Negotiation in distributed production planning environments

This article was downloaded by: [Florida Atlantic University]On: 18 September 2013, At: 08:42Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of ProductionResearchPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tprs20

Negotiation in distributed productionplanning environmentsGiovanna Lo Nigro a , Manfredi Bruccoleri a & Giovanni Perrone aa Dipartimento di Tecnologia Meccanica, Produzione e IngegneriaGestionale, Faculty of Engineering, Università di Palermo, Vialedelle Scienze, 90128 Palermo, ItalyPublished online: 06 Oct 2011.

To cite this article: Giovanna Lo Nigro , Manfredi Bruccoleri & Giovanni Perrone (2006) Negotiationin distributed production planning environments, International Journal of Production Research,44:18-19, 3743-3758, DOI: 10.1080/00207540600575787

To link to this article: http://dx.doi.org/10.1080/00207540600575787

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/tprs20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/00207540600575787

http://dx.doi.org/10.1080/00207540600575787

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

International Journal of Production Research,Vol. 44, Nos. 18–19, 15 September–1 October 2006, 3743–3758

Negotiation in distributed production planning environments

GIOVANNA LO NIGRO*, MANFREDI BRUCCOLERIand GIOVANNI PERRONE

Dipartimento di Tecnologia Meccanica, Produzione e Ingegneria Gestionale,

Faculty of Engineering, Universita di Palermo, Viale delle Scienze, 90128 Palermo, Italy

Mass customization and global competition push enterprises to adopt properbusiness models able to capture all the opportunities arising from emergingcompetition rules. An increasing number of industrial enterprises distribute theirproduction capacity world wide to achieve lower production costs, lowerdistribution costs (due to the closer proximity to customers), and deeperknowledge of customer needs. As a drawback, coordination of the differentproduction plants and the balance among plants and enterprise goals representa critical issue in the network management. In this context the paper looks at theproduction planning problem, adopting a traditional hierarchical time-basedperspective in the analysis of the global process and suggesting a decentralizedplanning approach to deal with the originated subtasks related to different timehorizons. In particular, the paper suggests a production planning architectureable to highlight relationships among subtasks’ variables in which mechanismsassure consistency among solutions of different planning levels. Moreover, thepaper proposes negotiation frameworks as effective tools to manage productionplanning subtasks.

Keywords: Negotiation; Distributed Production Planning (DPP); Multi AgentSystem (MAS)

1. Introduction

The consequences of mass customization, such as shorter and shorter productlife-cycles and low-cost variety, have brought critical pressures to improveproduction efficiency, responsiveness to market changes, and substantial costreduction. Some of these issues can be achieved through innovation relatedto production systems (advanced manufacturing systems, reconfigurable manufac-turing systems, etc.) or to operations and organization coordination. Distributionof activities involved in a business process is a successful management expedient toexamine with current market characteristics. Indeed, the need for global competitivestrategies, for rapid response to market changes, for costs and time to marketreductions, and for highly customized products requires the enterprise value chain

*Corresponding author. Email: [email protected]

International Journal of Production Research

ISSN 0020–7543 print/ISSN 1366–588X online � 2006 Taylor & Francis

http://www.tandf.co.uk/journals

DOI: 10.1080/00207540600575787

Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

to be more and more distributed. This results in manufacturing enterprises in severalindustries either splitting the production capacity geographically or working togetherin supply chain organizations involving several independent entities (Perrone 2005).These network organizations raise the need for new business models and manage-ment tools (Abid et al. 2004). On the other hand, ICT and market globalizationoffer the opportunity of adopting new business models to manufacturing enterprises.The coordinated co-working among various responsible units is the driving elementof the decision-making in the network enterprise.

In this context, effective operations management is the challenging task andinvolves distributed problem-solving tasks. Specifically, in production planningthe concern on internal production planning is replaced by the complexity ofexternal supply planning, since this supports the network operation. As soon asa manufacturing unit tries to achieve coordination with its partners, it quickly facesdifficulties associated with different operational conventions, locally specificconstraints, software legacy and properties, conflicting objectives and misalignedincentives. Even in a Distributed Production System (DPS) consisting of manufac-turing units belonging to the same firm, coordination is a critical issue. Coordinationcan be achieved easily through standardization. In heterogeneous contextsstandardization enables the sharing of activities which belong to the same processand it is useful to manage interdependencies. In order to achieve the enterprise goal,distribution needs integration and coordination among contributions from differententities and this requires that different systems share an ontology providinga common semantics. In literature integration issues have been largely treated andthree levels of integration are recognized: physical, application and businessintegration (Chen 2005). Michel (1997) considers that integration can be obtainedin terms of: data, organization (modelling of systems and processes) and commu-nication. Integration can also be developed through consistent enterprise-widedecision making (Doumeingts et al. 1998).

