International journal of computer integrated manufacturing , tập 23, số 6, 2010

Keywords: grid; collaborative virtual assembly; complex product; real-time collaborative simulation Notation CDM collision detection model CVA collaborative virtual assembly CVAE collabo

Sources of variability in the set-up of an indoor GPSCarlo Ferria*, Luca Mastrogiacomoband Julian Farawayca

Via XI Febbraio 40, 24060 Castelli Calepio, BG, Italy;bDISPEA, Politecnico di Torino, Corso Duca degli Abruzzi 24, Torino

10129, Italy;cDepartment of Mathematical Sciences, University of Bath, Bath BA2 7AY, UK

(Received 17 June 2009; final version received 5 January 2010)

An increasing demand for an extended flexibility to model types and production volumes in the manufacture oflarge-size assemblies has generated a growing interest in the reduction of jigs and fixtures deployment duringassembly operations A key factor enabling and sustaining this reduction is the constantly expanding availability ofinstruments for dimensional measurement of large-size products However, the increasing complexity of thesemeasurement systems and their set-up procedures may hinder the final users in their effort to assess whether theperformance of these instruments is adequate for pre-specified inspection tasks In this paper, mixed-effects andfixed-effects linear statistical models are proposed as a tool to assess quantitatively the effect of set-up procedures onthe uncertainty of measurement results This approach is demonstrated on a Metris Indoor GPS system (iGPS) Themain conclusion is that more than 99% of the variability in the considered measurements is accounted for by thenumber of points used in the bundle adjustment procedure during the set-up phase Also, different regions of theworkspace have significantly different error standard deviations and a significant effect on the transient duration ofmeasurement This is expected to affect adversely the precision and unbiasedness of measurements taken with IndoorGPS when tracking moving objects

Keywords: large scale metrology; large volume metrology; distributed coordinate measuring systems; Indoor GPS;iGPS; uncertainty

1 Introduction

During the last decades research efforts in

coordinate-measuring systems for large-size objects have led to a

broadening of the range of instruments commercially

available (cf Estler et al 2002)

These coordinate measurement instruments can be

grouped into two categories: centralised and

distrib-uted systems (Maisano et al 2008)

A centralised instrument is a measuring system

con-stituted by a single hardware element that in

perform-ing a measurement may require one or more ancillary

devices such as, typically, a computer An example of a

centralised instrument is a laser tracker that makes

use of a spherically-mounted reflector (SMR) to take

a measurement of point spatial coordinates and that

needs to be connected to a monitor of environmental

conditions and to a computer

A distributed instrument is a collection of separate

independent elements whose separately gathered

mea-surement information needs to be jointly processed in

order for the system to determine the coordinates of a

point A single element of the system typically cannot

provide measurements of the coordinates of a point

when standing alone Precursors of these apparatuses

can be identified in wireless indoor networks of sensors

for automatic detection of object location (cf Liu

et al 2007) These networks can be deployed for pection tasks in manufacturing operations once theirtrueness has been increased The term trueness isdefined in BS ISO 5725-1:1994 as ‘the closeness ofagreement between the average value obtained from alarge series of test results and an accepted referencevalue’ (Section 3.7)

ins-When inspecting parts and assemblies having largedimensions, it is often more practical or convenient

to bring the measuring system to the part rather thanvice versa, as is typically the case on a smaller scale.Therefore, instruments for the inspection of large sizeobjects are usually portable In performing a measure-ment task, a single centralised instrument, say a lasertracker, can then be deployed in a number of differentpostions which can also be referred to as stations Bymeasuring some fixed points when changing station,the work envelope of the instrument can be signifi-cantly enlarged enabling a single centralised instru-ment to be used for inspection of parts significantlylarger than its original work envelope To illustrate thisconcept, in Figure 1(a) the top view of three geome-trical solids, a cylinder, a cube and an octahedron(specifically a hexagonal prism) is displayed These

*Corresponding author Email: info@carlo.comyr.com

International Journal of Computer Integrated Manufacturing

Vol 23, No 6, June 2010, 487–499

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

Ó 2010 Taylor & Francis

DOI: 10.1080/09511921003642147

solids are inspected by a single centralised instrument

such as a laser tracker, which is moved across different

positions (1, 2, , 6 in the figure) from each of

which the coordinates of the points P1, P2 and P3 are

also measured In this respect, a single centralised

system appears therefore comparable with a

distri-buted system, whose inherent multi-element nature

enables work envelopes of any size to be covered,

provided that a sufficient number of elements are

chosen This characteristic of a measuring system of

adapting itself to suit the scale of a measuring task is

often referred to as scalability (cf Liu et al 2007) The

concept above can therefore be synthesised by saying

that a centralised system is essentially scalable in virtue

of its portability, whereas a distributed system is such

due to its intrinsic modularity

With a single centralised instrument, measurement

tasks within a working envelope, however extended,

cannot be performed concurrently but only serially

Each measurement task to be performed at a certain

instant in time needs a dedicated centralised

instru-ment This is shown in Figure 1(a) where the cylinder is

measured at the current instant with the instrument in

position 2, whereas the hexagonal prism is going to be

measured in a future instant when the instrument will

be placed in position 3 With a distributed system this

limitation does not hold With a distributed system,

concurrent measurement tasks can be performed

prov-ided that each of the concurrent tasks has a sensor or

subgroup of sensors dedicated to it at a specific instant

within the distributed instrument In Figure 1(b), the

same three objects considered in the case of a

cen-tralised instrument are concurrently inspected using a

distributed system constituted by six signal transmitter

elements (1, 2, , 6) and three probes, each carrying

two signal receiving elements whereby the coordinates

of the probe tips are calculated

This characteristic of distributed systems is cially advantageous when concurrently tracking theposition of multiple large-size components duringassembly operations The sole way of performing thesame concurrent operation with a centralised systemwould require the availability and use of more than asingle centralised instrument (laser tracker, for in-stance), with potentially-detrimental economic conse-quences on the manufacturing organisation in terms ofincreased fixed assets, maintenance costs and increasedcomplexity of the logistics

espe-A number of different distributed systems havebeen developed recently, some as prototypes for re-search activities (cf., for instance, Priyantha et al 2000;Piontek et al 2007), some others with a level ofmaturity sufficient for them to be made commerciallyavailable (cf., for instance, Welch et al 2001; Maisano

et al 2008) In this second case, the protection ofintellectual property (IP) rights prevents users’ trans-parent access to the details of the internal mechanismsand of the software implemented in the systems Thismay constitute a barrier to a full characterisation ofthe performance of the equipment This investigationendeavours to provide better insight into the perfor-mance of such systems by using widespread statisticaltechniques The main objective is therefore not tocriticise or evaluate the specific instrument consideredthereafter, but to demonstrate the use of techniquesthat may be beneficially deployed also on otherdistributed systems In particular, the effect of discre-tionary set-up parameters on the variability andstability of the measurement results has been analysed

In the next section the main characteristics ofthe Metris iGPS, which is the instrument considered,are described A cone-based mathematical model ofthe system is then presented in Section 3 The experi-mental set-up is described in Section 4 and the results

of the tests are analysed in Section 5 Conclusions are

drawn thereafter

2 Physical description of the instrument

The instrument used in this study is the iGPS (alias

indoor GPS) manufactured by Metris The description

of such a system provided in this section is derived

from publicly available information

The elements constituting the system are a set of

two or more transmitters, a number of wireless sensors

(receivers) and an unit controlling the overall system

and processing the data (Hedges et al 2003; Maisano

et al 2008)

Transmitters are placed in fixed locations within

the volume where measurement tasks are performed

Such a volume is also referred to as a workspace

Each transmitter has a head rotating at a constant

angular velocity, which is different for each

transmit-ter, and radiates three light signals: two infrared

fan-shaped laser beams generated by the rotating head,

and one infrared strobe signal generated by light

emitting diodes (LEDs) The LEDs flash at constant

time intervals ideally in all directions, but practically in

a multitude of directions Each of these time intervals

is equal to the period of revolution of the rotating head

on which the LEDs are mounted For any complete

revolution of the rotating head a single flash is emitted

virtually in all directions In this way, the LED signals

received by a generic sensor from a transmitter

constitute a periodic train of pulses in the time domain

where each pulse is symmetric (cf Hedges et al 2003,

column 6)

The rotating fan-shaped laser beams are tilted by

two pre-specified opposite angles, f1and f2(e.g 730

and 308, respectively) from the axis of rotation of the

head These angles are also referred to as slant angles.The fact that the angular velocity of the head isdifferent for different transmitters enables each trans-mitter to be distinguished (Sae-Hau 2003) A schematicrepresentation of a transmitter at the instant t1 whenthe first fanned beam L1 intersects the sensor inposition P and at the instant t2 when the second fannedbeam L2 passes through P is shown in Figure 2, wheretwo values for the slant angles are also shown Ideally,the shape of each of the fanned beams should beadjustable to adapt to the characteristics of the mea-surement tasks within a workspace Although twobeams are usually mounted on a rotating head, confi-gurations with four beams per head have also beenreported (Hedges et al 2003, column 5) To differ-entiate between the two fanned beams on a transmit-ter, their time position relative to the strobe signal isconsidered (see Figure 2)

The fanned beams are often reported as planar(Liu et al 2008; Maisano et al 2008), as depicted inFigure 2 Yet, the same beams when emitted fromthe source typically have a conical shape that is firstdeformed into a column via a collimating lens andthen into a fan-shape via a fanning lens (Hedges et al

2003, column 6) It is believed that only an ideal chain

of deformations would transform completely and fectly the initial conical shape into a plane For thesereasons, the final shape of the beam is believed topreserve traces of the initial shape and to be moreaccurately modelled with a portion of a conical sur-face, rather than a plane Each of the two conicalsurfaces is then represented by a vector, called a conevector, that is directed from the apex to the centre ofthe circular directrix of the cone The angle between acone vector and any of the generatrices on the conesurface is called the cone central angle This angle is

International Journal of Computer Integrated Manufacturing 489

designated by a1 and a2 for the first and the second

beams, respectively The apex of each cone lies on the

axis of rotation of the spinning head In Figure 3, a

schema of the portion of the conical surface

represent-ing a rotatrepresent-ing laser beam is displayed In this figure,

two portions of conical surfaces are shown to

illus-trate a2 and f2 (f24 0, having established

counter-clockwise angle measurements around the x-axis as

positive)

The angular separation between the optical axes of

the two laser modules in the rotating head is denoted

by yoff, when observed from the direction of the

rota-tional axis of the spinning head The rotation of the

head causes each of the cone surfaces, and therefore

their cone vectors, to revolve around the same axis

The angular position of the cone vector at a generic

instant is denoted by y1(t) and y2(t) for the first and

second fanned beams, respectively These angles are

also referred to as scan angles and are defined relative

to the strobe LED synchronisation signal, as illustrated

below

Wireless sensors are made of one or more

photo-detectors and a wireless connection to the controlling

unit for the transmission of the positional information

to the central controlling unit The use of the

photo-detectors enables the conversion of a received signal

(stroboscopic LED, first fanned laser, second fanned

laser) into the instant of time of its arrival (t0, t1 and t2

in Figure 2) The time intervals between these instants

can then be converted into measurements of scan

angles from the knowledge of the angular velocity of

the head for each transmitter (o in Figure 2) It is

expected that y1¼ o 6 (t17t0) and that y2¼ o 6

(t27t0) At the instant t0 when the LED signal reaches

the generic position P, the same LED signal also

flashes in any direction Therefore, at the very same

instant t0, the LED fires also in the reference directionwhere the angles in the plane of rotation are measuredfrom (i.e y1¼ y2¼ 0)

In this study, any plane orthogonal to the axis ofrotation is referred to as a plane of rotation For anyspherical coordinate system having the rotational axis

of the transmitter as the z-axis and the apex common

to the aforementioned conical surfaces as the origin,the angle y1swept by the cone vector of the first fannedbeam in the time interval t17t0 is connected with theazimuth of P measured from any possible referencedirection x established in the xy-plane, which is theplane of rotation passing through the common apex ofthe conical surfaces

From a qualitative point of view, the elevation(or the zenith) of P can be related to the quantity

o 6 (t27t1) By analogy with Figure 2, it is arguedthat, also in the case of conical fanned shaped beams,when the elevation (or zenith) of P is increasing(decreasing), the time interval t27t1 is also increasing.Vice versa, the reason why a time interval t27t1 islarger than another can only be found in the fact thatthe position of the sensor in the first case has a higherelevation than in the second

In the most typical configuration, two receivers aremounted on a wand or a bar in calibrated positions Atip of the wand constitutes the point for which thelocation is calculated based on the signals received bythe two sensors When the receivers are mounted on abar, the bar is then often referred to as vector bar Ifsuch a receivers-mounted bar is short, say with a lengthbetween 100 and 200 mm, it is then called a mini vectorbar These devices are equipped with firmware pro-viding processing capabilities The firmware enablesthe computation of azimuth and elevation of the wand

or bar tip for each of the spherical reference systemsassociated with each of the transmitters in the system.This firmware is called a position computation engine(PCE)

