m. grunow technical university of denmark, copenhagen national tsing hua university, hsinchu, taiwan...

M. GrunowM. Grunow Technical University of Denmark, Technical University of Denmark,

CopenhagenCopenhagen

National Tsing Hua University, Hsinchu, Taiwan December 13, 2007

Supply Network Planning andSupply Network Planning and

Advanced Planning SystemsAdvanced Planning Systems

H.-O. GüntherH.-O. GüntherTechnical University of Berlin, Technical University of Berlin,

GermanyGermany

http://www.dtu.dk/

OutlineOutline

Introduction

Advanced planning systems

Case 1: Global network design

Case 3: Production planning and scheduling

Case 2: Supply network planning

Case 4: Value chain management

Outlook: Teaching SNP & APS

Supply Chain ManagementSupply Chain Management

www.wikipedia.org:

“The primary objective of supply chain management is to fulfill customer demands through the most efficient use of resources.“

www.wikipedia.org:

“A supply chain, logistics network, or supply network is a coordinated system ...

… of entities, activities, information and resources involved in moving a product or service from supplier to customer.“

Supply Chain ManagementSupply Chain Management

Introduction






OutlineOutline


Generations of PPC softwareGenerations of PPC software

1960s

1970s

1980s

1990s

2000s

• Predecessors of PPC systems Focus on inventory control Basic order processing

• Material requirements planning (MRP) Bill of material files Calculation of net requirements

• Manufacturing resources planning (MRP II) Enhanced planning functions Integration of financial accounting and management functions

• Integrated systems CIM: Integration of manufacturing ERP systems covering the whole enterprise

• Advanced Planning and Scheduling systems Integration into supply chain management concept Use of true optimization techniques

Advanced Planning Systems (APS)Advanced Planning Systems (APS)

APS modulesAPS modules

Strategic Network Design

Supply Network Planning

DemandPlanning

External Procurement

Production Planning / Detailed

Scheduling

Transportation Planning /

Vehicle Scheduling

Order Fulfilment ATP / CTP



Scheduling


long-term

mid-term

short-term

Industrial case studiesIndustrial case studies

Case 1: Global network design for the production of

electrical components

Case 2: Multi-site production-distribution

planning in the chemical industry

Case 3: Characteristics

dependent planning in the

consumer goods industry


OutlineOutline

Introduction







Case 1: Global network designCase 1: Global network design


DemandPlanning



Scheduling


Vehicle Scheduling


long-term

mid-term

short-term Order FulfilmentATP / CTP

Case 1: Global network design for the production of

electrical components

Grunow, M., Günther, H.O., Burdenik, H., Alting, L.:Evolving production network structures.CIRP Annals (to appear)

Strategic network designStrategic network design


Mathematical methods Heuristics, MILP, Clustering techniques

Determination of transportation links

Assignment of locations to each other e.g. customers to DCs

Decisions

Number of plants and DCs

Assignment of products to plants

Locations and capacities

APS based on generic model

formulations provide limited support for strategic network

design.

Case 1: InsightsCase 1: Insights

Redesign of global network also affects the organizational structure and

business processes.Potential cost

savings achieved through quantitative

modeling underestimated by decision makers

(engineers).

ProductsProducts

3 Levels Vacuum interrupters

Kits (for local content and taxation reasons)

Switches (usually containing 3 vacuum interrupters)

Supply networkSupply network

Current centralized network Production sites at

Head production centers (produce vacuum interrupters): GER (with R&D) and CHINA 1

USA, INDIA, CHINA 2

MEX, BRA (from kits only)


Future network

Reduced production complexity at head production centers Increased flexibility through networked supply and distribution

Reduced production and logistics costs


Product relocation depends on Product maturity

new products at sites with R&D department

Personnel

required for product relocation (to qualify the new production facility for the production of the new products).

Complexity and flexibility Diversity of products at facility increases scheduling complexity Products at multiple facilities increase network complexity Production close to customer markets Exchange rate risks and taxation

Production capacities

test runs and ramp-up of new products


Product relocation depends on Product maturity

new products at sites with R&D department

Personnel

required for product relocation (to qualify the new production facility for the production of the new products).

Complexity and flexibility Diversity of products at facility increases scheduling complexity Products at multiple facilities increases network complexity Production close to customer markets Exchange rate risks and taxation

Production capacities

test runs and ramp-up of new products

Error !Not covered !