In a distributed environment decision-making processes can be treatedfollowing two basic approaches: a centralized one in which a unique decision-making entity, basing on a hierarchical structure with a bottom-up procedure,plans activities at lower levels or a decentralized one in which autonomous entities,basing on an etherarchical structure, participate actively in the decision-makingprocess.

The first approach presents well-known drawbacks (McEwan and Sackett 2001,Rahimifard 2004), while in order to adopt the second one Multi Agent System(MAS) tools are strictly required (Perrone et al. 2003).

Agent Based Manufacturing focuses on bridging the agent theory to manufac-turing systems and can be defined as a design approach that tries to describe thebehaviours of distributed operations and decision-making units in manufacturingsystems (Huang and Nof 2000); in a physically distributed production environmentMAS can give a more fitting contribution. MASs reflect the distributed andautonomous nature of distributed systems providing a natural way to design andimplement such environments (Karageorgos et al. 2003). MAS techniques have beenlargely used for their suitability in modelling complex systems involving multipleautonomous agents with internal knowledge and reasoning engines whichcommunicate and negotiate with each other by exchanging messages accordingto specific negotiation protocols.

3744 G. Lo Nigro et al.

Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

The aim of this paper is to examine production planning issues in DPS andto suggest a negotiation support system (NSS), based on MAS technology,to manage the related decision processes.

The paper is structured as follows: section 2 focuses on distributed productionsystems (DPS) and defines a specific DPS model. Production planning issues in DPSare discussed in section 3. Section 4 is dedicated to a brief overview on thenegotiation process and introduces the classification scheme adopted to design theNSS and compare it with the classical classification system proposed in scientificliterature. The negotiation support system and the related agents architectureare proposed in section 5. Section 6 draws some conclusions and indicates futureresearch paths.

2. Distributed production systems

Market globalization has offered companies the possibility of splitting theirproduction capacity geographically; business opportunities lead companies towork together in temporary organizations; in the same firm, business units behaveas autonomous profit centres and compete with each other for the productioncapacity allocation.

Three strategic options can be adopted jointly and can generate a complex varietyof enterprise organization models. In particular, the first option in literature isknown as the multi-site production system or the distributed production system;the second is known as virtual enterprise; and the third embraces holonicmanufacturing system or, in a broad sense, the divisional organization, whilecombinations of them are not deeply investigated. A common problem arisingin each of the considered configuration or in configurations originating froma combination of them is the degree of autonomy that needs to be embedded in eachpotentially autonomous entity.

Members of a virtual or real enterprise need to be properly coordinated toachieve reduction in lead times and costs, alignment of interdependent decision-making processes, and improvement in the overall performance of each member,as well as of the entire enterprise. Distributed firms in which each entity has a certaindegree of autonomy need to reconcile all the contributions from all over the world.

A distributed enterprise, that is an enterprise with multi-site production facilities,presents some characteristics similar to a Virtual Enterprise (VE): as a VE consistsof independent functional units behaving like a single company, in a distributedenterprise units belonging to the same enterprise behave like autonomous unitslooking for local goals in the solution domain framed by a global decision-maker.This relies upon the belief that empowerment and self-control in heterarchicalsystems provide increase in agility and flexibility (McEwan and Sackett 2001);meanwhile, with respect to the hierarchical ones, they allow a drastic reductionin computational efforts to find the right solution (Rahimifard 2004).

The distributed production system which has been considered here is inspiredby a world-wide electronic components company, which is a world leader indeveloping and delivering semiconductor solutions. It consists of geographicallydispersed reconfigurable manufacturing units (Bruccoleri et al. 2003a).

Negotiation in distributed production planning environments 3745

Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

Generally, companies employ geographically dispersed parallel manufacturingfacilities for reasons such as savings on transportation costs/time and/r to improvecustomer service by locating the plant closer to the customer (Kanyalkar andAdil 2005).

The considered company has a divisional organization and the top managementis named corporate; each division is named group; it is in charge of a product familycommercialization and it has its own objectives. Each group can fulfil the collectedorders thanks to the reconfigurable characteristics of the manufacturing plants.

Reconfiguration capabilities offer clear advantages to the enterprise whileincreasing the complexity of production planning activities.

