summerschool_ronnqvist

37
Pul and Pa er Su l Chain Mana ement  Supply Cha in and Transpo rtation Opt imizatio n within the Forest I ndustr y Mik ael Rön nqv ist The Norwegian School of Economics and Business Administration, Bergen, Norway The Forestry Research Institute of Sweden (Skogforsk), Uppsala, Sweden Outline n ro uc on o pu pan paper supp yc an management (SCM) •Strategic SCM  , • Tactical SCM  Production planning, collaboration in transportation  Routing, process control

Upload: jagan-mohan-reddy

Post on 12-Oct-2015

12 views

Category:

Documents


0 download

DESCRIPTION

SummerSch

TRANSCRIPT

  • Pulp and Paper p pSupply Chain Managementpp y g

    2009 Summer School2009 Summer School Supply Chain and Transportation Optimization within the Forest Industry

    Mikael Rnnqvist

    The Norwegian School of Economics and Business Administration, Bergen, NorwayThe Forestry Research Institute of Sweden (Skogforsk), Uppsala, Sweden

    OutlineOutline

    Introduction to pulp and paper supply chain Introduction to pulp and paper supply chainmanagement (SCM)

    Strategic SCM Terminal location transport mode selectionTerminal location, transport mode selection

    Tactical SCM Production planning, collaboration in transportation

    Operational SCMOperational SCM Routing, process control

  • ReferencesReferences D. Carlsson, S. DAmours, A. Martel and M. Rnnqvist, Supply chain management in the pulp and paper

    industry, to appear in INFOR.

    M. Frisk, M. Gthe-Lundgren, K. Jrnsten and M. Rnnqvist, Cost allocation in collaborative forest transportation, accepted for publication in European Journal of Operational Research.

    P. Flisberg, M. Rnnqvist and S. Nilsson, Billerud optimizes its bleaching process using online optimization, Interfaces,Vol. 39, No. 2, 119-132, 2009.

    P. Flisberg, B. Liden and M. Rnnqvist, A hybrid method based on linear programming and tabu search for routing of logging trucks, Computers & Operations Research, Vol. 36, 1122-1144, 2009.

    S. DAmours, M. Rnnqvist and A. Weintraub, Using Operational Research for supply chain planning in the forest product industry, INFOR, Vol. 46, No. 4, 47-64, 2008.

    H. Gunnarsson and M. Rnnqvist, Solving a multi-period supply chain for a pulp company using heuristics An application to Sdra cell AB, International Journal of Production Economics, Vol. 116, 75-94, 2008.

    G. Andersson, P. Flisberg, B. Liden and M. Rnnqvist, RuttOpt A decision support system for routing of logging trucks, Canadian Journal of Forest Research, Vol. 38, 1784-1796, 2008.

    H. Gunnarsson, M. Rnnqvist and D. Carlsson, A combined terminal location and ship routing problem, Journal of the Operational Research Society, Vol. 57, 928-938, 2006.

    M. Forsberg, M. Frisk, and M. Rnnqvist, FlowOpt a decision support tool for strategic and tactical transportation g, , q , p pp g pplanning in forestry, International Journal of Forest Engineering, Vol. 16, No. 2, pp. 101-114, July 2005.

    D. Carlsson and M. Rnnqvist, Supply chain management in forestry case studies at Sdra Cell AB, European Journal of Operational Research, Vol 163, pp. 589-616, 2005.

    D. Bredstrm, J. T. Lundgren, M. Rnnqvist, D. Carlsson and A. Mason, Supply chain optimization in the pulp mill , g , q , , pp y p p pindustry IP models, column generation and novel constraint branches, European Journal of Operational Research, Vol156, pp 2-22, 2004.

    IntroductionIntroduction

  • A ti i t d d l t f th l b l fl f d fibAnticipated development of the global flow of wood fibres(Source: Axelsson, Sdra Cell AB)

  • Illustration of the global producers by country. (Source Jaako Pyry)

    Procurement Production Distribution Sales Strategic Wood procurement strategy Location decisions Warehouse location Selection of markets

    The Pulp and Paper Supply Chain Planning Matrix.

