Experiences with supply chain management in Swedish forest industries

Transcription

1 Experiences with supply chain management in Swedish forest industries Mikael Rönnqvist Division of Optimization, Linköping University, Linköping, Sweden The Forestry Research Institute of Sweden, Uppsala, Sweden Outline Forest supply chain wood flow Supply chain modules Transportation and harvesting Integrated logistics Supply chain planning Process control Summary

2 Forest supply chain wood flow Forest supply chain customers customers customers

3 Divergent flows Bucking Log Sorting Sawing Drying Grading Planing Bucking Log Sorting Sawing Drying Planing Bucking log types 5-8 assortments raw products Grading dry products products Unbalanced annual harvesting '000 cubic meter Stock in forest Deliveries to mills Wood cuttings jan feb mar apr maj jun jul aug sep okt nov dec

4 Mechanization levels in Swedish forestry Percent 100 Final felling Thinning Year Geographical features sawmill pulpmill

5 Industry participants Forest companies and associations, owning forests and pulppapersaw mills. Independent sawmills, without any large forests Independent forest owners not connected with any mills Loggers and transporters Often decentralised management Planning modules (supply chain matrix) Forest management Transportation and routing Production Strategic > 5 years Planting, Valuation, Long term harvesting Road building, Fleet management, Trains?, Terminals? Investment planning Tactical 12 5 years Annual harvest plans Road upgrade, Equipment, Train scheduling Annual production planning Operational days Crew scheduling, Harvest sequencing Catchment areas, Back-hauling, scheduling Lot sizing, Scheduling Online < 1 day Bucking Truck dispatching Process control, Roll cutting, Cross-cutting

6 Solution time 1 day 6 hours 1 hour 30 min 10 min 5 min 1 min 1 s 1 ms * Strategic harvesting * Road investments * Tactical harvesting * Tactical production scheduling * Operative harvesting * Tactical production planning * Tactical transportation * Daily route planning * Daily production planning * Forwarding operations * Dispatching * Process control * Log sawing * Board cutting Planning horizon 1 sec 1-5 min 1-2 hours 1 day 1-2 weeks 1-3 months 1 year 2-5 year 100 years Current development Companies follow each other in terms of development Development of integrated support system for the entire chain but with different modules Modules are often based on manual interference but OR techniques is making advances

7 Transportation and harvesting modules Some facts The forest industry represents a major part (25%) of the demand for transportation of goods in Sweden. The cost of transportation represent one third of the total cost of raw material, round wood, to the forest industry.

8 Pulpmill SCV Pulpmill SCM Pulpmill SCMS Sawmill SCW Mills and forest districts Catchment areas Pulplogs Sawlogs

9 Multicommodity transportation problem x ijk = flow from supply point to demand point j with assortmenti k Mathematical model transportation problem min j J i I x ijk x x i I ijk ijk j J s ik = d c jk ijk x ijk 0, i I,, i I, k K, j J, k K j J, k K (supply) (demand)

10 Simple approach Norra Sala Heby Hallstavik Västerås Västra Östra Backhaulage tour Norra Sala Heby Hallstavik Västerås Västra Östra Backhauling case from southern part in Sweden No sawmills With sawmills ð ê ð ð ð ê ^ ^ ð ê ð ^ ð ê ^ ^ð ^ ð ð ð ð

11 Mathematical model backhaulage problem min j J i I x ijk x x ijk ijk, y i I j J k K l + + l L l L 0, a b c ikl jkl ijk y y l l x ijk s + ik = d jk l L,, d l y l i, k j, k i I, j J, k K, l L (supply at harvest areas) (demand at industries) Backhaulage planning results Savings Cost - 6 % Empty driving - 30 % Time - 9 % Fuel - 9 %

12 Case study Transportsamordning Nord Results backhaulage, 46 % of volumes Empty driving - 24 % Time 9 % Fuel 7 % Sorting decisions No backhaulage With backhaulage SprucePineThinning FinalThinning No sorting SprucePine

13 Åkarweb a web based planning system using optimized backhaulage plans Assortments Volumes Industry destination Harvest areas Potential backhaulage

14 Transport order Potential backhaulage trips connected to a supply node Saved distance Åkarweb current usage Online information system Continuous updates of supplies, demands Total of 180 trucks connected Experiences: Less administration 15% less empty driving 6-7% cost savings