As a result, the considered company has two of the key organizationalcharacteristics mentioned at the beginning of this section: it has a strong divisionalstructure and it has a multi-site production capacity. A decentralized coordinationarchitecture and, specifically, agent-based technology are strictly required forapproaching such complexity, as will be shown in the next sections.

3. Production planning in DPS

In a multi-site production set up, plants can be parallel (each producing thefinished products and supplying the market) or serial (some plants producingintermediate products supplying other plants, which convert them into finishedproducts). The parallel multi-plants production planning problem can be app-roached in two ways: a static approach which assigns a priori plants to specificproducts (in the considered case to particular groups) or a dynamic one which usesan integrated production planning that assigns a plant to a product basing on thecurrent (exogenous and endogenous) conditions. Usually the choice depends onstrategic decisions: in order to adopt dynamic allocation, production sites shouldpossess the technological capacity necessary to realize all the product families.In the case of reconfigurable manufacturing sites, the dynamic approach ismore suitable.

Moreover, in the case of a production site belonging to different enterprisesthe integration feasibility becomes a critical aspect.

As a decentralized production system aims at achieving goals derivingfrom localization, production management policies cannot adopt the classichierarchical approach fully. System complexity and time-based competition pushtowards sharing decision-making processes. Production planning and controlapproaches usually adopt different tools depending on the considered timehorizon and maintain, anyway, a centralized vision. Usually a manufacturingorganization, after having planned long-term activities (strategic level, capacityplanning) needs medium-term (tactical level, aggregate production planning) andshort-term plans (operational level, master production schedule): these plansdiffer in the type of information available at each stage. As discussed inKanyalkar and Adil (2005), there are different ways of integrating the aggregateand detailed planning problems. In the proposed approach, conceived for a make-to-order environment, aggregate and detailed plans are found in subsequentsolutions and the higher-level solution imposes constraints on the lower-levelmodels.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

To summarize, the following critical aspects should be taken into considerationduring production planning in DPS:

parallel/serial and static/dynamic approach;degree of autonomy of each plant;decomposition of the entire process in different levels;coordination of different level variables; andcoordination tools at each planning level.

The research presented here proposes a decision support system for productionplanning activities structured in different levels (time horizons) according to aclassical view point, providing the active involvement of the interested actors foreach step. This last requirement promotes decision autonomy: autonomy becomesa key issue in achieving the dynamic dimension required in the actual marketscenario. Also, as stated above, the decision-making process adopted in productionplanning activities should always guarantee global satisfaction level by meansof coordination, and this is accomplished by negotiation as explained in section 4.The proposed DSS is suitable both for serial/parallel and dynamic/static approach.

3.1 PP in the considered industrial case

In the considered example, every year, the corporate level assigns to the companygroups (responsible for product families such as electronic memories) a certain levelof the total production capacity (called capacity ownership) based on long-termdemand forecasting and products’ strategic positioning. Every three months thegroups, after having collected backlog and forecast orders coming from the regionaldivisions, according to the ownership they hold and to the demand they haveto supply, make their capacity allocation plan. If the group capacity ownershipis not enough to supply the demand orders, then the group can negotiate a portionof capacity with the other groups whose assigned ownership exceeds their actualdemand (Lo Nigro et al. 2005). In practice, such negotiation and consequent possibleexchange of capacity result in a reassignment of some production plants toa different group. Plants assigned at the beginning of the year to the productionof components belonging to a specific product family could be reconfiguredthroughout the year for producing different type of components, i.e. differentproduct families. Also, within the annual quarter in which the assignment of plantsto group remain fixed, orders of products belonging to a product family (group)must be allocated to the plants assigned to the group temporarily. Plants representreconfigurable production systems able to be reconfigured in the medium period(within the three months) in order to manufacture different types of product ofthe same part family.

Such a brief description demonstrates how the production planning process ina distributed organization can result in complex, multi-period, multi-decisionand multi-issue when a somewhat reconfigurable capability is considered. Figure 1reports an IDEF0 view of such process in DPS made up of reconfigurable plants.