    Strategic Wood procurement strategy (private vs public land)

    Forest land acquisitions and harvesting contracts

    Silvicultural regime and regeneration strategies

    Harvesting and transportation

    Location decisions Outsourcing decisions Technology and

    capacity investments Product family

    allocation to facilities Order penetration point

    Warehouse location Allocation of

    markets/customers to warehouse

    Investment in logistics resources (e.g. warehouse, handling

    Selection of markets (e.g. location, segment) customer segmentation

    Product-solution portfolio

    Pricing strategy Harvesting and transportation technology and capacity investment

    Transportation strategy and investment (e.g. roads construction, trucks, wagons, terminal, vessels, etc.)

    Order penetration point strategy

    Investment in information technology and planning systems (e.g. advance planning and scheduling

    technologies, vessels) Contracts with

    logistics providers Investment in

    information technology and

    l i t (

    Services strategy Contracts Investment in

    information technology and planning systems (e.g.

    planning systems (e.g. warehouse execution

    On-line tracking system, CRM)

    Tactical Sourcing plan (log class planning)

    Harvesting aggregate planning

    Campaigns duration Product sequencing in

    campaign Lot-sizing

    Warehouse management policies (e.g. dock management) S l i

    Aggregate demand planning per segment

    Customer contracts Demand forecasting,

    Route definition and transshipment yard location and planning

    Allocation of harvesting and transportation equipment to cutting blocs and All ti f d t/bl t

    Outsourcing planning (e.g. external converting)

    Seasonal inventory target

    Parent roll assortment

    Seasonal inventory target at DCs

    Routing (Vessel, train and truck)

    3PL contracts

    safety stocks Available to promise

    aggregate need and planning

    Available to promise allocation rules (i l di ti i Allocation of product/bloc to

    mills Yard layout design Log yard management

    policies

    optimization Temporary mill

    shutdowns

    (including rationing rules and substitution rules)

    Allocation of product and customer to mills and distribution centers

    Operational Detailed log supply planning

    Forest to mills daily carrier selection and routing

    Pulp mills/paper machines/winders/ sheeters daily production plans

    Mills to

    Warehouses/DCs inventory management.

    DCs to customer daily carrier selection

    Available to promise consumption

    Rationing Online ordering Customer inventoryconverters/DCs/

    customers daily carrier selection and routing

    Roll-cutting Process control

    and routing Vehicle loads

    Customer inventory management and replenishment

  • Planning horizon and scope of the reviewed literature

    Strategic planningStrategic planning

  • Supply chainSupply chaindesign at Sdra Cellg

    The Sdra Group

    Sdra Cell Processing wood into pulp

    Sdra Skog Purchasing and trading Processing wood into pulp

    Five mills Producing 2 million tons

    P i 9 illi bi t

    Purchasing and trading 12,5 million cubic meters Five regions, 51 districts

    Processing 9 million cubic meters

    Sdra Timber Processing logs into sawn timber

    Six mills Producing 1 million cubic meters Processing 2 million cubic metersg

  • Case: Sdra Cell - millsSdra Cell in total

    Capacity: ~2 million tons

    Case: Sdra Cell millsCapacity: 2 million tonsWood: ~9 million cu.m.*

    *cu.m. = cubic meters, solid under bark

    CustomersSKOGNSKOGNSKOGNSKOGNSKOGNSKOGNSKOGNSKOGNSKOGN

    Terminaler

    Folla (4)Svenska (17)Svenska, Tofte (2)Tofte (7)