Standard commercial APS

MILP optimization modelMILP optimization model

1

,,,,,,

,,,,,,,,

,,,,,,,,

,,,,,,,,,,,,

)1/(1

t

t

f gtfgtfg

ftftf

k hpctkpchpck

kpctkpchpckkpchpck

i hpctpchpci

pctpchpcipchpci

p f ctcfptcfptcfp

ptfp

ffp

ir

zFCwIC

DCCS

DCCS

CDSCVC

Min

Minimization of the relevant production and logistics costs variable manufacturing costs (including material and personnel) costs for shipments to customers (+ customs duty) costs for shipments of intermediates and kits from head production

centers to other facilities in the network investment costs required for the expansion of the production

capacities for final products costs associated with the complexity in a facility net present value calculation


Demand

Production

(final products)

tcpf

tcfp D ,,,,,

c

tcfstfs ,,,,,

kpc

tkpchpckc

tchpckthpck ,,,,,,,,

KSp

thpcppipc

tpchpcic

tchpcithpci A ,,,,,,,,,,,

(kits)

(intermediates)

Shipment of intermediates

tkpcshpc

tkpchpck ,,,,,

s

tpcssihpc

tpchpci A ,,,,,,

(kits)

(intermediates)


Capacities

(final product capacity incl. expansions, here: testing equipment)

(intermediates capacity)

t

ttfftf

stfs wECS

1',,,,

ft

tf FEw ,

tf

tf NEw ,

thpcithpci CI ,,,,

thpck

thpck CK ,,,

(no. of expansions per facility)

(no. of expansions in network per perioddue to personnel requirements, budget constraints)

(kit capacity)


Relocation

(continuous-binary production variables)

(number of relocations per facility and period, due to capacity requirements for test runs and ramp-ups)

(relocation-production)

(number of total relocations per period, due to personnel requirements)

(ban on relocation of products in R&D phase)

tfsfftftfs xEVFECS ,,,,,

1

0',,,,,,

t

ttfstfstfs xxy

Nys f

tfs ,,

fs

tfs By ,,

01

,, f

TR

ttfs

s

y

M


Complexity

(production of group)

(flexibility, limitation of exchange risks, demonstration of goodwill, marketing)

(network complexity)

(facility complexity)

gSs

tfstfg xMz ,,,,

tg f

tfg CNz ,,

tfg

tfg CFz ,,,

p p Cc

tcptptrFf

tfptp

rr

DLVLSLV ,,,,,,,

Local production

1

,,

,,,,,,,

,,,

,,,,,

,,,,

,,,,,

,,,,,,,,,

,,

)1/(1)_

)___(

(

tf

ff g

tfg

ftftfpchpcpfp

pchpcptpchpcp

cfpfpcfp

cfppchpcp

tpchpcppchpcp

cfpcfptcfptfp

fpfp

t

itcoefficiencomplexity

initgroupgrproducefacility

ECIIDSMCTR

DSMCTRISSC

DSSCpVMCMin


Advantages of optimization models Consideration of interdependencies between entities in the SC.

Quick and easy evaluation of scenarios

Provide understandable quantitative results (e.g. cost savings) as basis for final management decisions.

Manager: „Sometimes we make strategic

SND decisions very intuitively.“

Exemplary resultsExemplary results

Scenarios Current business plan Scenario 1: Actual configuration with optimized production/distribution Scenario 2: Reassignment of products with fixed capacity Scenario 3: Reassignment of products with capacity adjustments

Results Savings in production costs

overcompensate additional logistics costs.

Shift of production volumes and products from Germany to China

Increased capacity in Mexico and China

Direct manufacturing costs

Plan Szenario 1 Szenario 2 Szenario 3 Plan Szenario 1 Szenario 2 Szenario 3 Plan Szenario 1 Szenario 2 Szenario 3

2008 2009 2010

Plan Szenario 1 Szenario 2 Szenario 3 Plan Szenario 1 Szenario 2 Szenario 3 Plan Szenario 1 Szenario 2 Szenario 3

2008 2009 2010

OutlineOutline

Introduction







Case 2: Case 2: Supply network planningSupply network planning


DemandPlanning



Scheduling


Vehicle Scheduling


long-term

mid-term


Case 2: Multi-site production-distribution

planning in the chemical industry


Supply network planningSupply network planning

Consideration of production, transportation, and handling capacities as hard constraints

Decisions

Allocation of production quantities between plants

Smoothing out seasonal cycles in demand

Supply from the plants to DCs and from the DCs to customers

Mathematical methods LP and MILP, heuristics

Consideration of demand, due dates, and safety stocks as soft constraints

Considerable cost savings achieved

through application of optimization

model.