3.1.1 IDEF0 architecture. The production planning process is decomposed infive levels and each PP activity reported in figure 1 is related to a different timehorizon and concerns different PP levels.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

A1

TOP

PP L

evel

A2

HIG

H P

P Le

vel

A3

MED

IUM

PP

Leve

l

A4

LOW

PP

Leve

l

A5

SHO

P FL

OO

RPP

Lev

el

1 ye

ar d

eman

dfo

reca

st

Gro

ups

prio

ritie

s

Glo

bal

owne

rshi

p

Gro

ups

owne

rshi

p

Rem

odul

ated

owne

rshi

p

3 m

onth

s G

roup

s de

man

dfo

reca

sts

Plan

ts p

rodu

ctio

n ca

paci

ty

Plan

t/Gro

up a

lloca

tion

mat

rixPl

ants

cha

ract

eris

tics

Gro

ups

Ord

ers

Ord

er/P

lant

allo

catio

nm

atrix

Res

ourc

e te

chno

logy

char

acte

ristic

s

Job/

Res

ourc

esal

loca

tion

Tool

s to

allo

cate

owne

rshi

p

Tool

s to

rem

odul

ate

owne

rshi

p

Tool

s to

allo

cate

pla

nts

togr

oups

Tool

s to

allo

cate

ord

ers

topl

ants

Tool

s to

allo

cate

jobs

tore

sour

ces

Figure

1.

Alayered

architecture

fortheDPPproblem.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

The TOP PP Level (activity A1) starting from information about one-yeardemand forecast and priorities for each group derived from strategy considerations,assigns the global ownership to groups.

The output of activity A1 represents an input for A2 together with groups’priorities and a demand forecast for the considered time horizon of threemonths. Reducing the forecast horizon, future estimates are more reliable andat the HIGH PP Level a tuning in ownership allocation is allowed.

The MEDIUM PP Level focuses on the same time horizon of the previouslevel but it aims at assigning each plant to a group basing on the output of level 2.

LOW Level and SHOP FLOOR Level concern real-time planning issues: thefirst one allocates, for each group, orders to one of the plants assigned to that group,while the last one is responsible, for each plant, the allocation of jobs to resources.

4. Negotiation support system

Negotiation is one of the most flexible coordination processes in the economicfield. Actually, its application fields enfold all human activities: work activity,entertainment activity and relational activity.

Usually, negotiation is chosen when the transaction, which needs to becoordinated, involves two or more actors with conflicting goals. The conflictingaspects refer to the possible solutions; indeed, each actor prefers a different solution,while the final agreement presents advantages for both parties with respect to theirown ‘best alternative to a negotiated agreement’ (BATNA) (Fischer and Ury 1981).Then, parties need each other because a strict interdependency exists in the courseof actions they want to pursue: this interdependency asks for coordination.

In the manufacturing context negotiation represents a powerful decision-supporttool. Negotiation could be applied to internal conflicts (Cooper and Tabled-Bendiad1998, Sousa and Ramos 1999) where different enterprise functions or divisionssearch a trade-off accord for conflicting objectives. On the other hand, it could alsobe used to regulate external transactions in the two main directions of the supplychain: towards suppliers and towards customers. More generally, enterprises couldneed negotiation to interact with different elements of their industry sector.

In particular the paper will use the classification system proposed in (Lo Nigroet al. 2005) to characterize the negotiation process at each level. The proposedclassification scheme allows formalizing the negotiation process in three macroactivities as shown in the IDEF3 notation reported in figure 2: ‘StructureIdentification’, ‘Meta Negotiation for Protocol’ and ‘Negotiation Dynamics’.Structure identification as detailed in figure 3(a) analyses the problem to be solved(in this case DPP), identifies the conflict reason, the involved actors and their roles,and the inherent sub-negotiations. The meta-negotiation for protocol is depictedin figure 3 (b) and consists of establishing the interaction modes among the involvedactors: the communication channel (how and what to communicate with whom),time relations among sub-negotiations and among issues conflict resolution(static dimension), rules to manage the dispute (how and when a solution can besubmitted, how long the process goes on) in its evolution phase (dynamicdimension). Finally, negotiation dynamics activity, reported in figure 3(c), refersto the assessment by each actor of the strategy and the tactic adopted to formulate


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

offers and/or to evaluate counteroffers; this requires the definition of the actors’utility functions.

The above-mentioned schema can partially fit the Montreal taxonomy fore-negotiation (Strobel and Weinhardt 2003). The Montreal taxonomy aimsat achieving a well-structured approach for the design of electronic negotiation,

NegotiationDynamics

3.1

MetaNegotiationfor Protocol

2.1

StructureIdentification

1.1

L2L1

Figure 2. Negotiation design.