A vector bar therefore acts as a mobile ment for probing points as shown in the schema ofFigure 1(b) More recently, receiving instruments withfour sensors have been developed, enabling the user toidentify both the position of the tip and the orientation

instru-of the receiving instrument itself

3 The role of the bundle adjustment algorithms in theindoor GPS

The computation of the azimuth and elevation ofthe generic position P in the spherical reference system

of a generic transmitter enables the direction of theoriented straight line l from the origin (the apex of thecones) to P to be identified However, it is not possible

to determine the location of P on l In other words, it is

not possible to determine the distance of P from the

origin Therefore, at least a second transmitter is

nece-ssary to estimate the position of P in a user arbitrarily

predefined reference system {Uref} In fact, assuming

that the position and orientation of the ith and jth

transmitters in {Uref} are known, then the coordinates

of the generic point on liand on ljcan be transformed

from the spherical reference system of the transmitters

to the common reference system {Uref} (cf Section 2.3

in Craig 1986) Then, P can be estimated with some

nonlinear least squares procedure, which minimises the

sum of the squared distances between the estimates of

the coordinates of P in {Uref} and the generic point on

li and lj Only in an ideal situation would li and lj

intersect In any measurement result, the azimuth and

elevation are only known with uncertainty (cf Sections

2.2 and 3.1 in JCGM 2008) Very little likelihood exists

that these measured values for li and lj coincide with

the ‘true’ unknown measurands The same very little

likelihood applies therefore to the existence of an

intersection between liand lj When adding a third kth

transmitter, qualitative geometrical intuition supports

the idea that the distances of the optimal P from each

of the lines li, ljand lkare likely to be less variable until

approaching and stabilising around a limit that can be

considered typical for the measurement technology

under investigation Increasing the number of

trans-mitters is therefore expected to reduce the variability of

the residuals The estimation of the coordinates of P,

when the position of the transmitters is known, is often

referred to as a triangulation problem (Hartley and

Sturm 1997; Savvides et al 2001)

If the position and orientation of the transmitters

in {Uref} are not known, then they need to be

deter-mined before the actual usage of the measurement

system To identify the position and orientation of a

transmitter in {Uref}, six additional parameters need

to be estimated (cf Section 2.2 in Craig 1986) This

more general engineering problem is often referred to

as three-dimensional (3D) reconstruction and occurs in

areas as diverse as surveying networks (Wolf and

Ghilani 1997), photogrammetry and computer vision

(Triggs et al 2000; Lourakis and Argyros 2009) The

estimation of three-dimensional point coordinates

to-gether with transmitter positions and orientations to

obtain a reconstruction which is optimal under a

pre-specified objective function and an assumed errors

structure is called bundle adjustment (BA) The

objec-tive or cost function describes the fitting of a

mathe-matical model for measurement procedure to the

experimental measurement data Most often, but not

necessarily, this results in minimising the sum of the

squares of the deviations of the measurement data

from their values predicted with nonlinear functions of

the unknown parameters (Triggs et al 2000; Lourakis

and Argyros 2009) A range of general purpose misation algorithms, such as for instance those ofGauss–Netwon and Levenberg–Marquardt, can beused to minimise the nonlinear objective function.Alternatively, significantly increased efficiency can begained if these algorithms are adjusted to account forthe sparsity of the matrices arising in the mathematicaldescription of 3D reconstruction problems (Lourakisand Argyros 2009)

opti-In the measurement system investigated, a BAalgorithm is run in a set-up phase whereby the posi-tion and orientation of each transmitter in {Uref} aredetermined Therefore, during the subsequent deploy-ment of the system (measuring phase), the coordinates

of a point are calculated using the triangulationmethods mentioned above

However, as is typically encountered in commercialmeasurement systems, the BA algorithms implemented

in the system are not disclosed completely to the users.This makes it difficult for both users and researchers todevise analytical methods to assess the effects of thesealgorithms on the measuring system In this investiga-tion, consideration is given to experimental design andstatistical techniques to estimate the effect that deci-sions taken when running the built-in BA algorithmexert on measurement results

4 Experimental set-upFour transmitters were mounted on tripods and placed

at a height of about two metres from floor level Thedirection of the rotational axis of each transmitterspinning head was approximately vertical Each of thefour transmitters was placed at the corners of anapproximate square of side about eight metres

A series of six different targets fields labelled I, II,III, IV, V and VI and respectively consisting of 8, 9, 10,

11, 12 and 13 targets was considered during the BAprocedure Each of these fields was obtained by addingone target to the previous field, so that the first eighttargets are common to all the fields, the first ninetargets are common to the last five fields and so on Aschema of this experimental configuration is shown inFigure 4

All the fields were about 1.2 m above floor level.The target positions were identified using an isostaticsupport mounted on a tripod which was moved acrossthe workspace A set of the same isostatic supports wasalso available on a carbon-fibre bar that was used toprovide the BA algorithm built in the system with arequested measurement of length (i.e to scale thesystem) A distance of 1750 mm between two isostaticsupports on the carbon-fibre bar was measured on acoordinate-measuring machine (CMM) The carbon-fibre bar was then placed in the central region ofInternational Journal of Computer Integrated Manufacturing 491

the workspace The coordinates of the two targets

1750 mm apart were measured with iGPS and their

1750 mm distance was used to scale the system in all

the target fields considered In this way, the scaling

procedure is not expected to contribute to the

varia-bility of the measurement results even when different

target fields are used in the BA procedure Figure 5

shows an end of the vector bar used in this set-up (the

large sphere in the figure), while coupled with an static support (the three small spheres) during themeasurement of a target position on the carbon-fibrebar

iso-The BA algorithm was run on each of these sixtargets fields so that six different numerical descrip-tions of the same physical positions and orientations ofthe transmitters were obtained

Six new targets locations were then identified usingthe isostatic supports on the carbon-fibre bar men-tioned above Using the output of the BA executions,the spatial coordinates of these new target locationswere measured The approximate position of the sixtargets relative to the transmitters is shown in theschema of Figure 6

Each target measurement consisted in placing thevector bar in the corresponding isostatic support andholding it for about 30 s This enabled the measure-ment system to collect and store about 1200 records oftarget coordinates in {Uref} for each of the six targets

In this way, however, the number of records for eachtarget is different, owing to the human impossibility ofmanually performing the measurement procedure with

a degree of time control sufficient to prevent thissituation occurring

5 ResultsEach of the six target positions displayed in Figure 6and labelled 1, 2, 3, 4, 5, 6 was measured using each ofthe six BA set-ups I, II, , VI, giving rise to a

grouping structure of 36 measurement conditions

(cells)

When measuring a target location, its three

Carte-sian coordinates in {Uref} are obtained To reduce the

complexity of the analysis from three-dimensional to

mono-dimensional, instead of these coordinates the

distance of the targets from the origin of {Uref} is

considered Central to this investigation is the

estima-tion of the effect on the target–origin distance due to

the choice of a different number of target points when

running the BA algorithm The target locations 1,

2, , 6 do not identify points on a spherical surface,

so they are at different distances from the origin of

{Uref}, regardless of any possible choice of such a

reference system These target locations therefore

contribute to the variability of the measurements of

the target–origin distance whereby the detection of a

potential contribution of the BA set-ups to the same

variability can be hindered To counteract this masking

effect, the experiment was carried out by selecting first

a target location and then randomly assigning all the

BA set-ups for that location to the sequence of tests

This was repeated for all the six target positions Such

an experimental strategy introduces a constraint to a

completely random assignment of the 36 measurement

conditions to the the run order In the literature (cf

Chapters 27, 16 and 8 in Neter et al 1996, Faraway

2005 and Faraway 2006, respectively), this strategy is

referred to as randomised complete block design

(RCBD) The positions of the targets 1, 2, , 6

constitutes a blocking factor identifying an

experi-mental unit or block, within which the BA set-ups are

tested The BA set-ups I, II, , VI constitute a

random sample of all the possible set-ups that differ

only in the choice of the location and number of points

selected when running the BA algorithm during the

system set-up phase On the other hand, the analysis

of the obvious contribution to the variability of the

origin–target distance when changing the location of

the targets would not add any interesting information

to this investigation These considerations lead to

de-scribing the experimental data of the RCBD with a

linear mixed-effects statistical model, which is first

defined and then fitted to the experimental data

5.1 Mixed-effects models

The distance dijof the i th (i¼ 1, , 6) target from

the origin measured when using the jth (j ¼ I, , VI)

BA procedure is modelled as the sum of four

contributions: a general mean m, a fixed effect ti due

to the selection of the i th target point, a random effect

bj due to the assignment of the jth BA set-up and a

random error eijdue to all those sources of variability

inherent in any experimental investigation that is not

possible or convenient to control This is described bythe equation

dij¼ m þ tiþ bjþ eij: ð1Þ

In Equation (1) and hereafter, the Greek symbols areparameters to be estimated and the Latin symbols arerandom variables In particular, the bj’s have zeromean and standard deviation sb; the eij’s have zeromean and standard deviation s The eij’s are assumed

to be made of independent random variables normallydistributed, i.e eij* N(0,s2) The same applies to the

bj’s, namely bj* N(0,s2b) The eij’s and the bj’s are alsoassumed to be independent of each other Under theseassumptions, the variance of dij, namely s2

d, is given bythe equation

s2d¼ s2bþ s2: ð2ÞUsing the terminology of the ‘Guide to the expression

of uncertainty in measurement’ (cf Definition 2.3 inJCGM 2008), sd is the standard uncertainty of theresult of the measurement of the origin–targetdistance

As pointed out in the previous section, the number

of the determinations of the target–origin distance thathave been recorded is different for each of the 36measurement conditions For simplicity of the analy-sis, the number of samples gathered in each of theseconditions has been made equal by neglecting thesamples in excess of the original minimum sample sizeover all the cells This resulted in considering 970observations in each cell The measurement resultprovided by the instrument in each of these conditionsand used as a realisation of the response variable dijinEquation (1) is then defined as the sample mean ofthese 970 observations There is a single measurementresult in each of the 36 cells The parameters of themodel, i.e m, ti, sband s, have been estimated by therestricted maximum likelihood (REML) method asimplemented in the lme() function of the packagenlme of the free software environment for statisticalcomputing and graphics called R (cf R DevelopmentCore Team 2009) More details about the REMLmethod and the package nlme are presented inPinheiro and Bates (2000) The RCBD assumes thatthere is no interaction between the block factor (targetlocations) and the treatment (BA set-up) Thishypothesis is necessary so that the variability within

a cell represented by the variance s2 of the randomerrors can be estimated when only one experimentalresult is present in one cell In principle, such anestimation is enabled by considering the variation ofthe deviations of the data from their predicted valuesacross all the cells This would estimate the variabilityInternational Journal of Computer Integrated Manufacturing 493

of an interaction effect, if it were present If an

interaction between target locations and BA set-ups

actually exists, the estimate ^s of s provided in this

study would account for both interaction and error

variability in a joint way and it would not be possible

to separate the two components Therefore, from a

practical point of view, the more the hypothesis of no

interaction is violated, the more ^s overestimates s

After fitting the model, an assessment of the

assumptions on the errors has been performed on the

realised residuals, i.e the deviation of the experimental

results from the results predicted by the fitted model

for corresponding cells (eˆij¼ dij7dˆij) The realised

residuals plotted against the positions of the targets do

not appear consistent with the hypothesis of constant

variance of the errors In fact, as shown in Figure 7(a),

the variability of the realised residuals standardised by

^

s, namely ^eij¼ ðdij ^dijÞ=^s seems different in different

target locations

For this reason, an alternative model of the data

has been considered which accounts for the variance

structure of the errors This alternative model is

defi-ned as the initial model (see Equation 1), bar the

variance of the errors which is modelled as different in

different target locations, namely:

si¼ snew di; d1¼ 1: ð3Þ

From Equation (3) it follows that snewis the unknownparameter describing the error standard deviation inthe target position 1, whereas the di’s (i¼ 2, , 6) arethe ratios of the error standard deviation in the ithtarget position and the first

The alternative model has been fitted using one ofthe class variance functions provided in the packagenlmeand the function lme() so that also snewand the

di’s are optimised jointly with the other modelparameters (m, ti and sb) by the application of theREML method (Section 5.2 in Pinheiro and Bates2000)

For the alternative model, diagnostic analyses ofthe realised residuals were not in denial of its under-lying assumptions The standardised realisations of theresiduals, i.e ^eij¼ ðdij ^dijÞ=^si, when plotted againstthe target locations (Figure 7(b)) do not appear anylonger to exhibit different variances in differenttarget locations, as was the case in the initial model(Figure 7(a)) The same standardised realisations werealso found not to exhibit any significant departurefrom normality

The fact that all the target fields have more than50% of the targets in common together with the factthat each field has been obtained by recursively adding

a single target to the current field may cause theexperimenters to expect that the measurement resultsobtained when different target fields have been used in

alternative mixed-effects models

the BA procedure have some degree of correlation If

that were the case, then the experimental results

should be in denial of the assumed independence of

the random effects bj’s The random effects, like the

errors, are unobservable random variables Yet,

algo-rithms have been developed to predict the realisations

of these unobservable random effects on the basis of

the experimental results and their assumed model

(Equations 1, 2 and 3 with the pertinent description

above) The predictor used in this investigation is

re-ferred to as the best linear unbiased predictor (BLUP)

It has been implemented in nlme and it is described,

for instance, in Pinheiro and Bates (2000) The

predicted random effects bˆj’s for the model and the

measurement results under investigation are displayed

in Figure 8(a) To highlight a potential correlation

between predicted random effects relative to target

fields that differ by only one target, the bˆjþ1’s have been

plotted against the bˆj in Figure 8(b) (j¼ 1, , 5)

From a graphical examination of the diagrams of

Figure 8 it can be concluded that, in contrast with what

the procedure for establishing the targets fields may

lead the experimenter to expect, the measurement

results do not appear to support a violation of the

hypothesis of independence of the random effects

Similar values for the BLUPs and therefore similar

conclusions can be drawn also for the initial

mixed-effect model (the BLUPs for the initial model have not

been reported for brevity)