    Terminals

    HALDENHALDENHALDENHALDENHALDENHALDENHALDENHALDENHALDEN

    HNEFOSSHNEFOSSHNEFOSSHNEFOSSHNEFOSSHNEFOSSHNEFOSSHNEFOSSHNEFOSS

    GRANGEMOUTHGRANGEMOUTHGRANGEMOUTHGRANGEMOUTHGRANGEMOUTHGRANGEMOUTHGRANGEMOUTHGRANGEMOUTHGRANGEMOUTH

    CORPACHCORPACHCORPACHCORPACHCORPACHCORPACHCORPACHCORPACHCORPACH

    INVERNESSINVERNESSINVERNESSINVERNESSINVERNESSINVERNESSINVERNESSINVERNESSINVERNESS

    ABERDEENABERDEENABERDEENABERDEENABERDEENABERDEENABERDEENABERDEENABERDEEN

    DUNDEEDUNDEEDUNDEEDUNDEEDUNDEEDUNDEEDUNDEEDUNDEEDUNDEE

    SUNDERLANDSUNDERLANDSUNDERLANDSUNDERLANDSUNDERLANDSUNDERLANDSUNDERLANDSUNDERLANDSUNDERLAND

    TERNEUZENTERNEUZENTERNEUZENTERNEUZENTERNEUZENTERNEUZENTERNEUZENTERNEUZENTERNEUZEN

    EMDENEMDENEMDENEMDENEMDENEMDENEMDENEMDENEMDEN

    NORTHFLEETNORTHFLEETNORTHFLEETNORTHFLEETNORTHFLEETNORTHFLEETNORTHFLEETNORTHFLEETNORTHFLEET

    CHATHAMCHATHAMCHATHAMCHATHAMCHATHAMCHATHAMCHATHAMCHATHAMCHATHAM

    LOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFT

    HULLHULLHULLHULLHULLHULLHULLHULLHULL

    ANTWERPENANTWERPENANTWERPENANTWERPENANTWERPENANTWERPENANTWERPENANTWERPENANTWERPEN

    GRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBY

    BREMENBREMENBREMENBREMENBREMENBREMENBREMENBREMENBREMENSTETTINSTETTINSTETTINSTETTINSTETTINSTETTINSTETTINSTETTINSTETTIN

    WESELWESELWESELWESELWESELWESELWESELWESELWESEL

    KIELKIELKIELKIELKIELKIELKIELKIELKIEL

    GHENTGHENTGHENTGHENTGHENTGHENTGHENTGHENTGHENT

    BOULOGNEBOULOGNEBOULOGNEBOULOGNEBOULOGNEBOULOGNEBOULOGNEBOULOGNEBOULOGNE

    BASELBASELBASELBASELBASELBASELBASELBASELBASEL

    GENUAGENUAGENUAGENUAGENUAGENUAGENUAGENUAGENUA

    LA PALLICELA PALLICELA PALLICELA PALLICELA PALLICELA PALLICELA PALLICELA PALLICELA PALLICE

    PASAJESPASAJESPASAJESPASAJESPASAJESPASAJESPASAJESPASAJESPASAJES

    AVEIROAVEIROAVEIROAVEIROAVEIROAVEIROAVEIROAVEIROAVEIRO

    GANDIAGANDIAGANDIAGANDIAGANDIAGANDIAGANDIAGANDIAGANDIA

  • Sdra Cell - Supply ChainWood supply Production Distribution

    Sdra Cell Supply Chain

    Woodterminal

    port

    al

    Shi

    pmen

    t p

    Term

    ina

    External

    Procurement:Harvest ing Transport

    Production:Pulp product ion

    Distribution:Haulage of pulp

    Sales:Supplying the customerHarvest ing, Transport Pulp product ion Haulage of pulp

    Harvest calculation: (5yr)- Forest policy implicat ions

    Strategic planning: (5yr)- Investments at mills - Time chartering of vessels - Geographic regions

    Supplying the customer

    p y p- Domest ic/Import- Transport capacity

    Wood supply balancing: (1yr)

    g g p g- Terminal st ructure - Segments

    - Strategic customers

    Budgeting/Prognosis: (1yr) Budgeting/Prognosis: (1yr)- Volume Domest ic/Import- Harvest volume per Region- Wood-exchange

