the G&M codes STEP-NC is ahigh level code as well as contains much moreinformation as compared to G&M code It supportscoding of complex geometries with manageablechanges features, whereas G&M code was structuredfor simple geometries with difficult changes manage-ability Data movement in STEP-NC is bidirectionalinstead of just one direction only as per what G&Mcode does

2.2 Previous researchRecently, a number of projects involving the areas ofSTEP-NC based interoperability and research anddevelopment for various CNC manufacturing pro-cesses have been started (Yusof 2009) Systems related

to STEP compliance have been developed by academia

Table 1 Comparison between G&M code and STEP-NC

Figure 2 STEP-NC CAD/CAPP/CAM/CNC data flow

all over the world (Yusof et al 2009) The literature

shows that most of the systems developed are turning

and milling system There are few of the lathe systems

and shop floor jobs that have been studied Their

achievement is beyond what would have been thought

The ideas to create STEP-NC generator are many but

the basic concepts used are fundamentally the same

(Zamri Tan et al 2009) Table 2 shows the system

together with the programming language that is being

used to develop it

Most of the previous researchers use Visual Cþþ

and Java as their development tools The software

STEPTurn uses STEP AP203 as its inputs STEPTurn

is a CAPP system bridging the gap between CAD and

CAM (Heusinger et al 2006) TurnSTEP which was

developed in 2006 uses STEP-AP as inputs TurnSTEP

is claimed by Choi to be fully compliant with ISO

14649 and suitable for e-manufacturing (Choi et al

2006) G2STEP which is the latest system to cover the

machine functioning from pre-processor to STEP-NC

part program generation including part program

verification (Shin et al 2007) generates STEP-NC

part program as output with G code as input Details

of information on STEP-NC previous research can be

referred in Table 2

3 Software development

3.1 Methodology summary

The novelty of this research was that it visualised the

data transition from CAD/CAM to the STEP-NC

software by having some of the input data keyed in

manually and having the ISO 14649 Physical file for

Example 1 as the output Verification was done by

comparing the output generated by the software with

Physical File of Example 1 in ISO 14649 Most of the

STEP-NC prototype or system developed accepted

another file as its input Those files were transferred or

uploaded to the system directly which is good as no data

modification or loss happened during transfer; thus, it is

very suitable for advanced stages of STEP-NC studies

For early stage studies of STEP-NC, the proposed

GEN-MILL would be easier for understanding andlearning of the STEP-NC concept, as data flows caneasily be visualised from the system itself GEN-MILLsimplified the visualisation of data flow inside STEP-

NC, and users can fiddle freely with the inputs to seewhere the difference appears in the output Theapproach for system development was also different asmost of the other researchers use Visual Cþþ as theirdevelopment tools but GEN-MILL uses Visual Basic asthe development tools The overall research methodol-ogy will follow four important phases The first phase is adetailed study on the literature and standards on ISO

14649 (STEP-NC) and ISO 6983 (G&M code) Duringthis phase a system study on the current NC condition isbeing looked into and a comparison between the currentprogramming condition and the future one is beingmade, based on expectations raised The results of theliterature and standards study will initiate to the firstphase of programming and the development of aprototyped STEP-NC software The final phase of thewhole research will be the presentation of the developedmodel and STEP-NC code generator (GEN-MILL).The research methodology summary can be visualised as

in Figure 3

Software development methodology is referring tothe documented collection of guidelines, proceduresand standards intended to ensure the development ofquality application systems that meet the system’srequirements in an efficient manner The systemdevelopment methodology involves a series of opera-tions and procedures that are used to develop anapplication or system In this section, the type ofdevelopment process model to be used for the STEP-

NC generator (GEN-MILL) development will bediscussed A prototype is a toy implementation of asystem; usually exhibiting limited functional capabil-ities, low reliability, and inefficient performance Animportant purpose is to illustrate the input dataformats, messages, reports and the interactive dialogs

to the end user This is a valuable mechanism forgaining better understanding of the end user’s needs.Another important use of the prototyping model is

Table 2 Research on STEP-NC code generation

that it helps critically examine the STEP-NC or G&M

code technical issues associated with the product

development Small-scale mock-ups of the system are

developed following an iterative modification process

until the prototype evolves to meet the end users’

requirements While most prototypes are developed

with the expectation that they will be discarded, it is

possible in some cases to evolve from prototype to

working system A basic understanding of the

funda-mental operational problem is necessary to avoid

solving the wrong problem

The software prototyping development model is the

most suitable methodology to develop STEP-NC code

generator (GEN-MILL) It is more manageable as

opposed to the traditional waterfall model This

approach is broken down into eight development

phases of software prototyping They are requirement

gathering, quick design, prototype build, refine

re-quirements, prototype evaluation, design,

implementa-tion, test and maintain There is a looping iteration in

the quick design of prototype and the refine

require-ments stages Before the final design, system prototype

will be tested, evaluated and modified based on the

current inputs and requirements, and refinement of

requirements can be done whenever there are new

functional capabilities are added by the control in

order to improve the functionalities of the application

The reason why the software prototyping model waschosen was because it helps to identify easily confusing