As suggested in Pinheiro and Bates (2000) (Section

5.2, in particular), to support the selection between

the initial and the alternative model, a likelihood ratiotest (LRT) has been run using the generic functionanova()implemented in R A p-value of 0.84% led tothe rejection of the simpler initial model (8 parameters

to be estimated) when compared with the more plex alternative model (8þ 5 parameters to be esti-mated) The same conclusion would hold if theselection decision is made on the basis of the Akaikeinformation criterion (AIC) also provided in the out-put of anova() (read more about AIC in Chapters 1and 2 of Pinheiro and Bates 2000)

com-This model selection bears significant practicalimplications From a practitioner’s point of view, infact, selection of the alternative model means that therandom errors have significantly different varianceswhen measuring targets in different locations of theworkspace The workspace is not homogeneous: thereare regions where the variability of the random errors

is significantly lower than in others This also meansthat a measurement task can therefore be potentiallydesigned so that this measuring system can perform itsatisfactorily in some regions of its workspace but not

of random effects associated with consecutive targets fields

International Journal of Computer Integrated Manufacturing 495

Estimates ^ti confirm the tautological significance of

the location of the targets or block factor, whereas ^m,

depending on the the parametrisation of the model,

can for instance be the centre of mass of the point

locations or can also be associated with a particular

target location (cf Chapters 13 and 14 in Faraway

2005) All these estimates do not convey any practical

information They are therefore not reported

The significance of the random effect associated

with the BA set-up procedure has been tested using a

likelihood ratio approach, where the alternative model

has been compared with a null model characterised

by an identical variance structure of the errors but

without any random effect (i.e sb¼ 0) The p-value

was less than 10732 under the assumption of a

chi-squared distributed likelihood ratio In reality, as

explained in Section 8.2 of Faraway (2006), such an

approach is quite conservative, i.e it tends not to reject

the null hypothesis by overestimating the p-value

However, given the extremely low p-value (510732),

there is strong evidence supporting the rejection of the

null hypothesis of an insignificant random effect (H0:

sb¼ 0)

From a practical point of view, this indicates that

caution should be exerted when selecting the target

locations for running the BA algorithm during the

set-up phase: when repeating the BA procedure during the

set-up with identical positions of the transmitters, the

consideration of a different number of targets

sig-nificantly inflates the variability of the final

measure-ment results

Substituting the estimates of Equations (4) and (5)

in the adaptation of Equation (2) to the alternative

model, it is derived after a few passages that the choice

of a different number of targets when running the BA

algorithm during the set-up phase accounts for 99.22,

99.94, 99.42, 99.89, 99.99 and 99.77% of the variance

of the measured origin–target distance when the target

is in locations 1, 2, 3, 4, 5 and 6, respectively If there

was no discretion left to the operator when selecting

the number of targets and their locations during the

BA procedure, then the overall variability of the final

results in each of the location tested could have been

reduced by the large percentages reported above

It may be worth pointing out that the designed

experiment considered in this investigation could be

replicated K times, on the same or in different days

The obtained measuring results could then be modelled

with the following equation:

dijk¼ m þ tiþ bjþ ckþ eijk ð6Þ

with ck  N 0; s 2

; k ¼ 1; 2; ; K, being the dom effect associated with the kth repetition of the

ran-experiment The significance of the random effects c ’s

could then be tested in a similar way as the significance

of the bj’s has been tested above The practical use of themodel of Equation (6) is twofold First, it enables theexperimenter to detect if a significant source of varibilitycan be associated with the replication of the wholeexperiment For instance, if each replication takes place

in slightly different natural and/or artificial lightconditions, then testing the significance of the ckwouldtell if these enviromental conditions had significanteffects on the measurement results (dijk) The estimate ^sc

would quantify the increased variability of the responsevarible attributable to them Second, the increasednumber of measurements taken would raise theconfidence of the experimenter in the estimates of

^

sb; ^scand ^s For instance, it would dissipate (orconfirm) the suspicion that the experimenter may havethat the random effects attributed in Equation (1) to thedifferent setups, namely the bj’s, may be contributed to

by the natural variability due to repetition which wasestimated in Equations (4) and (5) This further studycan be considered as future work

5.2 Transient definition and analysis

In the above analysis, the average of all the 970experimental data in a cell has been considered Thevariability of each of these 970 determinations ofdistance, say st, is significantly larger than that of theiraverage (sd) If these determinations were mutuallyindependent, then it would be sd ¼ st= ffiffi

ð

p970Þ But thedeterminations are instead highly correlated, owing

to the fact that they are taken at varying samplingintervals of the order of milliseconds Identifying thecorrelation structure of these determinations is beyondthe scope of this investigation In this study, when theinstrument is measuring the tth determination, say

dt,ij, a running average of all the determinations sured until that instant, say dt,ij, is considered Aninteresting question that arises is: ‘How many deter-minations are sufficient for the instrument to provide ameasurement dt,ij that does not differ much from themeasurement result dij?’ A 2mm maximum deviationfrom dij has been considered for differentiating thesteady and the transient states of dt,ij The value t?has been used to identify the end of the transient Inother words, for any index t 4 t? it holds that

mea-jdt,ij7dijj 5 1mm

In Figure 9, for each of the 36 experimental ditions, two continuous horizontal lines 1 mm apartfrom the measurement result dijdelimit the steady-stateregion, whereas a single vertical dashed line indicatesthe transition index t?from the transient to the steadystate as defined above

con-From Figure 9, it is observed that for the sametarget location (panels in the same column) the

transition from transient to steady state may occur at

different t?’s for different BA set-ups (different vertical

dashed lines in each panel) This suspicion is even

stronger when considering t? for the same BA set-up

but for different target locations (panels on a row in

Figure 9)

To ascertain whether the variation of t? with the

BA set-ups and with the target locations examined is

significant or is only the result of uncontrolled or

un-controllable random causes, the experimental values of

t? calculated starting from the RCBD already

dis-cussed have been analysed with a fixed-effects ANOVA

model (cf Section 16.1 in Faraway 2005) The values

of t? have been computed by an ad hoc function

implemented in R by one the authors The t?’s are

assumed as though they have been generated by the

following equation:

t?ij¼ m þ biþ gjþ eij; ð7Þwhere the bi’s and the gj’s are the effects of the blocking

factor (the target locations) and of the BA set-ups,

respectively, whereas the eij’s are the random errors,assumed independent, normally distributed with con-stant variance and zero mean The parameters havebeen estimated using the ordinary least squares method

as implemented in the function lm() in R (cf R velopment Core Team 2009) The assumptions under-lying the models have been checked on the realisedresiduals and nothing amiss was found To test thepotential presence of interaction between the twofactors in the form of the product of their two effects,

De-a Tukey test for De-additivity wDe-as De-also performed (cf.Section 27.4 in Neter et al 1996) This test returned ap-value of 30.43% It is therefore concluded that theexperimental data do not support the rejection ofthe hypothesis of an additive model in favour of thisparticular type of interaction effect of target locationsand BA set-ups on tij

The effect of the target positions on t? wassignificant, i.e H0 : bi¼ 0(i ¼ 1, 2, , 6) gives rise

to a p-value of 3.88% (under the hypotheses of themodel) However, the effect of the BA set-ups did notappear to be significant, i.e H : g ¼ 0(i ¼ 1, 2, , 6)

International Journal of Computer Integrated Manufacturing 497

gives rise to a p-value of 84.96% (under the hypotheses

of the model)

From a practical point of view, there are two main

implications of these findings First, the selection of a

different number of targets when running BA

algo-rithms during the set-up phase does not appear to

have significant consequences on the duration of the

transient for obtaining a measurement Second, the

duration of the transient appears to be significantly

different for different target locations within the

work-space This may be stated as follows: there are regions

of the workspace that require longer transient periods

than others before a measurement result stabilises, and

this is expected to have consequences for the accuracy

and precision of the determination of the position of

moving objects (tracking)

In fact, if a target point is in motion at a speed

sufficient for a number of determinations greater than

t? to be recorded in each measured point of its

tra-jectory, then all the measurements results will be

representative of a steady state But this may hold for

only some portions of the target trajectory; for others,

characterised by a larger t?, such a condition may not

be satisfied with a consequent inflation of the

varia-bility of those estimated positions, which may also be

biased

6 Conclusions

The main characteristics of the Metris Indoor GPS

system have been reviewed on the basis of information

in the public domain In particular, the working

principles of the system have been presented in terms

of a cone-based mathematical model

The overall description of the system has been

instrumental in highlighting the key role of bundle

adjustment procedures during the set-up of the

sys-tem The selection of the number and location of target

points that are used when running the bundle

adjust-ment procedure during the set-up phase can be affected

by discretionary judgements exerted by the operators

To investigate the statistical significance of the

effects of this selection, a randomised complete block

design has been run on the distance between the origin

of the reference system and the measured positions of

target locations different from those used during the

bundle adjustment in the set-up phase This design

enhances the possibility for the potential effects of

different set-ups on the origin–target distances to be

detected by discriminating them from the obvious

effects of the target positions The set-ups considered

were different only in the number of the targets used

when executing the bundle adjustment procedure

A mixed-effects and a fixed-effects linear statistical

model were fitted to the measurements results using the

restricted maximum likelihood method and the nary least squares technique, respectively

ordi-The measurement results defined as the sampleaverage of the 970 determinations of distance recorded

in each target location for each set-up have been lysed with the mixed-effects model By analysing therealisations of the residuals, statistically different stan-dard deviations of the random errors were identifiedfor different target positions The work envelope ofthe instrument do not therefore appear homogeneous:

ana-in some areas the variability of the random error isgreater than in others, when performing measurements

of the distance of a target from the origin Owing tothis heterogeneity, the punctual estimates of the stan-dard uncertainty of the measured distances (sd) weredifferent for different target position and lay in a rangebetween 160.8 and 161.4 mm The different set-ups,tested to be statistically significant, always accountedfor more than 99.2% of the estimated standard un-certainty (the percentage varies for different targetpositions) This quantitative evidence suggests that theselection of points when running the bundle adjust-ment algorithms in the set-up phase should not beoverlooked Performing this selection in a consistentway according to some rule that ideally leads tochosing the same points when the transmitters are inthe same positions may be a course of action worthconsidering Also, for replication and comparisonpurposes, it may be advisable to quote the locations

of the targets used in setting up the system whenreporting the results of a measurement task

The duration of the transient, i.e the number ofdeterminations of distance needed for their currentaverage to be within +1 mm from the measurementresult (the average of the 970 determinations), has beenanalysed with the fixed-effects model The differentset-up configurations considered did not have anysignificant effect on the duration of the transient.However, this duration was significantly different indifferent target locations It can therefore be concludedthat the working space of the instrument is hetero-geneous also for the characteristics of the transient ofmeasurement It is expected that this conclusion hasnegative implications on the precision and unbiased-ness of the measurements obtained when using theinstrument for tracking moving points or movingobjects that the target points (or vector bars) areattached to Given a pre-specified configuration ofiGPS transmitters without any zone partitions havingbeen pre-established among them, if an object ismoving within an area of the working space of such

an iGPS, say area A, its position may be trackedcorrectly, because the transient is sufficiently shortthere But if the same movement of the same object istracked by the same iGPS in another area of the same

iGPS working area, say area B, the system may not be

able to track its location correctly because the transient

may not yet have finished

Acknowledgements

This study is part of the research initiatives of The Bath

Innovative Design and Manufacturing Research Centre

(IdMRC), which is based in the Department of Mechanical

Engineering of the University of Bath and which is supported by

the United Kingdom Engineering and Physical Sciences

Research Council (EPSRC) In particular, this investigation

was carried out within the scope of the IdMRC research theme

‘Metrology, Assembly Systems and Technologies’ (MAST),

which is coordinated by Professor Paul Maropoulos

References

BS ISO 5725-1, 1994 Accuracy (trueness and precision) of

measurement methods and results – Part 1: General

Stan-dards Institution Available from: http://shop.bsigroup

com/ [Accessed 14 March 2010]

Craig, J.J., 1986 Introduction to robotics mechanics & control

Reading, MA, USA: Addison-Wesley

Estler, W.T., et al., 2002 Large-scale metrology – an update

Annals of the CIRP, 51 (2), 587–609

Faraway, J., 2005 Linear models with R Boca Raton, FL,

USA: Chapman & Hall/CRC

Faraway, J., 2006 Extending the linear model with R:

generalized linear, mixed effects and nonparametric

regression models Boca Raton, FL, USA: Chapman &

Hall/CRC

Hartley, R.I and Sturm, P., 1997 Triangulation Computer

Vision and Image Understanding, 68 (2), 146–157

Hedges, T.M., et al., 2003 Position measurement system and

method using cone math calibration United States

Patent US6535282B2, March

JCGM (Joint Committee for Guides in Metrology), 2008

JCGM:100:2008 (GUM:1995 with minor corrections)

Evaluation of measurement data – guide to the expression

BIPM – Bureau International des Poids et Mesures

Available from: http://www.bipm.org [Accessed 7 June

2009]

Liu, H., et al., 2007 Survey of wireless indoor positioning

techniques and systems IEEE Transactions on Systems,

Man, and Cybernetics – Part C: Applications and Reviews,

37 (6), 1067–1080

Liu, Z., Liu, Z., and Lu, B., 2008 Error compensation ofindoor GPS measurement In: C Xiong et al., eds.,Proceedings of the first international conference onintelligent robotics and applications (ICIRA2008), Part

II, 15–23 October 2008 Wuhan, People’s Republic ofChina Lecture Notes in Computer Science 5315 Berlin:Springer-Verlag, 612–619

Lourakis, M.I.A and Argyros, A.A., 2009 SBA: a softwarepackage for generic sparse bundle adjustment ACMTransactions on Mathematical Software, 36 (1), 1–30.Maisano, D., et al., 2008 Indoor GPS: system functionalityand initial performance evaluation International Journal

of Manufacturing Research, 3 (3), 335–349

Neter, J., et al., 1996 Applied linear statistical models 4th ed.New York: Irwin

Pinheiro, J.C and Bates, D.M., 2000 Mixed-effects models in

S and S-plus New York: Springer-Verlag

Piontek, H., Seyffer, M., and Kaiser, J., 2007 Improving theaccuracy of ultrasound-based localisation systems Per-sonal and Ubiquitous Computing, 11 (6), 439–449.Priyantha, N.B., Chakraborty, A., and Balakrishnan, H.,

2000 The Cricket location-support system In: ings of the 6th annual international conference on mobilecomputing and networking (MobiCom’00), 6–11 August.Boston, MA New York: ACM, 32–43