    Wood supply planning: (3m)

    - Product ion capacity per product Customer demands perproduct

    - Volumes per product per millTransport volumes (purchase)

    Production planning: (3m) Inventory planning ( policies):

    - Contracts regular customers- Est imat ion other customers

    O d l tWood supply planning: (3m)- Requirement per assortment dist ributed per Region

    Production planning: (3m)- Campaigns per mill

    Inventory planning (-policies):- Vessel routes/t ime tables

    - Orders regular customers- Est imat ion other customers- SMI* -forecasts medium

    term

    Catchment areas: (1m)- Region -> Mill per

    assortment- Back-haulage- Chips-exchange

    Change-overs: (1m)- Fine tuning of change-overs

    (day/t ime)s.t . Customer orders, stocks

    etc

    Operative transport plan.: (1m)- Vessel rout ing- Load-planning at mills.t . Sea/Train/Truck - volumes

    - Call-of fs regular customers- Orders other customers- SMI-forecasts short term

    p g

    Operative transportation- Vehicle rout ing (Daily)- Combinat ion of other

    t ransports

    Operative transportation- Vehicle rout ing (Daily)- Combinat ion of other

    t ransports

    Operative: (Daily)- Order/delivery process

    t ransportst ransports

  • Inventories / Lead times

    30 60

    >15% of volume => 320 000 tonnes

    30 60

    >15% of volume => 320 000 tonnes 150 million

    Storage cost (20%) 30 million

    PulpSMI:pSupplier Managed Inventory

    Sdra Cell responsibilityCustomer responsibility

    Security stock(5-10 days)

  • TerminalsSdra Cell Folla

    Sdra Cell Tofte

    Sdra Cell MnstersSd C ll M

    Sdra Cell Vr

    G th Sdra Cell Mrrum

    SwinoujscieBrake

    Grangemouth

    TerneuzenChatham

    Basel

    Genova

    La Pallice

    Pasajes

    Delivery alternativesSdra Cell Folla

    Sdra Cell Tofte

    Sdra Cell MnstersSd C ll M

    Sdra Cell Vr

    Sdra Cell Mrrum

    SwinoujscieBrake

    TerneuzenTerneuzen

    Basel

  • Distribution - ComplexityHALDEN

    TofteDistribution Complexity

    Vr Three vessels on long term contract

    Additional vessels on spot marketp

    Train and truck transports

    EMDENHULLHULLHULLHULLHULLHULLHULLHULLHULLGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBYGRIMSBY STETTIN

    KIEL Several different products

    Supply Demand

    NORTHFLEET

    LOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTLOWESTOFTBREMEN

    WESEL

    Supp y e d

    Alternative shipment ports

    GHENT

    TERNEUZENNORTHFLEET

    CHATHAM ANTWERPEN

    WESELshipment ports

    Alternative terminals

    Supply chain design atSupply chain design at Sveaskogg

  • Case study - Sveaskog Major Swedish forest

    company (Sveaskog) withcompany (Sveaskog) with 16% of overall productiveforest area

    Using one train system, Trtget

    Study to use a new system, Bergslagspendeln, with a

    b f i lnumber of potential terminals

    Classical transportation problemp p

    ixijk demand topoint supply from flow=kj assortment with point

  • Norra

    Simple approach

    NorraNorra

    Simple approach

    SalaHeby Hallstavik

    SalaHeby Hallstavik

    SalaHeby Hallstavik

    stra

    Sala

    stra

    Sala

    stra

    Sala

    VstraVsters VstraVsters VstraVsters

    Norra

    Backhaulage tour

    NorraNorra

    Backhaulage tour

    SalaHeby Hallstavik

    SalaHeby Hallstavik

    SalaHeby Hallstavik

    straVst aV te

    straVst aV te

    straVst aV te VstraVsters VstraVsters VstraVsters

    Catchment areas - comparisonCatc e t a eas co pa so

  • Case studyCase study

    1,500 supply points, 220 industrial demands, 5 train routes 10 potential terminals5 train routes, 10 potential terminals, 12 products, 8 product groups, five scenarios. 3,000 constraints, 30 million variables Solution time 1 minute several hours Solution time 1 minute several hours Truck transports reduced by 35% and overall

    energy 20%

  • Tactical planning integrated productionTactical planning integrated productionand transportation planning at Sdra Cell

