Development of a decision support tool for Development of a decision - PowerPoint PPT Presentation

H.- -O. G O. Gü ünther nther H. Technical University of Berlin Technical University of Berlin Germany Germany M. Grunow M. Grunow Technical University of Denmark Technical University of Denmark Copenhagen Copenhagen Development of a decision support tool for Development of a decision support tool for supply network planning: supply network planning: A case study from the chemical industry A case study from the chemical industry ISORA 2008, Lijiang, China October 31, 2008

World network of chemicals trade flows World network of chemicals trade flows Latin America CEEC 71 51 7.3 4.1 12.8 3.5 6.5 10.4 24.4 European USA Union 393 441 14.4 4.6 19.0 12.2 7.4 Asia 183 In billion € 5.5 3.3 17.3 4.8 6.7 Japan 188 5.8 Sources: UNSTAT Comtrade & Cefic-ITC Analysis

Supply network in the chemical industry Supply network in the chemical industry � Application background: intra-organizational supply network � Large number of production sites located in different countries � Production of a chemical specialty � Numerous industrial customers across the globe � Contracted annual quantities, however, highly variable short- term replenishment quantities � Introduction of a central supply chain unit for coordinating � Development of a customized linear optimization model

Company profile Company profile � DEGUSSA AG, Düsseldorf, Germany � World largest producer of special chemicals � Subsidiaries in all continents � 46.000 employees / 300 plants worldwide � Turnover in 2005: 11,800 billion € Carbon black

Supply network in the chemical industry Supply network in the chemical industry � � Supply Network Planning SNP � Mid-term coordination of plant operations � Integration of production and distribution activities � Lack of coordination results in � � Excessive inventories � Poor utilization of capacities � Violation of delivery dates � � Main planning tasks � Allocation of production volumes to plants � Determination of the production mode for the main product and the generation of energy

Advanced Planning Systems (APS) Advanced Planning Systems (APS) � Mathematical methods Based on LP and MILP, heuristics generic model formulation Strategic Network Design Demand Planning Supply Network Planning Production Transportation Order External Planning / Planning / Fulfilment Procurement Detailed Vehicle ATP / CTP Scheduling Scheduling � Decisions Allocation of production quantities between plants Supply from the plants to DCs and customers

Advanced Planning Systems (APS) Advanced Planning Systems (APS) � Generic model formulation Production capacity at plants Storage and handling capacity at DCs ∑ ∑ ⋅ ≤ ∑ α ⋅ ≤ a x PC y SC pi pijt it p pjt j ∈ ∈ ( ) ( ) ∈ p P i j J i ( i ) p P ∑ ∑ α ⋅ + ∑ ∑ α ⋅ ≤ x z HC p pijt p pjkt j ∈ ∈ ∈ ∈ ( ) ( ) ( ) ( ) p P i i I j p P i k K j Storage and handling costs per DC Transportation capacity per link ∑ ∑ ∑ ∑ ∑ ∑ ∑ ⋅ + ⋅ ∑ α ⋅ ≤ h y c z x TC p pjt pjk pjkt p pijt ijt ∈ ∈ ∈ ∈ ∈ ∈ ∈ ∈ p P j J t T p P j J k K ( j ) t T ( i ) p P Aggregate demand per DC Production costs per plants ∑ = z b ∑ ∑ ∑ ∑ ⋅ pjkt pkt c x ∈ pij pijt ( k ) j J ∈ ∈ ∈ ∈ ( ) ( ) p P i i I j J i t T � Assignment of attributes to pre-defined entities

Production process Carbon Black Production process Carbon Black � Production volume: several 100,000 t per annum � 100 product specifications � Continuous production process Trans- Energy formation Gas Feed A Process Main Feed B Silo Silo Silo product Feed C

Implementation of an APS Implementation of an APS � Company negotiates annual volumes with Forecasting the Forecasting the Forecasting the period distribution period distribution key customers. period distribution of annual demand of annual demand of annual demand Allocation of Allocation of customer demands customer demands to production sites to production sites � Customers request deliveries upon short notice.