Proceed-R Development Core Team, 2009 Proceed-R: a language and

Foundation for Statistical Computing Available from:http://www.R-project.org ISBN 3-900051-07-0

Sae-Hau, C., 2003 Multi-vehicle rover testbed using a newindoor positioning sensor Thesis (Master’s) Massachu-setts Institute of Technology

Savvides, A., Han, C.C., and Strivastava, M.B., 2001.Dynamic fine-grained localization in ad-hoc networks

of sensors In: Proceedings of the 7th annual internationalconference on mobile computing and networking (Mobi-Com’01), 16–21 July Rome New York: ACM, 166–179.Triggs, B., et al., 2000 Bundle adjustment – a modernsynthesis In: B Triggs, A Zisserman, and R Szeliski,eds Proceedings of the international workshop on visionalgorithms (in conjunction with ICCV’99), 21–22 Sep-tember 1999, Kerkyra, Corfu, Greece Lecture Notes onComputer Science 1883 Berlin: Springer-Verlag, 298–372

Welch, G., et al., 2001 High-performance wide-area opticaltracking – the HiBall tracking system Presence: Tele-operators and Virtual Environments, 10 (1), 1–21.Wolf, P and Ghilani, C., 1997 Adjustment computations:statistics and least squares in surveying and GIS NewYork: Wiley

International Journal of Computer Integrated Manufacturing 499

A real-time simulation grid for collaborative virtual assembly of complex products

X.-J Zhen, D.-L Wu*, Y Hu and X.-M Fan

CIM Institute, Shanghai Jiao Tong University, Shanghai, China(Received 9 April 2009; final version received 9 February 2010)Simulation of collaborative virtual assembly (CVA) processes is a helpful tool for product development However,existing collaborative virtual assembly environments (CVAE) have many disadvantages with regard to computingcapability, data security, stability, and scalability, and moreover it is difficult to create enterprise applications inthese environments To support large-scale CVAEs offering high fidelity and satisfactory interactive performanceamong various distributed clients, highly effective system architectures are needed In this paper, a collaborativevirtual assembly scheme based on grid technology is proposed This scheme consists of two parts: one is a grid-basedvirtual assembly server (GVAS) which can support parallel computing, the other a set of light clients which cansupport real-time interaction The complex and demanding computations required for simulation of virtualassembly (VA) operations, such as model rendering, image processing (fusion), and collision detection, are handled

by the GVAS using network resources Users at the light clients input operation commands that are transferred tothe GVAS and receive the results of these operations (images or video streams) from the GVAS Product data aremanaged independently by the GVAS using the concept of RBAC (role-based access control), which is secureenough for this application The key related technologies are discussed, and a prototype system is developed based

on the web services and VA components identified in the paper A case study involving a car-assembly workstationsimulation has been used to verify the scheme

Keywords: grid; collaborative virtual assembly; complex product; real-time collaborative simulation

Notation

CDM collision detection model

CVA collaborative virtual assembly

CVAE collaborative virtual assembly

CVA technology is used to develop complex products

such as automobiles and ships It provides an effective

experimental assembly environment for designers

working at different locations (Lu et al 2006), who

can exchange product data and discuss and verify the

assembly scheme to improve the previous design

scenario Many CVA systems or prototypes have

been built for product development (Bidarra et al

2002, Shyamsundar and Gadh 2002, Chen et al 2004,

Chryssolouris et al 2009) However, existing CVA

systems still have many disadvantages In general,

virtual environments (VEs) have no modelling tion, which means that products must generally bemodelled in a CAD environment, creating productmodels that cannot be imported into VE directly; as aresult, much preparatory work such as model transfor-mation must be done before the VA task can beperformed Moreover, in the context of expandingrequirements for assembly simulation of complexproducts, current CVA systems lack adequate comput-ing power Most CVA systems support only single-PC,not parallel, computing, which is seriously insufficient

func-to meet requirements For example, the frame rate forrendering a model of a whole car is about 2*6F/S(frames/second), which is not compatible with inter-active operation In addition, the computing resources

of all user nodes are allocated statically before the task

is begun, which limits the stability and extensibility ofthe system

Grid technology provides a new way to promotecollaborative virtual assembly using the concept ofsharing distinct resources and services This approachhas been applied successfully in areas of computerscience requiring massive computing, such as parallelcomputing and massive data processing, but less so inthe areas of design and manufacturing, especially when

*Corresponding author Email: wudianliangvr@gmail.com

Vol 23, No 6, June 2010, 500–514

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

Ó 2010 Taylor & Francis

DOI: 10.1080/09511921003690054

real-time computing is required However, the

char-acteristics of grid technology are a perfect fit for the

requirements of a CVA system, and CVA based on grid

technology offers many advantages with regard to

computing power, data security, stability, and

scal-ability, as well as ease of constructing enterprise

applications

In this paper, a collaborative virtual assembly

scheme based on grid technology is proposed, and a

prototype system called VAGrid (Collaborative

Vir-tual Assembly-based Grid) is developed based on web

service and VA components This system consists of

two parts: a grid-based virtual assembly server

(GVAS) and a set of light clients Computing tasks

are handled by the GVAS using resources available

over the internet, with users performing only simple

interactive operations using a graphical interface The

key related technologies are discussed in detail

The rest of this paper is organised as follows In

Section 2, related research on CVA and grid

technol-ogy is reviewed Section 3 describes the structure and

workflow of the system The representative features

and capabilities of VAGrid are described in Section 4

Section 5 provides a case study, and Section 6 states

conclusions and directions for future work

2 Related research

2.1 CVA

Many researchers have already conducted extensive

research into CVA, and significant results have been

achieved An internet-based collaborative product

assembly design tool has already been developed

(Shyamsundar and Gadh 2002) In this system, a new

assembly representation scheme was introduced to

improve assembly-modelling efficiency Liang has also

presented a collaborative 3D modelling system using

the web (Liang 2007) Lu et al developed a

collabora-tive assembly design environment which enabled

multiple geographically dispersed designers to design

and assemble parts collaboratively and synchronously

using the internet (Lu et al 2006) Web-based virtual

technologies have also been applied to the automotive

development process (Noh et al 2005, Dai et al 2006,

Pappas et al 2006)

These researchers have proposed various

ap-proaches to enable collaboration among multiple

designers, but the only interactive modes supported

by the CVA environment, such as chat channels, were

not found to be effective or intuitive Moreover, the

performance of these systems, especially when

sup-porting real-time assembly activity for complex

pro-ducts, was considered inadequate In an effort to solve

these problems, the relative performance of various

distribution strategies which support collaborative

virtual reality environments, such as client/servermode, peer-to-peer mode, and several hybrid modes,has been discussed (Marsh et al 2006) Theseresearchers proposed a hybrid architecture whichsuccessfully supported real-time collaboration forassembly For supporting the interactive visualisation

of complex dynamic virtual environments for trial assemblies, a dynamic data model has beenpresented, which integrates a spatial data set ofhierarchical model representations and a dynamictransformation mechanism for runtime adaptation(Wang and Li 2006) Based on this model, complexityreduction was accomplished through view frustumculling, non-conservative occlusion culling, and geo-metry simplification

indus-2.2 Grid computingGrid computing was first proposed by Ian Foster in the1990s (Foster and Kesselman, 1999) It aims to shareall the resources available on the internet to form alarge, high-performance computing network An im-portant characteristic of a grid-computing environ-ment is that a user may connect to the grid-computingsystem through the internet, and the grid-computingsystem can provide all kinds of services for the user.Some grid toolkits, such as Globus (Foster 2006),Legion (Grimshaw and Natrajan 2005), and Simgrid(Emad et al 2006), which provide basic capabilitiesand interfaces for communication, resource location,scheduling, and security, primarily use a client-serverarchitecture based on centralised coordination Thesegrid-computing applications use client-server architec-tures, in which a central scheduler generally managesthe distribution of processing to network resources andaggregates the processing results These applicationsassume a tightly coupled network topology, ignoringthe changing topology of the network, and are suitablefor large enterprises and long-time collaboration.However, in the area of production design,especially for real-time design simulation, few relevantstudies have been reported Li et al (2007a, 2007b)presented the concept of a collaborative design grid(CDG) for product design and simulation, and acorresponding architecture was set up based on theGlobus toolkit, version 3.0 Fan et al (2008) presented

a distributed collaborative design framework using ahybrid of grid and peer-to-peer technologies In thisframework, a meta-scheduler is designed to accesscomputational resources for design, analysis, andprocess simulation, which can help in resourcediscovery and optimal use of resources To meetindustrial demands for dynamic sharing of variousresources, Liu and Shi (2008) proposed the concept ofgrid manufacturing According to the characteristics ofInternational Journal of Computer Integrated Manufacturing 501

the resources to be shared and the technologies to be

used, grid manufacturing distinguishes itself from

web-based manufacturing by providing transient services

and achieving true interoperability To support

real-time design simulation, a hybrid of HLA (high-level

architecture) and grid middleware was used (Rycerz

et al 2007) These systems can solve some problems

from a special viewpoint, but to support collaborative

simulation design for complex products, further

research is needed

3 Structure and workflow of VAGrid

3.1 Functions and performance requirements

Complex products such as automobiles and ships have

similar features: (1) numerous components, (2) complex

structure, (3) high research and development costs, and

(4) requirement for a large number of designers Some

form of collaborative design has been used by most

companies making complex products However, to date

only some indirect applications, such as physically

co-located meetings and CAD-based conferencing, have

been attempted Unfortunately, these approaches have

many deficiencies with regard to service efficiency,

application effectiveness, and convenience (Trappey

and Hsiao 2008) In contrast, this research targets the

entire distributed team-design scenario, involving all

the participating designers and supervisors A direct

way is needed to enable geographically distributed

designers to assemble their individual designs together

in real time To make this possible, the following set of

system requirements should be incorporated into the

development process:

Ease of use System configuration, including

allocation of computing resources, is done

automatically, and the user can obtain the

desired results by means of simple interactions

Convenient data conversion Product data

re-quirements in the virtual environment are

differ-ent from those in the CAD environmdiffer-ent, so data

conversion is required This process should be

simple or automatic and should support common

CAD software such as UG, CATIA, and PRO/E

Good real-time scene rendering performance

The virtual scene at each user station should be

responsive enough to meet the needs of

inter-active operation, which requires powerful

com-puting resources to perform model rendering,

collision detection, and similar tasks

Strong data security These systems have many

users, including product designers, assembly

technologists, and even component suppliers

User authorisation or similar measures must be

taken to maintain data security

Multi-user scene synchronisation Consistency ofthe scene across all the system nodes must bemaintained This means that when the scene at aparticular node changes because of user manip-ulation, the information must transfer to all othernodes and update their scenes synchronously Multi-modal interaction Users interact with VEthrough multiple modalities for different hard-ware, such as the data glove and FOB (flock ofbirds), as well as the common keyboard andmouse as in CAD

3.2 Structure of a grid-based virtual assemblysimulation for complex products

The basic idea of collaborative assembly is shown inFigure 1 Geometric modelling and assembly design ofproducts are carried out in a CAD system by designers

at different locations Then geometric and assemblyinformation are transferred to the collaborative assem-bly environment by means of a special data interface.All designers can share the same virtual assemblyverification environment collaboratively to performassembly analyses and assembly process planning forproducts, as well as assembly operation training Thesystem can run over the internet or on a LAN (LocalArea Network), or particularly on a company intranetfor reasons of performance and stability

In the context of the functional requirements andthe basic concept described above, several structurescan be used A distributed parallel architecture (Zhen

et al 2009) based on HLA and MPI (Message PassingInterface), with many client nodes and one masternode, has been used, in which each client node wassupported by PCs in a LAN The advantage of thisapproach was fast execution speed, especially atinitialisation because data were saved at each localposition However, this approach also had manydisadvantages: (1) data protection is difficult to achievefor distributed data; (2) manual configuration of thesupporting resource nodes and all the resource nodes isstatic and unreliable; (3) the system can support onlyone task at a time and is therefore not suitable forgeneral use For these reasons, an implementationscheme based on grid technology has been proposed,

as shown in Figure 2 The computing resources neededduring the assembly process for tasks such as render-ing, collision detection, and image processing (thesemay be any idle resource available on the internet or acorporate intranet) are managed dynamically by thegrid Multi-tasking can be supported, but there is onlyone virtual assembly scene for each task, maintainedand updated by the grid Product data and evaluationresults are stored at an independent location Theconfiguration of each user is simple, involving only an

Figure 2 CVA solution based on grid.