  • Supply chainWood supply Production Distribution

    Supp y c

    Woodterminal Ship-

    ment

    Term-inals

    port

    External

    Supply chain structureSupply chain structure

    Forestry Districts M

    ills Domestic

    Customers

    SawmillsSwedish Harbours

    Districts

    a P

    ulp

    M Customers

    Sd

    ra

    Import International Customers

    Production Plans

    S90Z S90TZ S85Z S85SZ S70Z SBZ S90Z

  • Sdra Cell, Supply Chain - conditionsDaily variation due to campaign production

    , pp y

    Stockat mill

    Usagein mill

    Deliveriesto mill

    campaign production

    8 0 0

    9 0 ,0

    8 0

    9 ,0

    (0-90 000) (0-8 000)(0-9 000)

    6 0 ,0

    7 0 ,0

    8 0 ,0

    6 ,0

    7 ,0

    8 ,0

    3 0 0

    4 0 ,0

    5 0 ,0

    3 0

    4 ,0

    5 ,0

    1 0 ,0

    2 0 ,0

    3 0 ,0

    1 ,0

    2 ,0

    3 ,0

    0 ,0 0 ,0January 1st to June 30th 2000

    Production plansProduction plans Comparison with manual plans (campaign changesComparison with manual plans (campaign changes

    35 million SEK, total 120 110 million SEK) Strategic implications regarding campaign scheduling Strategic implications regarding campaign scheduling

    VARO VAS90Z 2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3VARO VAS85RZ 3 3 3VARO VAS85TZ 2 2 2 3VARO VAS80TZ 3 3VARO VAS80TZ 3 3

    M ONSTERAS M ONSBZ 2 2 2 2 2 2 2 3 3 3M ONSTERAS M ONS90Z 3 3 3 3M ONSTERAS M ONS85Z 2 2 2 2 2 2M ONSTERAS M ONS85S 2 3M ONSTERAS M ONS85S 2 3M ONSTERAS M ONS90S 2 3 3M ONSTERAS M STOP

    M ORRUM M ORSBZ 3 3 3 3 3M ORRUM M ORS90TD 2 2 2 2 2 2 3M ORRUM M ORS90TD 2 2 2 2 2 2 3M ORRUM M ORS70TZ 2 2 2 2 2 2 2 2 2 3 3 3 3M ORRUM M ORS90RD

    1 31 61

  • Collaboration in transportationCollaboration in transportation

    Case studyCase study

    St d f d t i 2004 ( thl l ) Study from data in 2004 (monthly volume) Analysis done by Skogforsk

    (Th F R h I i f S d )(The Forestry Research Institute of Sweden) 8 participating companies in Sweden 898 supply points 101 demand points0 de d po s 39 assortments, 12 assortment groups Savings: Savings:

    Approximately 8% by co-operation Total of 14% as compared to actual transportationTotal of 14% as compared to actual transportation Environment CO2: 5,330 tons to 4,350 tons.