Sub-negotiationsrelationship

Sub-negotiationID

Roles Numberand TypologyID

Actors Numberand TypologyID

Issue Numberand typologyID

ObjectsNumberID

(a) Structure Identification – Static Dimension Variables

StrategyDefinition

TacticDefinition

&

(b) Meta negotiation for Protocol – Protocol Variables

(c) Negotiation Dynamics – Dynamics Variables

CommunicationChannel ID

DynamicDimension

StaticDimension

&

Figure 3. Negotiation variables.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

and at allowing a comparison of a broad variety of e-negotiation designs andsystems. The high granularity of the proposed taxonomy, able to embrace all kindof e-negotiation systems, is outsized for the aim of this paper that focuses on aparticular business environment. On the other hand, the presented research sharesthe perspective assumed by Lomuscio et al. (2003) where the interest is on automatednegotiation among autonomous agents; this gives relevance to the negotiationdynamic variables (strategy and tactic) and to the agent characteristics.

Moreover, the Montreal taxonomy is based on the Media Reference Modelwhere media are platforms where transactions are coordinated through agentinteractions; it identifies four phases of interaction limiting the attention to theintention and agreement phases neglecting the first phase (knowledge) and the last(settlement). Even if it declares to address the agreement phase, it does not coverit effectively but, focusing just on the negotiation medium services, it limits theanalysis using a static perspective.

5. Negotiation support system for DPP

This paper proposes negotiation as a mechanism to be used in all of the activitiespresented in figure 1 (tool to allocate ownership, tool to re-modulate ownership,etc.); negotiation is proposed as a coordination tool to resolve conflicts arising ateach level of the DPP process. The actors involved in the process (corporate, groups,plants, orders and jobs) have to find an agreement, step by step. They are notallowed to quit the process or opt for another alternative (for example, is not possibleto estimate the best alternative to a negotiated agreement); then negotiation supportsactors in satisfying their own utility function by matching the best counterpart.

The negotiation process of each level is complex and requires a detailed analysispartially conducted by the authors in other works (Bruccoleri et al. 2003a, Perroneet al. 2003). In this paper a general framework able to catch common aspectsat different PP levels and to support a MAS implementation is proposed.

The following items refer to the service oriented negotiation (Faratin et al. 1998)based on the Rubinstein protocol (Rubinstein 1982) and on the two-party many-issues models of Raiffa (1982). It is worth specifying some basic assumptions validfor each level:

. the negotiation goes through the following steps: one (or more) of the actorsmakes an offer; interested actors playing the role of potential partner respondby either accepting the offer, rejecting it (through a reactive functionand according to the Rubinstein Protocol) or proposing a counter offer(using a creative function);

. the negotiation is multilateral (it is possible to identify two roles each witha number of actors greater than one);

. the negotiation is time- and resource-constrained.

5.1 Negotiation at different PP levels

In this section negotiation variables, which can be found in figure 3, are definedfor each level. It is interesting to observe that some variables are common for all


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

the levels, in particular dynamic and static variables of protocol. It is assumed that,as in the Rubinstein Protocol, actors can take actions in the negotiation onlyat certain times in the set T¼ {0, 1, 2, . . . , 2n} that are determined in advance andare known to the agents; in particular buyers/sellers submit order/offers to all thesellers/buyers and at the successive step sellers/buyers reply (parallel negotiations areadmitted, that is each actor can negotiate with more than one potential partner).Negotiation ends after n cycles (rounds) of offer/counteroffer or order/counterorder(with or without agreement) or before (if any agreement is achieved). In the caseof multi-issue negotiation, a comprehensive text approach is adopted to bargaineach issue (all issues are indicated contemporary in an offer/counteroffer). Thecomprehensive approach offers the chance to compensate different issues utilitiesand this can be helpful if parties give to the issues different order priorities.

5.2 Agent architecture

Actors involved in the DPP are: corporate, groups, plants, orders, jobs, resources.They have characteristics which are proper of MAS: goal-oriented, collaborative,flexible and capable of making independent decisions on when to act (Etzioni andWeld 1995). Actually, involved agents pursue their own goals, interact with otheragents to bargain a common action plan, get different roles along the process andknow when and how to act in accordance with the concerted protocol.

In particular, referring to the UML notation of figure 4, it can be observed thata generic actor has a rational component, in charge of the negotiation process,and an interface component that takes care of the relationships with other agents.The negotiation agent can behave as buyer or as seller and the interface agent

Actor Agent

Group Corporate Plant Order Job Resource

Negotiation Agent Interface Agent

Buyer Agent Seller Agent

Figure 4. Agent architecture.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

manages the communication channel and is in charge of updating the actors’mental state (in this case, information about the current negotiation process)when information arrives, receives proposals and sends proposals to the correctcounterparts. The behaviour of the actor is not related to its original nature(plant, order, etc.) but simply to the played role (buyer or seller) and thisconsideration can make the NSS design and implementation easier.