I/O device or a multi-channel stereo system if using an

immersive virtual environment The scene at each user

client station is a sequence of continuous images or a

video stream that a user can ‘see’ at his location Users

are classified as an assembly task manager or a

participant

3.3 Architecture and workflow of VAGridThe system architecture of VAGrid is illustrated inFigure 3 All users first register with the system usingthe grid portal Once a user has finished designingassemblies or subassemblies using a CAD system, theproduct models will be submitted to a given location inInternational Journal of Computer Integrated Manufacturing 503

the grid database The assembly-task manager accesses

the database with a valid authorisation, prepares the

corresponding data in the VE, and then sets up and

launches a new task The system will search

auto-matically for resources to support this task If

insufficient resources are available, a rejection message

will be generated; otherwise, a collaborative assembly

environment will be established The related users then

can join the task and enter the CVE to perform

assembly verification together

The system consists of five key parts:

(1) Grid platform management: the basis of the

system This part provides management of

communication status and computing

re-sources for the CVE Management of users,

tasks, and resources is also performed by this

part

(2) Product management: convenient data

trans-formation is an essential requirement for

complex products, and management of these

data in a distributed grid environment is a

complex task

(3) Remote real-time collaboration: collaboration

based on virtual user models in a virtual

environment and remote real-time interactive

assembly operations are handled by this

module

(4) Virtual scene graphics management: this

mod-ule dynamically maintains the virtual scene

displays, including assembly based on solution

of constraints and constraint navigation The

design and implementation of this module have

already been described in the paper on IVAE

(Yang et al 2007)

(5) Tools for evaluation and analysis: a set of tools

for distance and dynamic gap measurement,

assembly path and sequence tracking, dynamiccollision detection, and assembly process eva-luation report generation, which can be inter-actively used by each user

4 Representative features and capabilities4.1 Grid management

The grid platform is the main framework of thesystem, which handles the management of users(registration, logon, and user status monitoring), tasks(startup, maintenance), and resources (monitoring,dynamic configuration)

User managementUsers can be classified into two types: the taskmanager and ordinary users The former is incharge of the whole task and holds the highestauthority; the latter is the operator in the virtualenvironment Although there is only one virtualenvironment, with no scene-synchronisation pro-blems among client nodes, the operating author-ity of each user is different, and thereforeproblems with operating collisions do arise andmust be solved In this case, the manager definesroles related to the task and sets levels ofauthority according to these roles When a userjoins a task, he enters into a role and operateswith its corresponding authority

Task managementUnlike existing systems, the VAGrid system cansupport multiple tasks running in parallel Eachtask has its own group of users, simulation data,and supporting resources, all managed uniformly

by the task scheduling module The workflow isshown in Figure 4 A task manager sets up a newtask and submits it to the task queue Theresource scheduling module queries the resourceand starts the task by calling CVA services, whichare the basic grid services deployed among thegrid nodes, such as virtual scene graphics manage-ment, model rendering, and collision detection Resource management

This is the most complex part of the system.Large amounts of real-time data must beprocessed, and large amounts of many kinds ofcomputing resources, for example for rendering,are required These resources are distributedamong the computing nodes on the internet Aresource cannot be used before a plug-in isinstalled, which is a small program that encap-sulates all the resources with the same kind ofaccess interface, as shown in Figure 5 Each noderegisters with the registration centre with an

attributes message When a node fails, its work

will be delivered to new resource nodes

dynamically

4.2 Data management

Data management (DM) is one of the key factors

which provide a flexible and extensible mechanism for

data manipulation in the grid and data delivery to grid

sites based on the participating entities’ interests

Depending on the system architecture and operation,

users need only to upload CAD models related to the

task to grid data-storage nodes Once in the virtual

environment, CAD models cannot be loaded directly,

because some preparatory work, including data

storage, processing, and access, is required when

moving from CAD to the virtual environment

A virtual assembly task requires three main types

of data: product data, including a CAD model and

data in the virtual environment; information on

assembly tools and processes (saved as a file); and

simulation results, as shown in Figure 6 The first type,

product data in the virtual environment, includes the

display model, the part information model, assembly

hierarchy information (saved as a file), and thecollision detection model (CDM) The second datatype consists of important auxiliary information forevaluation Simulation results include elements such as

an assembly process video, an assembly sequence, apath information file, and an evaluation report.Among these data, product data is managed andmaintained by, and only by, the task manager, becausesecurity requirements dictate that he has the soleauthority to write to the database; others can onlyupload data The second data type is shared by allusers of a particular task Assembly results are saved in

a folder accessible to users and can be downloaded bythem consistently with their authorisations

Some preparatory work must also be done for datatransformation Assembly hierarchy information andconstraint information can be obtained from the CADenvironment after a static interference check Thedisplay model is the ‘visible’ part, and the CDM is used

to perform a dynamic interference check; these modelsmust be transformed from CAD models A specialinterface based on the ACIS solid modelling kernel hasbeen developed for this complex task Each CADmodel is transformed into a sat file, from which thedisplay model and the collision detection model can beobtained

Display model: several types of model can beused such as step, flt (Open Flight), etc All ofthese are polygon models with colour, texture,and other appearance properties Several typesare supported by the system, and all the modelswill be transformed automatically before thesystem starts

CDM: This is an internally defined type whichcan be transformed by a polygon model asdescribed in the last step and then simplified into

a hierarchical bounding box depending on thecollision-detection precision

International Journal of Computer Integrated Manufacturing 505

Figure 7 shows a display model of a car chassis part

and its collision models at various precisions All these

data are saved into grid data-storage nodes by the

MYSQL system When the system starts, all the

computing resource nodes will load their

correspond-ing data accordcorrespond-ing to task requirements

4.3 Multi-user collaborative interactive operations

Unlike most existing CVA systems, users at each client

send operation commands to a single ‘remote’ virtual

scene and then receive the results of these operations,

which are scene fragments as seen from the users’

viewpoints It frequently occurs that many users stay at

their assembly workstations, observing with two eyes,

operating with two hands, where each of them can feel

each other’s presence, but not see each other This is

real intuitive and effective collaboration

4.3.1 Co-operation based on virtual user models

According to the system architecture, users at each

client only send operating commands and receive

simulation results over the network; they do not save

product data nor perform computing tasks The

management of users and of users’ operating

com-mands is very heavy work Therefore, a virtual-user

model is used, and each real user has a corresponding

virtual user in the virtual environment The virtual user

processes, not only the current commands from the

user, but the user’s attribute information such as

location, viewpoint parameters, and so forth The

architecture of the virtual user model is shown in

Figure 8

Here, in the virtual user attributes information,

‘Type’ can be task manager or general collaborative

user ‘RoleID’ is the ID number of the role which theuser takes on in the task; for example, if the ID of doordesigner is ‘6’, then the ‘RoleID’ of a virtual user acting

as a door designer is ‘6’ Viewpoints are the ‘eyes’ of avirtual user in the virtual assembly environment; soundchannels are his ‘ears’, operations are his two hands,and the supporting resource records the list ofcomputing nodes which provide services to this user.The system set up a new virtual user object auto-matically when a client joins in

In VAGrid, there is no synchronisation problemamong the clients because there is one single virtualscene, but manipulation conflicts still exist, forexample when:

An object is grasped by two or more userssimultaneously; or

Two objects that have an assembly constraintrelation are manipulated by two userssimultaneously

In the former case, if the object has been grasped,refuse all grasp requests; otherwise continue Then thelevel of authority and the role of the user are checked;

if they are inappropriate, refuse him; otherwise, accepthis application Finally record the exact time, so as togive precedence to the earlier user In the latter case,when a user tries to assemble one part with another,first determine whether it is a free part; if yes, thencarry out the assembly operation, otherwise continuechecking Then determine whether the operator is thesame person as the user If yes, this means that the user

is grasping two objects simultaneously, and assemblycan be continued using a two-handed assemblyprocess; if no, a prompt message that the object isalready being operated on will be generated

4.3.2 Remote real-time interactive assembly operation

Many kinds of equipment can be used for interactive

operation, such as FOB (Flock of Birds, a kind of

position-tracking device), data glove,

mouse/key-board, and other I/O equipment While in the

VAGrid system, the data and its processing program

all reside at the grid site and the result at the client

site is the rendered scene image To achieve

interac-tion, the user’s command must be sent to the grid site

in real time An image-based remote interactivescheme is provided; the basic workflow and hardwareconfiguration of the client are shown in Figure 9 Themanipulator command is first coded and sent to thegrid over the network, where it is then decoded by thesystem and sent to the assembly simulation scenenodes Then the parameters of related virtual userchanges and all rendering nodes within ‘RNodeList’will be updated in response In the same way, othercomputing nodes will also be updated, and a newvideo image will arrive at the client workstation Inaddition, users can communicate with each other bysending text messages

Using this scheme, users can manipulate virtualobjects conveniently according to their hardwarestatus Because the inputs and outputs are separate,multi-channel immersive stereo can be easily achieved,

as shown in Figure 9

4.3.3 Interactive operation using virtual hands or byassembly using tools and equipment in the virtualenvironment

4.3.3.1 Interactive operation using virtual hands.There are two virtual hand models in the virtualscene, corresponding to the real hands of the user.Virtual hands are driven synchronously by real handswith the same position and orientation The basicmanipulation of an object (part or assembly) in thevirtual scene includes grasping, moving, con-straint confirmation, motion navigation, and objectrelease

Grasping: a user who wants to pick up an objectsends an application request over the grid to the virtualscene and waits for feedback information If the objecthas been grasped, either a message, ‘cannot be

International Journal of Computer Integrated Manufacturing 507

grasped now,’ will be generated or the grasp will be

successful

Moving: after a successful grasp, the object is

affixed to the virtual hand and can be moved to

wherever the user wants The whole scene will be

updated in real time

Constraint confirmation and motion navigation:

when the object has been moved near to the desired

position, a marker appears (the location label of the

object), and the user can perform a gesture to confirm

the precise mating location based on assembly

constraint recognition

Object Release: the client (user) sends a command

to release the object, and the relation between the

object and the virtual hand is broken

4.3.3.2 Assembly using tools and equipment Taking

the real assembly environment into account, including

assembly tools, fixtures, and assembly equipment, is an

important aspect of virtual assembly simulation

Interactive assembly operations using assembly tools

should be provided In the assembly process for a

complex product, equipment can be classified into

three types: automatic, semi-automatic, and manual

tools, as shown in Figure 10 Assembly tools act as a

special ‘assembly’, inheriting all the attributes,

including the collaborative attributes, of the assembly

object Similarly, a part of an assembly tool inherits all

the attributes of a part object In addition, the

assembly tool has particular attributes that make it

able to manipulate other objects

The working process of a virtual tool is similar to

that in the real world Most tools select their target

object dynamically by colliding with it, except for some

special tools When a tool operates, for example a

screw tool, it first selects an object by colliding with a

bolt, then creates the axis constraint between the tool

and the target, and finally the tool can be navigatedusing the axis constraint to the ending facet which isthe final position

4.4 Tools for evaluation and analysis in assemblysimulation

To evaluate the assembly operation effectively, severalauxiliary tools are provided, such as distance anddynamic gap measurement, assembly path record andreplay, trajectory display, and an assisted evaluationreport

Distance and dynamic gap measurementThe distance between two points, a point and aline (a line segment), a point and a facet, or twofacets can be measured interactively Dynamicgap computation for two models is also pro-vided The user can select the measured object,and the gap value will be shown in the virtualscene in real time

Assembly path and sequence

In the virtual environment, the assembly quence and the paths followed by the parts arefreely and arbitrarily selected using the dataglove, but there must be an optimal sequence andset of paths Recording and replaying thesequence of motion and the paths followed can

se-be used to optimise the assembly process.Trajectory display is used as an intuitive way toshow the path of motion of a component In theVAGrid system, replaying the assembly recordmust be agreed upon by all the users becausethere is only one virtual scene

Part interference and collisionThe basic function of virtual assembly is to verifythe design intent But here the interferences

among tools and fixtures can also be calculated,

except for those among parts

Assembly process evaluation report

An evaluation report will be created

automati-cally when an assembly task is finished The

information in the report is divided into three

parts The first contains general statistical

in-formation about the whole assembly task The

second contains status information for all the

parts The last one contains interference

informa-tion, which shows the interference of parts in the

assembly process using a special symbol

5 Case studies

With the aim of verifying the feasibility of the

system while taking into account the assembly

space, an assembly simulation was performed usingthe VAGrid system on a classical workstation tomodel a rear suspension and front suspension Thecontent of the simulation includes the layout offixtures and tools, the operating space, and thedynamic interferences during the assembly process

An assembly technologist acted as task managerand was in charge of setting up the task andcoordinating all the participants The participantsincluded designers of the rear suspension and frontsuspension, designers of tools and fixtures, andassembly operators

To use resources over the internet, plug-ins must beset up at each resource node In this case, computingresources were used over an enterprise intranet, andproduct data were saved in the internal database of theenterprise The detailed steps followed using VAGridcan be described as follows

constraint (d) Run navigated by the axis

International Journal of Computer Integrated Manufacturing 509

5.1 RA750 car assembly workstation simulation

5.1.1 Registration and logon for users

Each user first registers and obtains an account The

task manager sets up a simulation task team and

defines roles and authority related to the present task

Figure 11 shows the interface for a user registering at a

portal

5.1.2 Preparation for simulation

Simulation initialisation consists of three steps: (1) all

CAD models are uploaded to the grid database by

designers from their dispersed locations; (2) the task

manager accesses the database, transforms the models

for simulation, and saves them to a given location; (3)

the folder path together with the car assembly

information are written into a simulation file (.xml as

the default format) A segment of the simulation file is

shown in Figure 12

5.1.3 Simulation task initialisation

The task manager starts up the task using the file

described above, and all related service nodes run

automatically to support this task: assembly scene

resources, rendering resources, and so forth Once allrelated resources are running, the status of this task isset to be a collaborative task, and users can join in Thewhole assembly scene (at a default viewpoint) can bebrowsed by opening an interactive interface, shown inFigure 13, as part of the assembly scene at the taskmanager workstation Other users can apply for a roleand join the task

5.1.4 Multi-user collaborative assembly operationOnce all the related users have joined the task, thecollaborative assembly operation will be performedunder the coordination of the task manger Severalinteractive modes can be supported by the system,among them the ordinary keyboard and mouse as inCAD, the 5DT Cyberglove and FOB (flock of birds),and the Cyber Glove/Touch glove with haptic sensingand FOB Figure 14 illustrates two interaction modes:(a) a user operating with a keyboard and mouse; (b) auser operating with the Cyber Glove/Touch glove andFOB

5.1.5 Aids to evaluation and analysisAssembly evaluation tools can be used at any pointduring the operation process as needed To access these

tools, the user must display the space analysis menu

and select ‘distance measure’ as the analysis mode In

this case, two modes will be used One displays the

dynamic gap in real time between the grasped object

and other important objects in the environment, as

shown in Figure 15(a) The other displays the

minimum distance between the given object and

another object at final assembly status, as shown in

Figure 15(b) When the user displays the assembly path

management menu and selects a path record, the

system generates a recording of the assembly process,

including assembly sequence and path of motion By

replaying the path and the trajectory display, the best

path can be determined In addition, trajectory display

with body mode can be used to check the tools’

operational space Figure 16 shows the trajectory

of moving a rear suspension using interactive

equipment

A design evaluation report is generated after the

assembly task is completed and will be saved to the

grid database in the appropriate folder depending on

the users A segment of a report of this type for anassembly operation is shown in Figure 17 A total of