  • Company VolumeProportion

    (%)Company 1 77 361 8 76Company 1 77,361 8,76Company 2 301,660 34,16Company 3 94,769 10,73p y , ,Company 4 44,509 5,04Company 5 232,103 26,29Company 6 89,318 10,12Company 7 36,786 4,17Company 8 6,446 0,73

    Total 882,952

    opt1: direct flows (individual)opt2: backhaul flows (individual)

    opt3: direct flows (collaboration)opt4: backhaul flows (collaboration)

    Company Real opt 1 opt 2 opt 3 opt 4Company 1 3,894 3,778 3,640p y 3,89 3, 8 3,6 0Company 2 15,757 14,859 14,684Company 3 4 828 4 742 4 703Company 3 4,828 4,742 4,703Company 4 2,103 2,067 2,043Company 5 10 704 10 340 10 153Company 5 10,704 10,340 10,153Company 6 5,084 4,959 4,826Company 7 1 934 1 884 1 877Company 7 1,934 1,884 1,877Company 8 0,333 0,333 0,332

    AllAll 39,253 38,315Total 44,637 42,963 42,257 39,253 38,315

    Saving (%) 0,00 3,75 5,33 12,06 14,16

  • Cost allocation based on volumeCost allocation based on volume

    Company Individual Volume %Company 1 3,778 3,439 9,0Company 2 14,859 13,411 9,7Company 3 4 742 4 213 11 2Company 3 4,742 4,213 11,2Company 4 2,067 1,979 4,3Company 5 10,340 10,318 0,2Company 6 4,959 3,971 19,9p y ,Company 7 1,884 1,635 13,2Company 8 0 333 0 287 14 0Company 8 0,333 0,287 14,0

    Sum 42,963 39,253

    Supply and demandSupply and demand

  • Game theory approach Basicsy ppplayers)(or tsparticipan ofsubset a :coalition NS

    for ation transportfor thecost :tsparticipan all :coalition grand

    Sc(S)NS =

    p S(S)

    { }

    :allocationEfficient c(N)yNj

    j =

    { }:Core

    )(:rationalIndividual

    NSc(S)y

    jcy

    j

    j

    y)(efficienc

    , :Core

    c(N)y

    NSc(S)ySj

    j

    =

    y)(efficienc c(N)yNj

    j =

  • Volume vs Shapley valueVolume vs Shapley valueCompany Volume % Shapley %

    Company 1 3,439 9,0 3,586 5,1Company 2 13,411 9,7 13,528 9,0, ,Company 3 4,213 11,2 4,102 13,5Company 4 1,979 4,3 1,889 8,6p y , 4,3 , 8,6Company 5 10,318 0,2 9,747 5,7Company 6 3 971 19 9 4 503 9 2Company 6 3,971 19,9 4,503 9,2Company 7 1,635 13,2 1,587 15,8Company 8 0 287 14 0 0 310 6 9Company 8 0,287 14,0 0,310 6,9

    Sum 39,253 39,253Stable No YesStable No Yes

    Equal Profit Method (EPM)Equal Profit Method (EPM)

    Company Volume Shapley Shadow Nucleo. EPMCompany 1 9 0 5 1 4 1 3 4 6 7Company 1 9,0 5,1 4,1 3,4 6,7Company 2 9,7 9,0 12,7 11,1 8,8Company 3 11,2 13,5 14,2 14,0 8,8Company 4 4,3 8,6 13,3 6,4 8,8Company 5 0,2 5,7 -1,8 4,8 8,8Company 6 19,9 9,2 11,7 8,3 8,8Company 7 13,2 15,8 15,6 11,5 8,8Company 8 14,0 6,9 9,1 4,6 8,8Company 8 14,0 6,9 9,1 4,6 8,8

  • Operational planning routingOperational planning routing

    Standard vehicle routing problemg p

  • forest vehicle routing problemg p

    542,000 km roads in Sweden-----------------------------------

    102,000 government56,000 local communities

    384 000 private384,000 private

  • Route selection with road database

    LengthLengthRoad classSpeed limitS fSurface

    Road widthRoad widthOwner

    Gantt schedule for a dayGantt schedule for a day

  • Industrial demand

    Accumulated demandmaxmin

    max

    i i dmin

    Time periodDay 1 Day 2 Day 3 Day 4 Day 5

    Gantt schedule for three daysGantt schedule for three days

  • Two phase methodTwo phase method Idea:

    Use an efficient tabu search algorithm to form routes Phase 1:Phase 1:

    Construct transport nodes mainly by solving a destination problem (LP) and form loads taken fromdestination problem (LP) and form loads taken from supply point(s) to demand point

    Phase 2: Phase 2: Use an extended and modified tabu search to

    construct routes where e g time windows and excessconstruct routes where e.g. time windows and excess supply is taken care of

  • Industries

    Harvest areas

  • Home bases and change-over pointsg p

  • Real time planningReal time planningprocess control at Billerud

  • Plug flow D(EOP)DP sequencegKappa number

    BrightnessKappa number

    BrightnessBrightnessResidual

    BrightnessResidual

    D1a

    EOP

    D0

    D1b

    D2

    EOP

    Hydrogen peroxideOxygen

    Chl i di idChlorine dioxide

    (Chlorine dioxide)Hydrogen peroxide

    Time: 0 h Time: 2.93 h

    Time: 5.61 h

    Brightness: 87.26 Time: 8.64 h

    Chlorine dioxide

    Brightness: 39.18 Brightness: 74.75 Brightness: 87.84Brightness: 90.61

  • State vector, control, and output, , p

    Stage D0 Stage EOP Stage D1 Stage D2

    D1a

    x

    2z 1z

    x D1b

    D2 3x 3y 4y 2y 1y

    3z 4z EOP D0

    4x 1x 2x D1b

    3x :Variables

    iii

    ii

    Stageafteroutputcharges) (chemical Stagein control

    Stage before vector State :

    ===

    yxzVariables

    ii Stageafter output =y

    P lParameter values

    ),,(sDefinition

    = iiiii wxzFy TowerStage i

    iz iy

    spolynomial e thfor Parameters =iw

    ix

    solve we problem onoptimizati squareleast a

    ),,( s.t. )),,((min ][LSOP

    2i

    iiiii

    iiiiibwxzAywxzF

    Unknown- data Historical - ,,

    i

    iiiw

    xzy

  • Master ProblemLinking the output with the

    approximate functionsMaster Problem

    Connect brightness between

    Cost of chemicals

    min [MP]4

    1=i

    iT

    i xcConnect brightness between

    stages

    4,3,2,1)( 1,2,3,4),,( s.t.

    1 ====

    + ii

    iii

    iiiii

    yzGwxzFy

    Relaxed and

    1,2,3,4

    1,2,3,4 maxmin

    maxmin

    ==

    ii

    iii

    iii

    xxxyyypenalized in the

    objectiveiii

    Stagein chemicalsfor costs - where

    i ic

    Upper and lower limits of theoutput and control

    numberkappaely with thequivalent and stages, ebetween th

    brightness econnect th toused functions - )(][LSOP ofsolution theasfix -

    gi

    ii

    i

    i

    zGw

    number kappa

    Support tool - information flowSupport tool information flow

    Support tool

    O ti l t l

    Operators

    OptimizationData storage

    Plug flow

    Optimal control values

    Optimization

    The bleaching process Real control l

    Process

    D1a

    EOP

    valuesdata

    D0

    D1b

    D2

  • Brightness after Tower D1bBrightness after Tower D1b

    90

    88

    84

    86

    Estimated Real

    80

    82

    7816.jan 16.jan 17.jan 17.jan 18.jan 18.jan 19.jan 19.jan 20.jan

  • Cost of manual run

    Kr/(ton o kappa)

    13

    14

    15

    Real cost

    10

    11

    12

    8

    9

    Opt cost

    Period of three weeks, > 9 % difference

  • Concluding remarksConcluding remarks

    There are many planning problems that can be There are many planning problems that can be coordinated and/or optimized in the pulp and papersupply chain and which provide large savingssupply chain and which provide large savings.

    The models are a combination of nonlinear/ integerd l land large scale.

    There are many open and challenging problems in the pulp and paper supply chain, e.g. robustness, uncertainty, collaboration