5.2.1 TOP PP level. At this level, the Corporate behaves as a seller while groups asbuyers. In fact, Corporate behaves like the owner that splits ownership to groupsbased on the priorities assessed at the strategic level and communicated to all thegroups. Each group aims at maximizing its own negotiated ownership: in order tolimit this trend and to achieve company goal, Corporate concedes ownershipobliging groups to assure to the company a certain profit level according to theirpriorities. The conditions (the negotiation outputs) of the contracts signed in thisphase would be used as control for groups’ performance at the end of the year. Theseconsiderations represent guidelines to keep in mind to set, in related future research,strategies and tactics.

Protocol variables to be set (the others are common for all the levels) concernthe communication channel (cc): there exists a cc for each offer/counteroffer pair(it joins the corporate to each group) and an informational cc to broadcast groups’priorities (from the corporate to all the groups). Negotiation structure identificationis detailed in table 1.

5.2.2 HIGH PP level. Negotiation running at this level, already analysedin (Bruccoleri et al. 2003a), cannot be considered a service-oriented one. On thecontrary, it can be observed that resource (capacity) is scarce and already allocatedto groups while groups can change ownership in the current quarter to contractoptions for receiving it back in the future. To do this it is necessary to introducea lateral payment using an expedient: credits. Credits correspond to a virtualproduction capacity and are equally distributed to the groups at the beginningof each year; they are used to buy capacity, then the group with a great amountof credits has a great power contract to obtain capacity.

At this level, based on the quarter forecast and on the ownership assigned at theprevious level, the group can get buyer or seller behaviour. If the workload relatedto the forecasted demand is higher than the group ownership the group wants to buyproduction capacity; in the opposite condition it is interested in selling the extracapacity and receiving credits.

The process proposed in Bruccoleri et al. (2003a) can be schematized asin figure 5. Depending on the assumed role the Negotiation process

Table 1. Negotiation structure variables for the TOP PP level.

Objects Issues Actors Roles Sub-nego.

Ownership Ownership andrequested profit

Corporate andgroups

Buyer (groups)and Seller (Corporate)

No


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

(unit of behaviour 3.1 in figure 5) is articulated in different activities as figure 6shows for the buyer (a) and the seller (b) roles.

Seller and buyer adopt a time dependent tactic and use generative functionfor the order and counter order formulation.

At this level, there exists only the cc for the offer/counteroffer. Negotiationstructure identification is reported in table 2.

5.2.3 MEDIUM PP level. This level has the same time horizon of the previous onebut involves plants as active participants which have to be assigned to groups basedon the ownership obtained at the end of the HIGH PP level. Assuming that eachplant is a cost and profit centre, it is interested in being assigned to the mostpromising group in terms of future profit; the plant, to avoid imperfect commitment,asks a price to assure its availability. Moreover, it can happen that group catchesplants at different round offering an increasing price but giving it a lower priorityin terms of guaranteed workload. So plants assess a risk attitude, which is thereforeused in their reactive function formulation. Negotiation starts with the pricesubmission offered by each group; each plant evaluates each offer and repliesby accepting it or asking for a new offer. The negotiation structure identification,for the MEDIUM level is reported in table 3.

Buyers (groups) adopt a generative function and a time and resourcedependent tactic (Perrone et al. 2003), while sellers adopt a reactive function anda time-dependent and imitative tactic; strategy is constant for both roles.

At this level there exists only the cc for the offer/counteroffer.

5.2.4 LOW PP level. This level considers a real time horizon; each group collectsorders and each order is assigned to one of the plants obtained at the previous level.The problem is now decomposed at a lower level: here, within each group, orders aimat achieving their objectives and plant aim at being workloaded. The previous levelshave a common decision making structure: the global objective is the company goaland the local objectives are the group goals.

At the low level the global perspective is the group perspective and the localinterests concerns order and plant goals.

Goto / ActorVariablesUpdate

Negotiation

3.1

endprocess

17.1

&J11

Actor RoleIdentification

2.1

ActorVariablesUpdate

1.1

L44L43

L41L42L1

Object /possibleroles: buyer,seller, no role

Object /ownership andcredits update

Figure 5. High Level PP.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

quit

New ordercomputation

&

Sendmessage tocounterpart(sellers)

&Signcontract

Choice bestcounter order

RoundComputation

&First ordersubmission

Process / counterorder submission

(a)

quit

Goto /Orderevaluation

&

counterordersubmission

Counterordercomputation

Orderevaluation

Counter part(buyer)selection

Process / Sendmessage tocounterpart(sellers)

Process /First ordersubmission

(b)

Figure 6. Negotiation process at high level PP for the buyer (a) and the seller (b).