12 interferences occurred The image shows theinterference between the screw tool and the frontsuspension fixture, which left too small a space for theoperation The fixture will be modified later in theCATIA environment and re-verified in VAGrid.Eventually all the interferences will be eliminated

monitor scene at manager client node (a) Segment ofresource configure file (b) Part of assembly scene atmanager client

5.2 Case discussion and analysisAssembly technologist, related designer and operatorjoin in the evaluation and verify task The applicationresult shows that VAGrid is an effective tool forassembly process evaluation

International Journal of Computer Integrated Manufacturing 511

5.2.1 VAGRID prototype system

Prototype system can meet the requirements of

multi-user real-time collaborative assembly

simula-tion based on grid technologies GVAS part

under-takes scene visualisation and simulation computing

task by configuring and starting the grid resources

automatically

(1) Interactive assembly operations smooth and

no lag feeling during simulation The scene

update frequency is greater than 17 F/S (frames/

second) for stereo display mode (34 F/S for

non-stereo mode), which is much higher than

efficiency using single PC (about 2*6 F/S)

(2) Consistency of the scene at all user clients is

well maintained with the only scene manager

node at grid

users collaboratively assembly (b) Auxiliary Manipulator for

rear-suspension

computation (a) Dynamic gap display between suspension spring and other parts (b) Gap between oilpipe and the protective board

interactive equipment

(3) Only the task manager can access the database

at GVAS, and then data security and

consis-tency can be guaranteed

5.2.2 Assembly process evaluation application

The simulation process is carried through according to

the assembly process requirements using the assistant

tool

(1) The assembly sequence is reasonable and

there is no component that cannot be

assembled

(2) The product assembly path is feasible and no

interference by reason of design occurs

(3) There is sufficient room for manoeuvre except

for front suspension fixture, the size of the

upper frame of which needs to reduce

6 ConclusionsCVA technologies are being applied more and more inindustry, but still encounter many problems In thispaper, a CVA system called VAGrid based on gridtechnology is presented, in which the heavy computingtasks such as model rendering and image processingare performed over the grid using resources availableover the internet Product data are managed indepen-dently by the grid with role-based access authorisation,which is secure enough for this purpose Users need toperform only simple interactive operations using agraphic interface Compared with existing CVAsystems, VAGrid offers many advantages with regard

to computing capability, data security, and scalability,and moreover is well suited to constructing enterpriseapplications

However, VAGrid still has some disadvantages inapplication It relies on having a high-performancenetwork, and plug-ins must first be set up for a PC to

be used as a computing resource node In addition, agood dynamic resource configuration mechanism isessential because otherwise part of a virtual scene willdisappear if a node does not work; this mechanism isbeing studied further

Assembly evaluation using CVA is still a luxuryapplication which is mostly used by large companies.Providing service for other enterprises using VAGrid is

a research topic for the future

Acknowledgements:

This work has been supported by a Key Project grantfrom the National Nature & Science Foundation ofChina (grant no 90612017) and a Key Project grantfrom the Science & Technology Commission ofShanghai Municipality (grant no 08DZ1121000)

References

Bidarra, R., Kranendonk, N., Noort, A., and Bronsvoort,W.F., 2002 A collaborative framework for integratedpart and assembly modeling In: Proceedings 7th ACMsymposium on solid modeling and applications, ACMPress, 389–400

Chen, L., Song, Z., and Feng, L., 2004 Internet-enabled time collaborative assembly modeling via an e-assemblysystem: status and promise Computer-Aided Design, 36(9), 835–847

real-Chryssolouris, G., Mavrikios, D., Pappas, M., Xanthakis, E.,and Smparounis, K., 2009 A web and virtual reality-based platform for collaborative product review andcustomisation In: L Wang and A.Y.C Nee, eds.Collaborative design and planning for digital manufactur-ing London: Springer, 137–152

Dai, K.Y., Li, Y.S., Han, J., Lu, X.H., and Zhang, S.S.,

2006 An interactive web system for integrated dimensional customization Computers in Industry, 57(8–9), 827–837

node (a) Segment1 (task statistic and status information) of

an evaluation report (b) Segment2 (interference information)

of an evaluation report

International Journal of Computer Integrated Manufacturing 513

Emad, N., Shahzadeh, S., and Dongarra, J., 2006 An

asynchronous algorithm on the NetSolve global

comput-ing system Future Generation Computer Systems, 22 (3),

279–290

Fan, L.Q., Senthil, K.A., Jagdish, B.N., and Bok, S.H., 2008

Development of a distributed collaborative design

frame-work within a peer-to-peer environment CAD

Computer-Aided Design, 40 (9), 891–904

Foster, I., 2006 Globus toolkit version 4: software for

service-oriented systems Journal of Computer Science

and Technology, 21 (4), 513–520

Foster, I and Kesselman, C., 1999 The grid: Blueprint for a

future computing infrastructure San Francisco: Morgan

Kaufmann

Grimshaw, A and Natrajan, A., 2005 Legion: lessons

learned building a grid operating system Proceedings of

the IEEE on grid computing, 93 (3), 589–603

Li, Z., Jin, X.L., Cao, Y., Zhang, X.Y., and Li, Y.Y., 2007

Architecture of a collaborative design grid and its

application based on a LAN Advances in Engineering

Software, 38 (2), 121–132

Li, Z., Jin, X.L., Cao, Y., Zhang, X.Y., and Li, Y.Y., 2007

Conception and implementation of a collaborative

manufacturing grid International Journal of Advanced

Manufacturing Technology, 34 (11–12), 1224–1235

Liang, J.S., 2007b Web-based 3D virtual technologies for

developing a product information framework

Interna-tional Journal of Advanced Manufacturing Technology, 34

(5–6), 617–630

Liu, Q and Shi, Y., 2008 Grid manufacturing: a new

solution for cross-enterprise collaboration International

Journal of Advanced Manufacturing Technology, 36 (1–2),

205–212

Lu, C., Fuh, J.Y.H., Wong, Y.S., Qiu, Z.M., Li, W.D., and

Lu, Y.Q., 2006 Design modification in a collaborative

assembly design environment Computing and

Informa-tion Science in Engineering, 6 (2), 200–208

Marsh, J., Glencross, M., Pettifer, S., and Hubbold, R.,

2006 A network architecture supporting consistent richbehavior in collaborative interactive applications IEEETransactions on Visualization and Computer Graphics, 12(3), 405–416

Noh, S.D., Park, Y.J., Kong, S.H., Han, Y.-G., Kim, G., andLee, K.I., 2005 Concurrent and collaborative processplanning for automotive general assembly InternationalJournal of Advanced Manufacturing Technology, 26 (5),572–584

Pappas, M., Karabatsou, V., Mavrikios, D., and louris, G., 2006 Development of a Web-based collabora-tion platform for manufacturing product and processdesign evaluation using virtual reality techniques Inter-national Journal of Computer Integrated Manufacturing,

Chrysso-19 (8), 805–814

Rycerz, K., Tirado-Ramos, A., Gualandris, A., PortegiesZwart, S.F., Bubak, M., and Sloot, P.M.A., 2007.Interactive N-body simulations on the grid: HLA versusMPI International Journal of High-Performance Comput-ing Applications, 21 (2), 210–221

Shyamsundar, N and Gadh, R., 2002 Collaborative virtualprototyping of product assemblies over the Internet.Computer-Aided Design, 34 (10), 755–768

Trappey, A and Hsiao, D., 2008 Applying collaborative designand modularized assembly for automotive ODM supplychain integration Computers in Industry, 59 (2–3), 277–287.Wang, Q.H and Li, J.R., 2006 Interactive visualization ofcomplex dynamic virtual environments for industrialassemblies Computers in Industry, 57, 366–377

Yang, R.D., Fan, X.M., Wu, D.L., and Yan, J.Q., 2007.Virtual assembly technologies based on constraint andDOF analysis Robotics and Computer-Integrated Manu-facturing, 23 (4), 447–456

Zhen, X.J., Wu, D.L., Fan, X.M., and Hu, Y., 2009 Distributedparallel virtual assembly environment for automobiledevelopment Assembly Automation, 29 (3), 279–289

Mastering demand and supply uncertainty with combined product and process configuration

C.N Verdouwa*, A.J.M Beulensb, J.H Trienekenscand T Verwaarta

a

(Received 18 June 2009; final version received 15 January 2010)The key challenge for mastering high uncertainty of both demand and supply is to attune products and businessprocesses in the entire supply chain continuously to customer requirements Product configurators have proven to bepowerful tools for managing demand uncertainty This paper assesses how configurators can be used for combinedproduct and process configuration in order to support mastering high uncertainty of both supply and demand Itdefines the dependence between product and process configuration in a typology of interdependencies Theaddressed dependences go beyond the definition phase and also include the effects of unforeseen backend eventsduring configuration and execution Based on a case study in the Dutch flower industry, a conceptual architecture isproposed for coordination of these interdependencies and development strategies are identified

Keywords: concurrent engineering; configuration; ERP; flower industry; mass customisation; supply chainmanagement

1 Introduction

Mastering both demand and supply uncertainty is a

key challenge for many companies Markets are

increasingly turbulent and also the vulnerability of

production and logistics processes is growing The

management of uncertainty has been addressed as an

essential task of supply chain management (SCM)

(among others by Davis 1993) The well-known

bullwhip effect shows that amplification of demand

uncertainty can be reduced by supply chain

coordina-tion (Lee et al 1997) There are two main categories of

supply chain uncertainties: i) inherent or high frequent

uncertainties arising from mismatches of supply and

demand; ii) uncertainties arising from infrequent

disruptions to normal activities, such as natural

disasters, strikes and economic disruptions (van der

Vorst and Beulens 2002, Kleindorfer and Saad 2005,

Oke and Gopalakrishnan 2009) This paper is

concerned with the first category of uncertainties,

which can either be demand- or supply-related

(Lee 2002)

For coping with the addressed uncertainties, SCM

literature initially has focused on creating so-called

lean supply chains that efficiently push products to the

market Lean supply chains build upon reduction of

demand uncertainty, especially by product

standardi-sation Customers must choose from a fixed range of

standard products that are made to forecast in high

volumes Business processes in lean supply chains can

be highly automated by Enterprise Resource Planning(ERP) systems (Davenport and Brooks 2004)

In the late 1990s, the then dominant approach ofleanness was criticised more and more It was arguedthat in volatile markets it is impossible to removeuncertainty Companies therefore should accept differ-entiation and unpredictability and focus on betteruncertainty management Agility was proposed as analternative approach that aims for rapid response tounpredictable demand in a timely and cost-effectivemanner (Fisher 1997, Christopher 2000) It is founded

on a mass customisation approach that combines theseemingly contradictory notions of flexible customisa-tion with efficient standardisation (Davis 1989, Pine

et al 1993, Chandra and Kamrani 2004) This is byfabricating parts of the product in volume as standardcomponents, while achieving distinctiveness throughcustomer-specific assembly of modules (Duray et al.2000)

Besides product modularity and flexible assemblysystems (cf Molina et al 2005), product configuratorsare addressed as important enabling technologies(Duray et al 2000, Zipkin 2001, Forza and Salvador2002) Product configurators provide an interface forrapid and consistent translation of the customer’srequirements into the product information needed fortendering and manufacturing (Sabin and Weigel 1998,

*Corresponding author Email: cor.verdouw@wur.nl

International Journal of Computer Integrated Manufacturing

Vol 23, No 6, June 2010, 515–528

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

Ó 2010 Taylor & Francis

DOI: 10.1080/09511921003667706

Forza and Salvador 2002, Tseng and Chen 2006,

Reinhart et al 2009)

Until then, SCM focused on strategies for coping

with demand uncertainty Lee (2002) was one of the

first who stressed the impact of supply uncertainty on

supply chain design Supply chains characterised by

high supply uncertainty require the flexibility to deal

with unexpected changes in the business processes

Disturbances of logistics, production or supply of

materials should rapidly be observed and lead to

process changes, including re-planning and

re-scheduling, purchasing new material, hiring alternative

service providers or negotiating new customer

require-ments The rigid planning and scheduling systems of

traditional ERP systems may cause problems in this

type of supply chain (Akkermans et al 2003, Zhao and

Fan 2007) Modular software approaches, in particular

service-oriented architecture (SOA), have been

pro-posed to overcome these limitations In these

ap-proaches, process models guide the workflow planning

and execution in run-time information systems This

puts the emphasis on process configuration to achieve

the required backend flexibility Process configuration

supports a rapid and consistent specification of the

workflow that is needed to fulfil specific customer

orders (Schierholt 2001 and others) For example, local

deliveries from stock follow a different workflow than

exports that are produced to order Moreover, it

supports reconfiguration of the workflow in case of

unexpected supply events, e.g components that were

originally planned to be produced can be re-planned to

be purchased

Supply chains characterised by both uncertain

demand and supply require a combination of

respon-siveness to changing demand and the flexibility to deal

with unexpected changes in the business processes

Following Lee (2002), in this paper the term ‘agility’ is

used to characterise these types of supply chains In

agile supply chains, demand requirements and supply

capabilities, i.e products and processes including

resources, should be continuously attuned Therefore,

both front-office and back-office systems need to be

flexible and smoothly integrated This paper explores

the application of configurators to both products and

processes to achieve this

The majority of the existing configuration research

focuses either on product or process configuration

However, interdependence among product and process

configuration is relatively under-researched (cf Jiao

et al 2007, Chandra and Grabis 2009) A literature

review, which is presented later, shows that available

literature on this subject focuses on the definition

domain, i.e translation of customer requirements to an

integrated design of products and manufacturing

processes (Jiao et al 2000, de Lit et al 2003, Jiao

et al 2005, Bley and Zenner 2006) However, thepresence of supply uncertainty also results in a highmutual dependence after the definition phase Duringconfiguration and execution, the effects of unforeseenbackend events on the defined product and fulfilmentprocesses must continuously be evaluated, based onthe actual state of the required resources No research

is found that provides an integrated consideration ofthe interdependences during definition, configurationand execution, nor that develops the correspondinginformation architecture for coordination of thisinterdependence using configurators