or difficult functions and missing functionality as well

as encourages innovation and flexible designs

3.2 STEP-NC requirements gathering

In order to study the requirement of STEP-NC forsystem development, the content of STEP-NC code inExample 1 ISO 14649 Part 11 will be studied Forreview purposes STEP-NC code study for the firstthree lines in Example 1 ISO 14649 Part 11 will beshown and visualised as shown in Figure 4 The firstthree top entities in Example 1 ISO 14649 arePROJECT, WORKPLAN and WORKPIECE In thefirst three lines, these three entities act as main entities,but inside these entities itself; it was shown that itcontains another entity or entities as its sub-entities.Identify (MAIN) Mandatory entities involved:(1) - PROJECT

(2) - WORKPLAN(3) - WORKPIECEFor example, for main entity PROJECT, the entityWORKPLAN and WORKPIECE act as its sub-entities When referred to, the entity WORKPLAN

Figure 4 STEP-NC code for the first three lines from Example 1 ISO 14649 Part 11

Figure 3 Research methodology summary

and WORKPIECE also act as main entity This means

that an entity can play the part as a main entity as well

as sub-entity to another entity Based on Figure 4, it is

explained that, the main entity PROJECT has three

known sub-entities, which are, IDENTIFIER,

WORKPLAN and WORKPIECE, as for the main

entity WORKPIECE, it has four known sub-entities,

which are IDENTIFIER, MATERIAL, SHAPE

TOLERANCE and list of CARTESIAN POINT

Identify entities that act as subtype to MAIN

Mandatory entities

From PROJECT - WORKPLAN, WORKPIECE

From WORKPLAN - MACHINING_

WORKINGSTEP, SETUP

From WORKPIECE - MATERIAL,

CARTESIAN_POINT

The prototype system developed will focus first on

identifying the input that is required to come out with

a STEP-NC code which is identical with Example 1

Initial prototype development requirement will

pro-ceed on a list of inputs collected from the STEP-NC

code study Programming language chosen to develop

STEP-NC code generator GEN-MILL will be Visual

Basic, because it is easily learned and used by beginner

programmers The language not only allows

program-mers to create simple GUI applications but can also

develop complex applications Programming in VB is a

combination of visually arranging components or

controls on a form, specifying attributes and actions

of those components, and writing additional lines of

code for more functionality Since default attributes

and actions are defined for the components, a simple

program can be created without the programmer

having to write many lines of code

3.3 Identify input data between the required scope

In order to organise and identify variables needed for

the prototype system development, detailed lists of

STEP-NC entities and sub-entities involved in

Exam-ple 1 ISO 14649 Part 11 for round hole will be listed

and extracted from the manual provided by ISO

Based on Tables 3 and 4, the system will identify the

SUBENTITY column as the variable and the MARKS column will be the description of thevariables

RE-ENTITY: PROJECTENTITY: WORKPIECE

3.4 Relationship between round hole entitiesAll the entities that are involved in the production ofround hole in Example 1 ISO 14649 Part 11 weregathered and a detailed study on each entity was done.There were cases where a sub-entity of an entity wasanother or referred to another entity A relationshipexisted between most of the entities that resembled aparent to child relationship in system or databasedevelopment which can be seen in Figure 5

3.5 Quick design of STEP-NC input screen interfaceFigure 6 represents the input screen interface that wasdeveloped for the prototype system After studying theSTEP-NC code for round hole that are listed inExample 1 ISO 14649 Part 11, the required inputs areclassified and categorised into four main groups, whichare the main section, round hole section, drillingsection and reaming section For the initial prototype,lots of inputs were required from the user side

3.6 Build STEP-NC screen prototypeBased on the inputs provided, a prototype system thatcan manipulate the inputs to form a full STEP-NCcode for round hole, identical to the one provided byExample 1 ISO 14649 Part 11 was developed Thecoding structure that points one entity to another in arecursive way was structured Figure 7 is the outputproduced by the prototype system

3.7 Evaluation of the prototypeThe prototype system developed did provide an insight

of what the systems required that was difficult tounderstand just by reading the standards provided byISO The data or entity flow from one point to another

is much more traceable as the prototype provides avirtual presentation of how the data flows from onepoint to another and where does it end, as well as what

is required by the data Lots of inputs were required to

be keyed in by the users in the prototype stage Thisshould not be so, as a good system would only requirelittle inputs by users The possibility that one commonentity was referred to or pointed by many differententities was also taken into consideration It could be aone to many relationships, many to one relationship or

Table 3 Sub-entity details for entity project

its_workpieces SET[0:?] of WORKPIECE

Figure 6 Input from the prototype system.

Figure 7 Output from the prototype system