15 Integrated logistics - FlowOpt Supply chain planning modules Forest management and harvesting Transportation and routing Production Strategic planning > 5 years Planting, Evaluation, long term harvesting Road building, road upgrading, fleet management, train, terminals Investment planning Tactical planning 6 months 5 years Annual harvest plans Road upgrade, Equipment and fleet utilization, train schedules Annual production planning Operative planning 1 day 6 months Crew scheduling, Harvest sequencing Catchment areas, backhaulage planning, scheduling Lot sizing, scheduling Online planning < 1 day Bucking Truck dispatching Process control, Roll cutting, Cross-cutting

16 Integrated supply chains Supply of roundwood Truck transport Train terminal Train transport Demand of roundwood " "" Route selection with road database Length Road class Speed limit Surface Road width Owner

17 Recent case 140 supplies 1500 piles 52 industries 220 demands 10 terminals 5 train routes 12 assortments 8 assortment groups 5 scenarios

18

19 Optimization Approx. 100,000 variables (flow) Approx. 3,000 constraints Approx. 30 million backhaulage tours Solution time: (no backhaulage): < 10 seconds (backhaulage): up to 1 hour Results Overall saving using train & trucks: 5-10% Addional saving using backhaulage tours: 2-5% Case studies with FlowOpt Co-operation between companies to coordinate transportation Co-ordinate between companies to swap pulp-logs Integrate harvesting, sorting and paper production Terminal location

20 Supply chain planning The Södra Group Södra Skog Purchasing and trading 12,5 million cubic meters Five regions, 51 districts Södra Cell Processing wood into pulp Five mills Producing 2 million tons Processing 9 million cubic meters Södra Timber Processing logs into sawn timber Six mills Producing 1 million cubic meters Processing 2 million cubic meters

21 Södra Cell - Production units Södra Cell in total Capacity: ~2 million tons Wood: ~9 million cu.m.* *cu.m. = cubic meters, solid under bark Supply chain - Transports To the mills From the mills

22 Customers GHENT GHENT GHENT GHENT GHENT GHENT GHENT GHENT GHENT TERNEUZEN TERNEUZEN TERNEUZEN TERNEUZEN TERNEUZEN TERNEUZEN TERNEUZEN TERNEUZEN TERNEUZEN EMDEN EMDEN EMDEN EMDEN EMDEN EMDEN EMDEN EMDEN EMDEN NORTHFLEET NORTHFLEET NORTHFLEET NORTHFLEET NORTHFLEET NORTHFLEET NORTHFLEET NORTHFLEET NORTHFLEET CHATHAM CHATHAM CHATHAM CHATHAM CHATHAM CHATHAM CHATHAM CHATHAM CHATHAM LOWESTOFT LOWESTOFT LOWESTOFT LOWESTOFT LOWESTOFT LOWESTOFT LOWESTOFT LOWESTOFT LOWESTOFT HULL HULL HULL HULL HULL HULL HULL HULL HULL BOULOGNE BOULOGNE BOULOGNE BOULOGNE BOULOGNE BOULOGNE BOULOGNE BOULOGNE BOULOGNE GENUA GENUA GENUA GENUA GENUA GENUA GENUA GENUA GENUA LA PALLICE LA PALLICE LA PALLICE LA PALLICE LA PALLICE LA PALLICE LA PALLICE LA PALLICE LA PALLICE PASAJES PASAJES PASAJES PASAJES PASAJES PASAJES PASAJES PASAJES PASAJES GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH GRANGEMOUTH CORPACH CORPACH CORPACH CORPACH CORPACH CORPACH CORPACH CORPACH CORPACH INVERNESS INVERNESS INVERNESS INVERNESS INVERNESS INVERNESS INVERNESS INVERNESS INVERNESS AVEIRO AVEIRO AVEIRO AVEIRO AVEIRO AVEIRO AVEIRO AVEIRO AVEIRO ANTWERPEN ANTWERPEN ANTWERPEN ANTWERPEN ANTWERPEN ANTWERPEN ANTWERPEN ANTWERPEN ANTWERPEN GRIMSBY GRIMSBY GRIMSBY GRIMSBY GRIMSBY GRIMSBY GRIMSBY GRIMSBY GRIMSBY ABERDEEN ABERDEEN ABERDEEN ABERDEEN ABERDEEN ABERDEEN ABERDEEN ABERDEEN ABERDEEN BREMEN BREMEN BREMEN BREMEN BREMEN BREMEN BREMEN BREMEN BREMEN GANDIA GANDIA GANDIA GANDIA GANDIA GANDIA GANDIA GANDIA GANDIA STETTIN STETTIN STETTIN STETTIN STETTIN STETTIN STETTIN STETTIN STETTIN WESEL WESEL WESEL WESEL WESEL WESEL WESEL WESEL WESEL DUNDEE DUNDEE DUNDEE DUNDEE DUNDEE DUNDEE DUNDEE DUNDEE DUNDEE HALDEN HALDEN HALDEN HALDEN HALDEN HALDEN HALDEN HALDEN HALDEN HÖNEFOSS HÖNEFOSS HÖNEFOSS HÖNEFOSS HÖNEFOSS HÖNEFOSS HÖNEFOSS HÖNEFOSS HÖNEFOSS KIEL KIEL KIEL KIEL KIEL KIEL KIEL KIEL KIEL SKOGN SKOGN SKOGN SKOGN SKOGN SKOGN SKOGN SKOGN SKOGN SUNDERLAND SUNDERLAND SUNDERLAND SUNDERLAND SUNDERLAND SUNDERLAND SUNDERLAND SUNDERLAND SUNDERLAND BASEL BASEL BASEL BASEL BASEL BASEL BASEL BASEL BASEL Terminaler Folla (4) Svenska (17) Svenska, Tofte (2) Tofte (7) Terminals Member deliveries Wood supply Production Distribution External Wood terminal Shipment port Terminal Wood stock Process Pulp stock Södra Cell - Supply Chain