Generic model formulations embedded in SNP modules of APS

do not consider application-specific

features.

High degree of acceptance by management.

DEGUSSA AG, Düsseldorf, Germany

Company profileCompany profile

World largest producer of special chemicals

Subsidiaries in all continents

46.000 employees / 300 plants worldwide

Turnover in 2005: 11,800 billion €

Production process Carbon BlackProduction process Carbon Black

Production volume: several 100.000 t per annum

100 product specifications

Continuous production process

Feed A

Feed B

Feed C

Process Mainproduct Silo

Gas

Trans-formation

Energy

Silo Silo

Implementation of an APSImplementation of an APS

Company negotiates annual volumes with key customers.

Forecasting the Forecasting the period distribution period distribution of annual demandof annual demand


Allocation of Allocation of customer demands customer demands to production sitesto production sites Customers request

deliveries upon short notice.

Demand planningDemand planning

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

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

t

Observed demand

Forecast incl. seasonality

1997 1998 1999

Adaptation of Winters forecasting technique

Considerably increased forecast accuracy

Forecast represents network-wide commitment


Allocation of Allocation of customer demands customer demands to production sitesto production sites

Allocation of Allocation of customer demands customer demands to production sitesto production sites

Decisions in supply network planningDecisions in supply network planning

Transportation volumes between production sites and customers

Production volume in the production sites and at each production train

Generation of energy from side-products

Objectives in Supply Network PlanningObjectives in Supply Network Planning

Transportation costsTransportation costs(Distance, carrier)(Distance, carrier)

Inventory costsInventory costs

t,1Train,GER,pPC

t,2Train,GER,pPC

t,3Train,GER,pPC

t,1Train,RSA,pPC

t,2Train,RSA,pPC

Production costsProduction costs(site, train)(site, train)

(negative) energy refund(negative) energy refund

Minimization ofMinimization ofMinimization ofMinimization of

Customer DemandCustomer 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

vpctvpcASss

tvpcs CDAT ,,,,,}p) c,,s(|{

,,,,

p,c,sv,p,ct,v,p,c,s BCDT

Selected constraints: DistributionSelected constraints: Distribution

Selected constraints: EnergySelected constraints: Energy

Energy Transformation capacity of the sites

Transformation

tss

sTRstrPp

tpstrspstrs ATMaxCPPVCP ,

)()(,

,),(,),(,

Graphical user interface

User and scenario administration

Solver(CPLEX)

MILPmodels(OPL)

CO

M-

Op

tio

n

SQL

DataSQL

Data base OPL Studio

Optimization software architectureOptimization software architecture

Application of the optimization modelApplication of the optimization model

Currently numerous managers are using the tool for operative planning.

Rolled out in Europe, US and Asia.

Estimated financial benefitper year from supply network planning exceeded project costs by far.

Scenario mode is used extensively, e.g., for capacity decisions, evaluation of approvals and of the profitability of energy transformation.

Further benefits arise from improved supply network design.

Performance improvement in optimizationPerformance improvement in optimization

Average algorithmic improvement 1988 2001

Estimated machine improvement 1988 2003: 800 x Net improvement for LP problems 1988 today: ~2,000,000 x

Bixby, B., Solving real-world linear programs: a decade and more of progress. Operations Research, 50 (2002), pp. 3-15

OutlineOutline

Introduction







Case 3: Case 3: Production planning / Production planning / detailed schedulingdetailed scheduling


DemandPlanning



Scheduling


Vehicle Scheduling


long-term

mid-term


Case 3: Characteristics

dependent planning in the consumer goods industry

Günther H.O., Grunow, M., Neuhaus, U.:Realizing block planning concepts in make-and-pack production using MILP modelling and SAP APO. IJPR (2006), 3711-3726

Lütke Entrup, M., Günther, H.O., van Beek, P., Grunow, M., Seiler, T., MILP approaches to shelf life integrated planning and scheduling in yogurt production. IJPR (2005), 5071-5100

Production planning / detailed schedulingProduction planning / detailed scheduling

Lot-sizing, sequencing, and procurement proposals

Decisions Generation of production orders

Allocation of resources according to a finite scheduling policy

Mathematical methods Genetic algorithms

Consideration of the availability of resources as hard constraints


Scheduling

Application specific algorithms

Consideration of due dates, time windows etc. as soft constraints

Constraint programming

Rules and heuristics

Development of a mathematical

modelling approach based on human expertise.