The present research aims to contribute to this gap

by assessing how configuration software can be usedfor combined product and process configuration tosupport mastering high uncertainty of both supply anddemand More specifically, it aims to: i) identify theinterdependences between product and process config-uration; ii) design an information architecture forcoordination of this interdependence using configura-tors; iii) identify configurator development strategies.The focus is on the order fulfilment cycle that startswith configuring orders in interaction with customersand ends with delivering the finished goods (Lin andShaw 1998, Croxton 2003)

In the remainder of this paper, first an account isgiven of the applied research method Next, the paperintroduces problem context of the case firm, which is

a typical example of a company operating in agilesupply chains Subsequently, an overview is provided

of the literature about the use of configurators forproducts and processes and a typology of its inter-dependencies is defined The case study results arethen presented The paper concludes by addressingchallenges for future development and summarisingthe main findings

2 Research methodThe research used a design-oriented case study method

to answer the research question addressed in Section 1.Design-oriented research aims to develop a body ofgeneric knowledge that can be used in designingsolutions to management problems (van Aken 2004)

It is a foundational methodology in informationsystems research (Hevner et al 2004) Design-orientedresearch is typically involved with ‘how’ questions, i.e.how to design a model or system that solves a certainproblem A case-study strategy fits best for this type ofquestions, in particular in case of complex phenomenathat cannot be studied outside its context (Benbasat

et al 1987, Yin 2002) This characterises the presentresearch, because it focuses on the interdependencesbetween product configuration, process configurationand the planning and control of fulfilment Therefore,

it was chosen for an in-depth explorative case study

research that puts the different related topic areas into

context In such a case study, it makes sense to focus

on an extreme situation that clearly highlights the

process of interest (Eisenhardt 1989, Yin 2002) In

present research, this is the existence of supply

uncertainty in addition to demand uncertainty

There-fore, the present authors searched within a sector that

is inherently involved with high supply uncertainty, i.e

the Dutch flower industry Firms in this sector face

high supply uncertainty because of the dependence on

the growth of living materials Production processes

are, therefore, vulnerable to weather conditions, pests

and other uncontrollable factors Next, a firm within

this sector was selected, which was characterised by

high demand uncertainty Additional criteria were

product variety and practical reasons, in particular the

firm’s willingness to cooperate and the authors’

familiarity in the domain

Data collection was done in semi-structured open

interviews with managers and employees of the case

company and additional desk research In total, 14

persons were interviewed in nine meetings (five

managers and nine employees) The division of roles

is as follows:

Management: sales (1), finance (1), logistics (1),

production (1), CEO (1)

Employees: order processing (1), planning (2),

expedition (1), ICT (1), production seedlings (2),

production cuttings (2)

The questionnaire comprises four main parts:

supply chain structure, business processes, control

and information management Every section includes

open questions both for mapping and evaluation (see

Appendix 1) Three in-depth interviews were held

covering the complete questionnaire The subsequent

interviews focused on specific business processes and

were combined with observation of the company’s

operations and systems

The research was organised as follows First, the

dependence between product and process

configura-tion was defined in a typology of interdependencies

based on literature review Second, the case-study firm

was investigated in interviews and additional desk

research Next, the investigation results were matched

with the developed theoretical framework to define the

basic design requirements The researchers then

designed a conceptual information architecture for

combined support of both product and process

configuration The designed architecture was tested in

a proof of feasibility implementation at Sofon, a Dutch

configurator vendor, and evaluated by the

manage-ment of both the case-study firm and Sofon Finally,

general development strategies were abstracted fromthe case findings based on the developed theoreticalframework

3 Configuration in the Dutch flower industryThis section introduces the case firm and its need forproduct and process configuration

3.1 Dutch flower industryThe Dutch flower industry is traditionally a strong andinnovative sector with a leading international com-petitive position and a great impact on the nationaleconomy It is internationally renowned as a strongcluster (Porter 1998) that produces cut flowers andpotted plants, mainly in greenhouses In particular,production of potted plants has many similarities withmanufacturing It is also a form of discrete production,

in which products are assembled from plants, pots, decorations, labels and packaging The creation

flower-of potted plants also has some features flower-of continuousproduction, because of the process of continuousgrowth, but potted plants remain discrete units,traceable at single product level

The extent to what processes are order-drivendiffers a lot, not only among different companies butalso within firms For the spot market, products aremade to stock and distribution is either to order(usually via traders) or anticipatory (usually viaauctions) For other cases, plants are often produced

to forecast, while assembling, labelling and packagingare order-driven

The flower industry is characterised by highuncertainty of both demand and supply Supplyuncertainty is high, because chains are vulnerable toproduct decay, weather conditions, pests, traffic con-gestion and other uncontrollable factors Further,demand uncertainty is also high because of weather-dependent sales, changing consumer behaviour andincreasing global competition amongst other things.This results in high variability of supply capabilitiesand demand requirements in terms of volume, time,service levels, quality and other product characteristics

3.2 Case company profileThe case company is a global supplier of a wide range

of young potted plants It is a rapidly growingcompany, with 350 staff and with production locations

in Holland, Brazil, Kenya, Israel and Zimbabwe.Annually, over 100 million young plants are delivered

as input material to growers or wholesalers

The firm is characterised by high product variety Itproduces about 800 varieties in six main categories,International Journal of Computer Integrated Manufacturing 517

including begonia and cyclamen In addition, over 400

varieties are sourced from other producers to offer a

complete assortment Varieties differ, amongst other

things in colour, shape and growing characteristics

The firm propagates young plants in two basic ways: as

seedlings or cuttings Seedlings can be sold at different

stages of the growing process Cuttings can be sold

rooted or unrooted and in different sizes All young

plants can be delivered in different types of trays

Furthermore, delivery conditions vary For example,

due to product-inherent characteristics, some varieties

can only be delivered in specific periods and quality

and prices are often time-dependent In addition,

royalties differ per variety and per continent

Also process variety is high Production differs

between seedlings and cuttings For seedlings, seeds are

sourced from breeders, seeded in trays and budded

Budded seeds can be sold directly or grown further

Seedlings are mostly seeded to customer order, but also

produced to forecast or sourced to order (especially for

specific variety mixtures) Cuttings are mostly

pro-duced by the firm, but are also sourced from third

parties Production of cuttings starts with propagation

and growing of parent plants, which is done in

southern countries for reasons of climate and labour

costs After almost 2 years, cuttings can be harvested

They are shipped directly to customers or transported

to Holland for rooting Unrooted cuttings can be

stored for 10 days at the most, including 3 days for

transportation The company strives for order-driven

harvesting and rooting of cuttings, but production to

stock also occurs Furthermore, logistics are complex,

due to the global distribution of both production

locations and customers, combined with high

require-ments concerning delivery lead-times and flexibility

3.3 Need for combined product and process

configuration

The interviews indicated that the case company is

characterised by high uncertainty of both demand and

supply Demand requirements (about product features,

quality and service levels) are diverse and difficult to

predict Also, predictability of the demand amount and

time is low, although basic seasonable patterns can be

determined Moreover, the lead-times, yields and

qualities of production very much depend on the

growth of living materials

The company deals with this high uncertainty by

providing variety in their product assortment and

flexibility in meeting customer demands with regard to

product specifications and delivery schedules To date,

it has relied heavily on improvisation by experienced

employees However, since the company is growing,

they face problems in keeping this manageable, which

set limits to further growth As a consequence, theinterviewees in particular stressed the lack of tools forcustomer requirement definition based on real-timeinformation of the supply capabilities, as well asflexible back-office systems for (re)planning, (re)scheduling and monitoring of order fulfilment Theaddressed most urgent bottlenecks are:

Knowledge of production processes and options

to reconfigure these processes is only implicitlyavailable in the minds of some experienced staffmembers This problem is manageable with thefirm’s current scale, but inhibits further growth Information systems are fragmented and poorlyintegrated They require a lot of manual data re-entry Information inconsistency leads to largersafety buffers than strictly required and manyredundant data checks and duplicate registra-tions are performed

Mid-term planning is not coordinated withoperational data, due to a lack of systemintegration

The company’s management assessed existing ERPsystems for solving these problems, but evaluated them

to lack the required flexibility Therefore, the firmdecided to consider implementation of configurationsoftware for products and processes, in combinationwith an ERP system, as a possible option to masteruncertainty

4 Role of configurators in supply chain managementThis section provides some conceptual backgroundabout the use of configurators and defines thedependence between product and process configura-tion in a typology of interdependencies

4.1 Product configurators in responsive supply chainsConfigurators have emerged from the development ofrule-based product design in the field of artificialintelligence A well-known early application was R1, aproduct configurator for VAX computers (McDermott1981) A product configurator is a tool that guidesusers interactively through specification of customer-specific products (Sabin and Weigel 1998, Forza andSalvador 2002, Tseng and Chen 2006, Reinhart et al.2009) Configurators generate specific product variants

by combining sets of predefined components andspecifying features according to permitted values.Next, they check the completeness and consistency ofconfigured products based on rules that define theinterdependencies between components or features.Product configurators are based on generic product

models, which define the class of objects that can be

configured (Hegge and Wortmann 1991)

Currently, configurators play an important role in

responsive supply chains, which are characterised by

high demand uncertainty and low supply uncertainty

(Lee 2002) They are widely used for product

config-uration to enable rapid response to customer demands

In interaction with the user, the software generates

consistent and complete specifications of customised

products, taking into account both the customer’s

requirements (e.g functional specifications and

deliv-ery conditions) and feasibility of production, sourcing

and delivery Together with the product specification,

current configurators can produce commercial offers

and draft contracts and schedules and contracts for

support and maintenance of the product The software

can be designed for use either by a sales representative

of the supplier or by a customer, e.g through the

Internet In both cases, the configuration process

results in a quick and effective order specification

that can directly be entered into the production

planning and scheduling systems

4.2 Configuration in agile supply chains

Next to demand uncertainty, agile supply chains are

also characterised by high uncertainties at the

supply-side (Lee 2002) High supply uncertainty makes great

demands on the flexibility of supporting information

systems The development of modular software

ap-proaches especially has been advocated for realising

this flexibility (for example, Verwijmeren 2004) SOA is

the latest development in software modularity (Wolfert

et al 2010) In a SOA approach, business process

models are leading in routing event data amongst

multiple application components that are packaged as

autonomous, platform-independent services (Erl 2005,

Papazoglou et al 2007) Consequently, new or adapted

business processes can be supported without changing

the underlying software Induced by the emergence of

SOA, ERP vendors have also begun to modularise

their software (Møller 2005, Loh et al 2006)

The leading role of business processes in modular

software approaches puts emphasis on rapid

config-uration of processes in achieving flexibility The

con-cept of process configuration is introduced by

Schierholt (2001), who applied the principles of

product configuration to support process planning

Rupprecht et al (2001) and Zhou and Chen (2008)

described approaches for automatic configuration of

business process models for specific projects Jiao et al

(2004) formalised the modelling of process

configura-tions for given product configuraconfigura-tions Verdouw et al

(2010) argue that reference process models should be

set up as dynamic configurable models to enable

ICT mass customisation and they assess the readiness

of existing models Furthermore, the ERP vendor SAPhas addressed process configuration to manage thecomplexity of their reference process models that areused as a basis for system implementation Theyconducted extensive research to make these modelsconfigurable (Dreiling et al 2006, Rosemann andvan der Aalst 2007) Building upon this, Rosa et al.(2007) proposed a questionnaire-driven approach toguide users interactively through process modelconfiguration

Nevertheless, the majority of existing literaturefocuses either on product or on process configuration.The mutual dependence between product and processconfiguration is relatively under-researched (cf Jiao

et al 2007, Chandra and Grabis 2009) The papers,found in the literature review, all focus on the definitiondomain, i.e translation of customer requirements to anintegrated design of products and manufacturingprocesses Jiao et al (2000) put forward an integratedproduct and process model that unifies bill-of-materialsand routings, called generic bill-of-materials-and-op-erations Jiao et al (2005) proposed a product-processvariety grid to unify product data and routing infor-mation de Lit et al (2003) introduced an integratedapproach for product family and assembly systemdesign Bley and Zenner (2006) developed an approach

to integrate product design and assembly planning

As argued before, the presence of supply tainty also results in a high mutual dependence afterthe definition phase During configuration and execu-tion, the effects of unforeseen backend events on thedefined product and fulfilment processes must con-tinuously be evaluated, based on the actual state of therequired resources However, no research is found thatexplicitly considers the interdependences during defini-tion, configuration and execution and that developsthe corresponding information architecture for co-ordination of this interdependence using configurators.Therefore, the next section first develops a typology ofproduct and process interdependences based onorganisational literature

uncer-4.3 Typology of interdependences between productand process configuration

Dependence is a central notion of the General SystemsTheory This theory argues that the whole of a system ismore than its parts, because of the existence ofdependencies between their elements (Bertalanffy1950) Thompson (1967) was one of the first to applythis idea to organisational theory He distinguishedthree basic types of dependency, pooled, sequential andreciprocal interdependence, which require differentcoordination modes: coordination by standardisation,International Journal of Computer Integrated Manufacturing 519

by plan and by mutual adjustment His work is refined

by many others, all focusing on coordination of generic

dependencies between organisational subunits Malone

and Crowston (1994) have introduced different types of

dependencies between activities and resources They

distinguish between flow, sharing and fit dependencies

(see also Malone et al 1999) Flow dependencies arise

whenever one activity produces a resource that is used

by another activity (precedence relation) Sharing

dependencies occur whenever multiple activities all

use the same resource Fit dependencies arise when

multiple activities collectively produce a single

resource

If these interdependencies are applied to product

and process configuration, distinction should be made

between different decision levels, i.e definition,

config-uration and execution First, in the definition phase

designers predefine reference product and process

models These are generic models, or family models,

which define the possible product and process

compo-nents and which include rules that define the possible

combinations of components A product reference

model is constrained by the available business processes

as defined in process reference models Conversely, a

process reference model must contain the business

processes that produce the variety of products as

defined in product reference models Second, the

configuration phase starts when a customer order

request comes in A customised product is configured

in interaction with the customer, taking into account

whether the enabling business process can be

config-ured Therefore, the required input products and

capacity must be available to promise The result is

an accepted order, which triggers configuration of the

business processes that fulfil the order These might

include distribution activities (make to stock), and

production activities (assemble/make to order) and

engineering activities (engineering to order) Finally,

the execution phase comprises planning, scheduling

and completion of the configured business processes.The progress is monitored continuously and, ifnecessary, the product and process configurations areupdated