23 Procurement: Harvesting, Transport Production: Pulp production Distribution: Haulage of pulp Sales: Supplying the customer Harvest calculation: (5yr) - Forest policy implications - DomesticImport - Transport capacity Strategic planning: (5yr) - Investments at mills - Time chartering of vessels - Geographic regions - Terminal structure - Segments - Strategic customers Wood supply balancing: (1yr) - Volume DomesticImport - Harvest volume per Region -Wood-exchange BudgetingPrognosis: (1yr) - Production capacity per product Customer demands per product - Volumesper product per mill Transport volumes (purchase) BudgetingPrognosis: (1yr) - Contracts regular customers - Estimation other customers Wood supply planning: (3m) - Requirement per assortment distributed per RegionSBO Production planning: (3m) - Campaignsper mill Inventory planning (-policies): - Vessel routestime tables - Orders regular customers - Estimation other customers - SMI-forecasts medium term Catchment areas: (1m) - RegSBO -> Mill per assort ment - Back-haulage - Chips-exchange Change-overs: (1m) - Fine tuning of change-overs (daytime) s.t. Customer orders, stocks etc Operative transport plan.: (1m) - Vessel routing - Load-planning at mill s.t. SeaTrainTruck - volumes - Call-offs regular customers - Orders other customers - SMI-forecasts short term Operative tp: (Daily) - Vehicle routing - Combination other transports Operative tp: (Daily) - Vehicle routing - Combination other transports Operative: (Daily) - Orderdelivery process Supply chain - Production planning Wood supply Production Distribution Member deliveries Wood terminal Wood stock Process Pulp stock Shipment port Terminal External

24 Supply chain structure Pulp Logs Forestry Districts Sawmills Import Sawdust Pulp Logs Södra Pulp Mills Domestic Customers Swedish ports International Customers S90Z S90TZ S85Z S85SZ S70Z SBZ S90Z Production Plans Södra Cell, Supply Chain production recipes Soft wood = Long fiber Hard wood = Short fiber Scots pine Norway spruce Saw mill chips Birch Mixed Eucalyptus Process 4-5 to 1 S90Z (std, TCF) S90TZ (strong, TCF) S85Z (std,tcf) S85TZ (strong,tcf) S85SZ (tear, TCF) S85S (tear, ECF) etc... SBZ (Birch) SEUC (Eucalyptus) Production schedule (campaigns) S90Z S90TZ S85Z S85SZ S70Z SBZ S90Z Soft wood Hard wood