Efficient support by characteristic based planning tool of APS

module.

High performance of MILP model enables

reactive planning mode.

Block planning conceptBlock planning concept

Given a natural sequence of set-ups (e.g. light to dark)

Decisions on cyclic production patterns (i.e. lot sizes)

Minimizing inventory and set-up costs

Block 1 Block 4Block 3Block 2

Production cycle

1

4 weeks

Production cycle

2 3 4

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

Setup families

time

Production order Major setup

Production of hair dyes

Numerical experimentsNumerical experiments

time4 weeks 8 weeks 12 weeks

...

Period 1 Period 3Period 2 ...

SolutionSolution

MILP model: ILOG OPL Studio and CPLEX 7.0MILP model: ILOG OPL Studio and CPLEX 7.0

Characteristic dependent planning of SAP APO 3.1Characteristic dependent planning of SAP APO 3.1

Setup families

1

4 weeks

Production cycle

2 3 4

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

Production order Major setup

Production of hair dyes


Master data

No. of master data records

With CDP Without CDP

Locations 1 1

PPMs 30 56

Resources 14 14

Products 37 140

Total 84 213

No

. of

Products 26

Periods 6

Production orders 156

Variables 780

Constraints 786

CPU time 0.25

APO implementation MILP block planning model

Recuced data volume Small CPU times

• Combination of characteristics dependent planning and MILP modelling seems appealing


OutlineOutline

Introduction








Case 4: Value chain managementCase 4: Value chain management


DemandPlanning



Scheduling


Vehicle Scheduling


long-term

mid-term


SCM is usually understood as a “coordinated system … involved in

moving a product or service from supplier to customer.“

Dynamic spot pricing for sales and procurement makes a business unit

highly profitable.

Coordinated decision making along the

entire supply chain appears to be very

effective.


Change of paradigm from “supply chain

management” to “value chain management”?

OutlineOutline

Industrial case study

Sales model

Distribution model

Numerical experiments

Production model

Procurement model

Supply chainSupply chain

Europe

China

Planning problemPlanning problem

• Standard products with defined quality

• Price is the key buying criterion

• Considerable sales price and demand volatility

• Increasing raw material costs due to higher crude oil prices

Chemical commodities

Planning problem

• Operational planning for 6-12 months

• Coordination of sales, production, procurement and distribution activities for the entire company-internal value chain

Value chain networkValue chain network

Procurement locations

Production location

Production location

Sales locations

suppliers’value chain

Sales

company-internal value chain network

Procure-ment

Prod-uction

Distribu-tion

Sales

customers’ value chain

Procure-ment

financial flowmaterial flow information exchange

Sales locations


Distribution locations

Transportation lanes

Spot sales opportunity

Decision dependenciesDecision dependenciesScenario: Spot sales opportunitiesScenario: Spot sales opportunities

Pipeline inventory?

Stocks?

Production capacity?

Production mode?

Costs?

Raw material availability?

Costs?Transportation capacity?

Sales locationsSales

locations

Productionlocations

Procurementlocations


Oversupply

Decision dependenciesDecision dependenciesScenario: Oversupply at production locationScenario: Oversupply at production location

Sales locationsSales

locations

Productionlocations

Procurementlocations


Push spot sales

Production mode?

Inventory allocation?

Delay contract procurement?

OutlineOutline


Sales model

Distribution model


Production model

Procurement model

Sales available total supply

Demand

€/t €/t €/t

€/t

€/t €/t

push

€/t

€/t-spot

sales price

spot sales quantity

€/t

cut

contractdemand price

contractdemand quantity

spot demand quantity

spot demand price

1 2 3 Periods

contractsales quantity

contract sales price €/t €/t€/t

- - -

1 2 3 Periods

Contract demand needs to be fulfilled, spot demand has bid character and is fulfilled depending on supply and profitability.