Demand planning Demand planning 1997 1998 1999 20000 t 18000 16000 14000 12000 10000 Observed demand � 8000 Adaptation of Winters forecasting technique Forecast incl. seasonality 6000 � Considerably increased forecast accuracy 4000 � Forecast represents network-wide commitment 2000 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Month

Decisions in supply network planning Decisions in supply network planning Forecasting the Forecasting the period distribution period distribution of annual demand of annual demand Allocation of Allocation of Allocation of customer demands customer demands customer demands to production sites to production sites to production sites � Production volume in the production sites and at each production train � Transportation volumes between production sites and customers � Generation of energy from side-products

Objectives in Supply Network Planning Objectives in Supply Network Planning � Production costs Minimization of Minimization of Minimization of (site, train) � Transportation costs (Distance, carrier) PC p , GER , Train 1 , t PC p , GER , Train 2 , PC p , GER , Train 3 , t � (negative) energy refund � Inventory costs PC p , RSA , Train 1 , t PC p , RSA , Train 2 , t

APS Implementation: Body of constraints APS Implementation: Body of constraints � Customer Demand � Limited substitutability of products � Delivery only from sites, which have a customer approval for the product � A customer demand may only be covered by deliveries from a limited number of sites ≤ ⋅ T CD B s , c , p , v , t c , p , v s , c , p ≥ ⋅ ∑ T A CD , , , , , , , , , s c p v t c p v t c p v ′ ′ ∈ ∈ ∈ { | ( s , c, p) } s s S A

APS Implementation: Body of constraints APS Implementation: Body of constraints � Production � Storage capacity � Distribution of the delivered ≤ SV MaxSV s , p , t s , p , t volumes at the production trains ∑ ∑ = + − SV SV PV T − s , p , t s , p , t 1 s , tr ( s ), p , t s , c , p , v , t ′ ′ ∈ ∈ ∈ ∈ tr ( s ) TR ( s ) c { c C | (s, c , p) A } ∈ v V

APS Implementation: Body of constraints APS Implementation: Body of constraints � Production � Production capacity at equipment level

APS Implementation: Body of constraints APS Implementation: Body of constraints � Production � Production capacity at equipment level ∑ PV s , tr ( s ), p , t ≤ AE e ( s ), t O s , tr ( s ), p ∈ ∈ p P , tr ( s ) TR ( e ( s ))

APS Implementation: Body of constraints APS Implementation: Body of constraints � Production � Minimum sales from production sites ∑ ≥ T MinT s , c , p , v , t s ′ ′ ∈ ∈ ∈ c { c C | (s, c , p) A } ∈ ∈ ∈ p P , v V , t T

APS Implementation: Body of constraints APS Implementation: Body of constraints � Energy � Transformation capacity of the sites Transformation ∑ ⋅ ≤ ⋅ CP PV MaxCP AT s , tr ( s ), p s , tr ( s ), p , t s s , t ∈ ∈ p P , tr ( s ) TR ( s )

APS Implementation: Body of constraints APS Implementation: Body of constraints � Energy � Distribution of the aggregate energy of the various energy types ∑ ⋅ ⋅ EV CP PV s , tr ( s ), p s , tr ( s ), p s , tr ( s ), p , t ∈ p P , = ∈ tr ( s ) TR ( s ) ∑ G ⋅ s , et , t 3600 EET s , et ∈ et ET Transformation Transformation

APS Implementation: Body of constraints APS Implementation: Body of constraints � Energy � Lower and upper limit on energy sales of the various energy types ⋅ ≤ ≤ ⋅ MinE Y G MaxE Y s , et 2 , t s , et 2 s , et 2 , t s , et 2 , t s , et 2 Transformation Transformation

Optimization software architecture Optimization software architecture Data base OPL Studio SQL MILP Solver Data models (CPLEX) SQL (OPL) Option COM- User and scenario administration Graphical user interface

Application of the optimization model Application of the optimization model � Currently numerous managers are using the tool for operative planning. � Rolled out in Europe, US and Asia. � Estimated financial benefit per year from supply network planning exceeded project costs by far. � Further benefits arise from improved supply network design. � Scenario mode is used extensively, e.g., for capacity decisions, evaluation of approvals and of the profitability of energy transformation.

Thank Thank you you for for your your attention! attention!