Table 2. Negotiation structure variables for the HIGH PP level.


Ownership Ownership andrequested credits

Groups Buyer (groups)and Seller (groups)

No

Table 3. Negotiation structure variables for the MEDIUM PP level.


Plant assignment Plant assignment Groups, plants Buyer (groups) Noand price Seller (plants)


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

As in the previous levels, the assignment will obviously consider logistic

(the distance between the plant and the final customer), economic (the demandelasticity of each product) and technology aspects (plant/product efficiency

matrix and reconfiguration costs) but, at this level, time is most of all a scarce

resource. Moreover, each plant does not know the price agreed with other plantsor their priorities. Each plant, in the previous step, has agreed with the assigned

group a price and a workload priority but there is no broadcast communication

channel in the protocol, because these represent confidential information.Basically, two situations can occur: orders are a scarce resource or plants are

a scarce resource. In other words, it could be that workload is greater than available

capacity and vice versa; but this information is not known because it is dispersed

among the actors. The solution to this conflict by means of negotiation will dependon the presence of a ‘mediator’. Indeed, if a third part (a mediator role) is present,

in the first case orders behave as buyers and plants as sellers, the roles are reversed

in the other case. If a third party is not present it can be argued that plants behaveas sellers and order as buyers because of the promises deriving from previous level

(price and priority).At this level order can undertake parallel negotiation; in a Response for

Quotation environment customer and supplier can negotiate to fix order character-istics. Usually customers and suppliers (here orders) negotiate about price, volume

and due date (Argoneto et al. 2004); then, negotiation between order and customer

and negotiation between order and plant are interrelated. In fact, order has to waitfor the plant offer before it can propose a counteroffer to the customer to specify the

volume and the due date, while order has to wait for the customer reply about price

to offer a price to plants. Negotiation structure identification, for the LOW level,is reported in table 4.

At this level, there exists only the cc for the offer/counteroffer.Tactics and strategies for this level will be investigated in future research.

5.2.5 SHOP FLOOR PP level. This last level looks at the resource/job allocation.Each order has already been assigned to a plant in the previous level but it could be

processed by one of the available resources in that plant. In order to choose

Table 4. Negotiation structure variables for the LOW PP level.


Assign ordersto plants

Plant/orderassignment and price

Orders and plants Buyer (orders)Seller (plants)

Yes

Table 5. Negotiation structure variables for the SHOP FLOOR level.


Assign jobsto resources

Job/resourceassignment and price

Jobs and resources Buyer ( jobs)Seller (resources)

No


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

what resource for what job, the parameters involved in the decision are resource skillin processing the job and reconfiguration time to process it (Bruccoleri et al. 2003b).

At the shop floor level the global perspective is the plant perspective and thelocal interests arise from the opposite goal pursued by jobs and resources.

Negotiation structure identification, for the SHOP FLOOR level, is reportedin table 5. At this level, there exists only the cc for the offer/counteroffer.

6. Conclusions

The production planning process in a distributed organization can result incomplex, multi-period, multi-decision and multi-issue when a somewhat reconfig-urable capability is considered. The decision-making process adopted in productionplanning activities should always guarantee global satisfaction level by meansof decentralized coordination, and this is can be accomplished by using MAStechnology and automatic negotiation.

The paper presents a modelling methodology for designing and implementinga negotiation support system for DPP that can be adopted in different contextswith analogous structure and that is suitable for an automated solution.

A reference classification has been proposed for analysing and understandingnegotiation dimensions and for assisting the designer during the conceptual design ofeach specific negotiation level. A common framework facilitates the implementationand the usage phase in an object-oriented environment.

In related works authors have already tested successfully the decision-makingstructure proposed for two of the five planning levels. Future works are directedtowards the assessment of a more widespread negotiation taxonomy (in particularnot referred exclusively to the service oriented model), the study of strategies andtactics for the first and for the last level.

Acknowledgments

This research has been supported by the grant of the University of Palermo underthe INTERLINK project titled ‘Innovative Negotiation Models for ProductionPlanning in Reconfigurable Enterprises’.

References

Abid, C., D’Amours, S. and Montreuil, B., Collaborative order management in distributedmanufacturing. Int. J. Prod. Res., 2004, 42(2), 283–302.