25 Daily variation due to batch (campaign) production: Assortment pine Deliveries to mill Stock at mill Consumption in mill 90,0 80,0 70,0 60,0 50,0 40,0 30,0 20,0 10,0 0,0 January 1st to June 30th ,0 8,0 7,0 6,0 5,0 4,0 3,0 2,0 1,0 0,0 l + H x = l i, a, t F F ia, t 1 iat ijat iat j M H H jp, t 1 jpt jpt jpt l + w v = l j, p, t a A ja M j 0.9 f x 1.1 f j, a, t j M a A j M j M f T j ja ijat ja i F D x T i, t ijat D y = D d, p, t jdpt E v = D p, t jpt pt i dpt l + x R δ z l = 0 j, a, t A in A ja, t 1 ijat jra jqrt jq jat i F q Qj r Rj P l + R δ z w y l = 0 j, p, t P out jp, t 1 jrp jqrt jq jpt jdpt q Qj r R d D j No systems from SAP, i2, PC based system Daily time discretisation Minimize total supply chain cost Production variables: z Column generation of 3-month plans Flow and storage variables Inflow, storage, transportation Constraints Flow conservation, capacities, demand Branch and price & constraint branching Initial tests gave 7 days solution time Aggregation of time periods q Qj z jpt jq = 1 j

26 Solution method - column generation Production schedule (campaigns) S90Z S90TZ S85Z S85SZ S70Z SBZ S90Z SBZ S90Z S85Z S85ZTZ S90ZTZ Each campaign has a minimum and maximum duration There is a fixed cost to switch products Solution balances transportation and setup costs 3 Swedish pulp mills Comparison with manual plan (Jan-April 2001) 10 forest districts producing 4 log types 15 products (specific recipes per pulp mill) 90 days planning giving 55 time periods Model A: Master: 9,500 constraints; 31,800 variables+1,500 generated Sub: Production Plan Generator: Model B: Test case 300,000 arcs (full subproblem) 5,500 arcs (lower bounds on campaign length) Master: 14,000 constraints; 35,000 variables

27 Production plans Comparison with manual plans Strategic implications regarding campaign scheduling Challenges: model more parts of the supply chain VARO VAS90Z VARO VAS85RZ VARO VAS85TZ VARO VAS80TZ 3 3 MONSTERAS MONSBZ MONSTERAS MONS90Z MONSTERAS MONS85Z MONSTERAS MONS85S 2 3 MONSTERAS MONS90S MONSTERAS MSTOP MORRUM MORSBZ MORRUM MORS90TD MORRUM MORS70TZ MORRUM MORS90RD Supply chain - Distribution Wood supply Production Distribution Member deliveries Wood terminal Wood stock Process Pulp stock Shipment port Terminal External

28 Distribution - Complexity Three vessels on long term contract Additional vessels on spot market Tofte HALDEN Värö Train and truck transports Several different products Supply Demand HULL GRIMSBY EMDEN KIEL STETTIN Alternative shipment ports LOWESTOFT BREMEN Alternative terminals NORTHFLEET CHATHAM TERNEUZEN WESEL ANTWERPEN GHENT Process control

29 Btv2 O 2 F7 MC 1 Btv1 BT1&2 6xBoiler washing pre-bleaching washing boiling HC3&4 D 0 F1 F3 EOP F2 F4 F5 D 1 F6 D 2 F7 To mill bleaching Kappa number Brightness Bleaching process Kappa number Brightness Brightness Brightness D1a D0 EOP D2 D1b Chlorine dioxide Hydrogen peroxide Oxygen Chlorine dioxide Hydrogen peroxide Time: 0 h Brightness: 39.2 Time: 2.93 h Brightness: Time: 5.61 h Brightness: Time: 8.64 h Brightness: 90.61

30 Support tool - information flow Support tool Operators Optimization Data storage Optimal control values Process data The bleaching process Real control values D1a D0 EOP D2 D1b Brightness after Tower D2 Manual Auto H 2 O 2 H 2 O 2 in control Tower D2 auto Manual Auto A period of 16 months

31 Cost of manual run cost Testing period gave > 9% difference Annual 300,000 savings time Summary

32 Summary and future challenges Forest companies are developing supply chain modules Increasing usage of OR techniques in planning system Savings are typically in the order of 5-10% Important that OR skillededucated persons are a part of the development Data collection can be hard and and faulty data is a potential problem Specialized models and methods required Quick and robust solution methods