Figure 1 more precisely defines the interdependenceamong product and process configuration in atypology of dependencies This typology is an applica-tion of the categorisation of Malone and Crowston(1994) and Malone et al (1999) as discussed above.Product configurators are primarily means forcoordination of fit dependencies: assembling consistentproduct variants that meet specific customer require-ments from available components and options Analo-gously, process configuration coordinates the assembly

of consistent process variants from available activities

or services The alignment of product and processconfiguration requires coordination of precedence(flow) dependencies: process configuration is condi-tional for product configuration and vice versa.Furthermore, process configuration depends on opera-tional execution because fulfilment of the configuredprocess needs capacity and input products Morespecific, the defined interdependencies are:

(1) Product assembling: multiple product modulesare required to produce a single product (fitdependency) Product configurators are pri-marily means for coordination of this depen-dency They specify components, options,interfaces and interdependency rules in refer-ence product models and guide customer-specific configuration of product variants.(2) Process rules precedence: process properties setconstraints to possible product configurations.Consequently, process reference models are aprecondition for product reference models(flow dependency) This dependency is mostlycoordinated by mutual adjustment of productand process models by designers, ideally

supported by tools that ensure consistency of

both model types

(3) Order precedence: specific product

configura-tions (order information) are input for

config-uration of specific fulfilment processes (flow

dependency) Therefore, order information

must be interpretable by back-office systems

for production and distribution This

depen-dency can be coordinated by standardisation of

order data in an executable form, including

bill-of-materials

(4) Process assembling: multiple activities, i.e

process modules, are required to compose a

single process (fit dependency) Process

config-uration is primarily a mechanism to coordinate

this dependency It specifies the activities,

interfaces and interdependency rules in

refer-ence process models and guides configuration

of order-specific processes

(5) Process precedence: the output of the process

configuration task is conditional for the

plan-ning and scheduling of the fulfilment (flow

dependency) Execution of a configured process

consumes input products (raw material or

semi-finished products) and uses capacity This

dependency can be coordinated by

standardisa-tion of configured processes in a model format

that is interpretable by planning and scheduling

systems

(6) Product precedence: for execution of a

fulfil-ment process, the required input products must

be available (flow dependency) This

depen-dency can be coordinated by integration with

planning and scheduling mechanisms

(7) Capacity precedence: for order-driven

pro-cesses, the required capacity must be available

(flow dependency) This can be coordinated by

integration with planning and scheduling

mechanisms

(8) Capacity rules precedence: the characteristics of

used capacity (e.g machine set-up, other facility

layouts and human resource competences) set

constraints for the possibilities for process

configuration (flow dependency) This can be

coordinated similarly to process rules

depen-dencies: mutual adjustment of capacity layouts

and process models by designers ideally

sup-ported by tools that ensure model consistency

The last dependencies to be mentioned are related

to the operational execution of configured processes:

(1) Product consumption: multiple configured

pro-cesses all use the same input products (sharing

dependency)

(2) Capacity usage: configured processes for ple orders all use the same capacity (sharingdependency)

multi-(3) Capacity assembling: multiple capacity unitsare required to set up specific layouts (fitdependency)

These last three dependencies are coordinated byplanning and scheduling systems They do not directlyimpact product and process configuration (only viaproduct, capacity and process precedences) and arethus beyond the scope of this paper

5 Information architecture for combined product andprocess configuration

This section describes a conceptual informationarchitecture for combined support of both productand process configuration, including a proof offeasibility implementation in a configurator

5.1 Basic design requirementsThe uncertainty of both demand and supply of the casecompany is high, as Section 3 demonstrates Demandrequirements (about product features, quality andservice levels) are diverse and difficult to predict Alsopredictability of the amount and time is low, althoughbasic seasonable patterns can be determined More-over, the lead-times, yields and qualities of productionvery much depend on uncontrollable factors

In order to make this variability manageable, thesolution to be designed must support coordination ofthe high interdependence between the company’sproducts and processes during:

definition: it must be possible to define integratedreference models, which cover the variety of thefirm’s products and enabling processes andwhich take into account the constraints arisingfrom its specific process characteristics;

configuration: it must be possible to configurecustomised products and the accompanyingprocesses, in interaction with the customer andtaking into account whether the required inputproducts and capacity are available to promise; execution: it must be possible to implement theconfigured business processes in the company’sbackend systems, to monitor its progress andupdate product and process configurations ifnecessary

More specifically, these basis requirements implythat the design must support coordination of thedependences as developed in previous section Table 1International Journal of Computer Integrated Manufacturing 521

identifies these dependencies for the case company by

matching the investigation results with the defined

typology The remainder of this section develops a

corresponding information architecture, including a

proof of feasibility implementation in the configurator

Sofon

5.2 Information architecture for product configuration

Sofon Guided Selling is a model-based product

configurator (Sofon B.V., Son, The Netherlands) It

provides functionality for the definition of

question-naires that guide users interactively through

require-ment specification and translates this information

to product configurations in the form of

bills-of-materials, quotation calculations, visualisations and

document generation Most users utilise Sofon as a

front-office system, in combination with an ERP

system for the back-office Figure 2 illustrates theunderlying information architecture

The focus is on coordination of product assemblingdependencies Therefore, functionality is provided tospecify the product range, possible features and rulesthat define permitted selections in reference productmodels Additionally, other order specifications such

as delivery dates can be defined here Product expertscan enter configuration rules into the configurator’srepository Product data (bill-of-materials, part num-bers, prices) and process data (routing, lead times,production cost) can be copied from ERP master data,

to ensure that production orders will be in terms thatcan be interpreted by ERP systems (process andcapacity rules precedence)

Questionnaires are then generated that guideconfiguration, either directly by the customer orthrough a sales representative The configured product

2) Process rules

precedence

processes (about 14 weeks for seedlings, 5–6 weeks for rooted cuttings, 10 days for unrootedcuttings)

road, rail or air freight, and different requirement to shipping documentation)

to be reserved

the cuttings order fulfilment

cuttings8) Capacity rules

precedence

cuttings9) Product

consumption

that can be budded synchronously, consequently capacity shortage for an urgent order mightresult in rescheduling another order

and other customer specifications (orders, bill of

material) are generated in a format that can be

executed by ERP systems (order precedence) Also,

basic order-specific routings can be generated, which

serve as a basis for planning and scheduling (process

precedence)

For the case firm, the reference product model

includes product categories (including begonia and

cyclamen), specific varieties and product features, such

as budded seeds or grown up, cutting size, rooted or

unrooted, possible tray types, delivery conditions and

royalty types Figure 3 presents a simplified example in

Sofon

The figure shows that the generic model is defined

in two ways The main part is the definition of like questionnaires in the language of customers Thegeneric questionnaire is defined to the left of the screenand possible answers are shown to the top-right of thescreen At the bottom, the product model is specified as

wizard-a generic bill-of-mwizard-ateriwizard-als thwizard-at is executwizard-able by ERPsystems During configuration, selections made in thequestions are specified automatically into this bill-of-materials For example, based on the selection of thecolour red, the variety ‘begonia elatior baladin’ isdefined (see Figure 3: article code BE72)

5.3 Information architecture for combined productand process configuration

Currently, configurators such as Sofon focus onproduct configuration in the responsive segment Agilesupply chains require combined product and processconfiguration Two essential differences can be dis-tinguished: i) introduction of process configurationbetween product configurators and planning andscheduling systems; ii) dynamic alignment of resultinginterdependencies In the case study, Sofon was used todevelop an information architecture for this and toevaluate the feasibility of configurators Figure 4shows the resulting conceptual model

Analogous to product configuration, the focus ofprocess configuration is on the coordination of process

International Journal of Computer Integrated Manufacturing 523

assembling dependencies, i.e to assemble specific order

fulfilment processes from multiple activities (process

modules) Therefore, standard process models can be

specified and the composition of customer-specific

processes can be guided by configurator tools

How-ever, the important difference with product

configura-tion is that most informaconfigura-tion required for process

configuration is available in the system Two important

information sources can be distinguished for process

configuration: customer orders (output of product

configuration) and availability of required input

products and capacity (output of ERP back-office

system) Neither of these types of information needs to

be specified manually during process configuration

Although Sofon focuses on companies in theresponsive segment that do not face high supplyuncertainties, the tool can be applied to configureprocesses in the same way that it is used to configureproducts Figure 5 presents an example for the casecompany using Sofon’s existing functionality

It shows that there are three additional questionsfor the configuration of the young plant order ful-filment processes, all of which are answered auto-matically The questions concerning capacity andproduct availability are queries to ERP back-officesystems The question ‘how far order-driven?’ isanswered by an automatic calculation using theretrieved data about product and capacity availabilityand information about the required vs possible lead-time The required delivery lead-time is as specifiedduring product configuration The possible lead-time isthe sum of all order-driven fulfilment processes Thecalculation result is input for activity specification inthe generic routing (i.e bill of activities) that isexecutable by ERP systems (see right-bottom ofFigure 5)

Consider, for example, an illustrative order forcyclamen The customer specification, resulting fromproduct configuration, shows that this is an order for

2000 budded ‘cyclamen miniwella twinkle blanc’ to bedelivered within 4 days in 66–66–44 trays to Hamburg,Germany Operational ERP data show that these are

in stock and that distribution will take 2 days Thus,

only distribution activities are on customer order and

these activities are selected for this order (see Figure 5:

D2.5 and further)

Consider another simplified example: an order for

begonia cuttings The configured order shows that this

is an order for 5000 rooted ‘begonia eliator baladin ‘to

be delivered in 7 weeks in 72–72–44 trays to Latina,

Italy Operational ERP data show that the required

cuttings are available at parent plants in Brazil and

that the required air freight and greenhouse capacity is

also available The lead-time of rooting cuttings from

these parent plants is 5 weeks The total lead-time from

harvesting until delivery at the customer site is 6 weeks

and 3 days This is less than 7 weeks, so all activities

from harvesting onwards can be on customer order

Consequently, all activities for production of cuttings

and for distribution are selected in the generic routing

(see bottom-right of Figure 5)

5.4 Configurator development strategies

The previous analysis shows that product

configura-tors can also be used to support process configuration

Below, it is evaluated more precisely to what extent the

identified basic requirements can be met by existing

configurators by discussing how coordination of the

defined interdependencies is supported (see Figure 1):

(1) Product assembling is well supported, since this

is the traditional focus of configurators For

example, Sofon provides rich functionality for

defining generic product models in wizard-like

questionnaires and accompanying rules, and

generic bill-of-materials

(2) Process rules precedence requires solid

integra-tions with back-office systems and mechanisms

to prevent redundant process logic or to ensure

consistency For example, Sofon provides

functionality to copy master data from ERP

packages, but alignment has to be done

manually by product experts; consistency

checks are not supported

(3) Order precedence: configurators and ERP

systems must be technically integrated and

order-related data must be defined in a format

that is executable by back-office systems

Especially in agile supply chains, functionality

is required for reconfiguration of order-related

data if changes in the back-office occur For

example, Sofon contains rich functionality for

defining standard orders and accompanying

bill-of-materials and it provides standard

ap-plication connectors for ERP packages

How-ever, reconfiguration of adjusted requirements

after contract conclusion is not supported

(4) Process assembling: this could be supported

by applying available product configurationfunctionality to processes However, adequateprocess configuration requires rich function-ality to specify reference process models and

to configure business process models based onconfigured orders and operational back-officedata In existing questionnaire-based productconfigurators, this functionality might berather basic For example, in Sofon, genericroutings for customer-specific processes can beconfigured, but this functionality just listsactivities, possibly including fixed lead-times

It does not specify possible interactions andsequences among activities and it is notpossible to derive activity lead-times fromoperational data

(5) Process precedence: configured processesshould be executable in back-office systems toorchestrate order-specific fulfilment, includingproduct reconfigurations For compatibilitywith different software environments, processmodels should be configured in XML-standardnotations (i.e BPMN and BPEL) that areexecutable in any SOA-compliant back-officesystem or integration platform For example, inSofon order-specific routings can be configuredand executed by planning and schedulingsystems However, as argued at the previousdependency, this is rather basic decompositioninformation Sofon does not yet support theconfiguration of BPMN and BPEL processmodels

(6) Product precedence: in product configurators,product availability data can be incorporatedinto configuration, e.g to determine the avail-ability to promise This functionality could also

be used for process configuration For example,

in Sofon it is possible to define questions thatretrieve data automatically from back-officesystems

(7) Capacity precedence: particularly in the case oforder-driven production, capacity availabilitydata are also needed This can be retrieved inthe same way as product availability data.(8) Capacity rules precedence: in this case, requiredfunctionality is similar to process rules pre-cedence; solid integration with back-office andmechanisms to prevent redundant capacitylogic or to ensure consistency

(9) Product consumption, (10) Capacity usage and(11) Capacity assembling: these dependenciesare supported by back-office systems for plan-ning and scheduling and are beyond the scope

of configurators

International Journal of Computer Integrated Manufacturing 525