Argoneto, P., Renna, P., Perrone, P., Sabato, L., Lo Nigro, G. and Bruccoleri, M., Evaluatingmulti-lateral negotiation policies in manufacturing e-marketplaces, in Proceedings of the37h CIRP International Seminar on Manufacturing Systems, 2004.

Bruccoleri, M., Amico, M. and Perrone, G., Distributed intelligent control of exceptionsin reconfigurable manufacturing systems. Int. J. Prod. Res., 2003a, 41(7), 1393–1412.

Bruccoleri, M., Lo Nigro, G., Federico, F., Noto La Diega, S. and Perrone, G., Negotiationmechanism for capacity allocation in distributed enterprise. Ann. CIRP, 2003b, 52(1),397–402.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

Chen, D., Enterprise-control system integration—an international standard. Int. J. Prod. Res.,2005, 43(20), 4335–4357.

Cooper, S. and Tabled Bendiab, A., CONCENSUS: multi party negotiation supportfor conflict resolution in concurrent engineering design. J. Intell. Manuf., 1998, 9,155–159.

Doumeingts, G., Vallespir, B. and Chen, D. In Handbook on Architecture forInformation Systems, edited by P. Bernus, K. Mertins and G. Schmidt, pp. 313–338,1998 (Springer: Berlin).

Etzioni, O. and Weld, D., Intelligent agents on the Internet: fact, fiction, and forecast.IEEE Expert, 1995, 10(4), 44–49.

Faratin, P., Sierra, C. and Jennings, N.R., Negotiation decision functions for autonomousagents. Robot. Auton. Syst., 1998, 30, 159–182.

Fisher, R. and Ury, W., Getting to yes: Negotiating Agreement without Giving In, 1981(Houghton Mifflin: Boston).

Huang, C.Y. and Nof, Y.S., Autonomy and viability-measures for agent based manufacturingsystems. Int. J. Prod. Res., 2000, 38(17), 4129–4148.

Kanyalkar, A.P. and Adil, G.K., An integrated aggregate and detailed planning in multi-siteproduction environment using linear programming. Int. J. Prod. Res., 2005, 43(20),4431–4454.

Karageorgos, A., Mehandjiev, N., Weichhart and Hammerle, A., Agent-based optimisationof logistics and production planning. Engng Applic. Artif. Intell., 2003, 16, 335–348.

Lo Nigro, G., Bruccoleri, M. and La Commare, U., Negotiation models in manufac-turing e-marketplaces. In Designing and Evaluating Value Added Services inManufacturing e-Marketplace, edited by G. Perrone, M. Bruccoleri and P. Renna,2005 (Springer: Berlin).

Lomuscio, R.A., Wooldridge, M. and Jennings, N.R., Classification scheme for negotiationin electronic commerce. Group Dec. Neg. J., 2003, 12, 31–56.

McEwan, A.M. and Sackett, P.J., An exploration of empowerment in manufacturingenterprises. AI and Society, 2001, 15, 40–57.

Michel, J.J., Manufacturing, modelling and integration. Presentation at a meeting of thecomputer department at CETIM, 1997, France.

Perrone, G., Manufacturing e-marketplaces: innovative tools for the extended enterprise.In Designing and Evaluating Value Added Services in Manufacturing e-Marketplace,edited by G. Perrone, M. Bruccoleri and P. Renna, 2005 (Springer: Berlin).

Perrone, G., Renna, P., Cantamessa, M., Gualano, M., Bruccoleri, M. and Lo Nigro, G.,An agent based architecture for production planning and negotiation in cataloguebased e-marketplace, in Proceedings of the 36th CIRP-International Seminar onManufacturing Systems, 2003.

Rahimifard, S., Semi-heterarchical production planning structures in the support ofteam-based manufacturing. Int. J. Prod. Res., 2004, 42(17), 3369–3382.

Raiffa, H., The Art and Science of Negotiation, 1982 (Cambridge University Press:Cambridge).

Rubinstein, A., Perfect equilibrium in a bargaining model. Econometrica, 1982, 50(1), 97–119.Sousa, P. and Ramos, C.A., Distributed architecture and negotiation protocols for scheduling

in manufacturing systems. Comput. Indust., 1999, 38, 103–113.Strobel, M. and Weinhardt, C., The Montreal taxonomy for electronic negotiations.

Group Dec. Neg., 2003, 12, 143–164.


Dow

nloa

ded

by [

Flor

ida

Atla

ntic

Uni

vers

ity]

at 0

8:42

18

Sept

embe

r 20

13

Negotiation in distributed production planning environments

Documents

Transcript of Negotiation in distributed production planning environments