• Average spot sales price increases

• Reduced production volume

• Savings in raw material consumption• Reduced spot procurement

• Average spot sales price decreases

• Reduced inventories

• Efficient use of contract procurement quantities

• Change of production mode (recipe)

Contract and spot demandContract and spot demand

Price-quantity function

100

100

300

93.3

400

90

600

83.3

0

100

0 300 600Cum. quantity [t]

Price [€/t]

Price-quantity-share function

17%

120%

50%

112%

67%

108%

100%

100%

0%

100%

0% 50% 100%

Cum. quantity share [%]

[%]

Priceshare

Algorithm steps Example

Customer [Unit] A B C D

Quantity [t] 100 200 100 200

1. List individual customer price and quantity

forecasts for product, sales location and

period

→

Price [€/t] 100 90 80 70

2. Sort forecasts by price and determine rank → Rank 1. 2. 3. 4.

3. Determine cumulated quantity for rank → Quantity [t] 100 300 400 600

4. Determine average price for rank → Ø Price [€/t] 100 93.3 90 83.3

5. Determine cumulated quantity share of rank → Quantity [%] 17 50 67 100

6. Determine average price share of rank → Price [%] 120 112 108 100

7. Conduct linear regression for shares → Regression y = -0.2407 x + 1.2408

8. Determine elasticity from regression → Elasticity 2407.0Ssplt

Algorithm steps Example

Customer [Unit] A B C D

Quantity [t] 100 200 100 200

1. List individual customer price and quantity

forecasts for product, sales location and

period

→

Price [€/t] 100 90 80 70

2. Sort forecasts by price and determine rank → Rank 1. 2. 3. 4.

3. Determine cumulated quantity for rank → Quantity [t] 100 300 400 600

4. Determine average price for rank → Ø Price [€/t] 100 93.3 90 83.3

5. Determine cumulated quantity share of rank → Quantity [%] 17 50 67 100

6. Determine average price share of rank → Price [%] 120 112 108 100

7. Conduct linear regression for shares → Regression y = -0.2407 x + 1.2408

8. Determine elasticity from regression → Elasticity 2407.0Ssplt

Linear price-quantity functionLinear price-quantity functionfor each product, sales location and periodfor each product, sales location and period

Quadratic turnover functionQuadratic turnover function

sp

ox

( )p x


Average spot price

Cumulated spot sales quantity

x

minx maxx

Turnover function

x

ox

( )p x x


Spot sales turnover

minxmaxx



ox

( )p x

Average spot sales price

sp

Spot sales quantity

Spot sales

price

x

minx maxx

Piecewise linear approximationPiecewise linear approximation

Turnover approximation

actual turnover curve


x

( )p x x

Spot sales turnover

Spot sales quantity

minx maxx

ox

Sales locationsDecisions for p,l and t

spot sales quantitySspltx

total sales quantitySpltx

Ssplty spot sales turnover

Input for p,l and t

contract demand quantityScpltq

Distribution locations,SsSspltpltX X min. and max. spot sales


Constraints of the sales model (1)Constraints of the sales model (1)

• Total sales quantity equals the sum of spot and contract sales quantity.

• Boundaries for spot sales quantities

Scplt

Ssplt

Splt qxx { , } Sp l PL Tt;

Ssplt

Ssplt

Ssplt XxX { , } Sp l PL Tt;

Product-sales location combination

, Sp l PL

DSpltx

• Linear approximation of spot sales quantities and spot sales turnover

Constraints of the sales model (2)Constraints of the sales model (2)

1

1

NSs Ssplt jplt

j

y y

{ , } Sp l PL Tt;

Ss Ss Ssjplt jplt jplty x { , } Sp l PL 1.. 1j N Tt; ;

1

1

NSs Ssplt jplt

j

x x

{ , } Sp l PL Tt;

1,Ss Ss Ssjplt iplt i pltx q q { , } Sp l PL Tt; ;1ijNi ..j 1 N 1 1i; ; ;

OutlineOutline


Sales model

Distribution model


Production model

Procurement model

Decisions for p and t

transportation quantity received via e

Dpetx

Input for p,l and e

,D Dpe peLT UT min. and max. transportation

quantity for p on e

Constraints of the distribution model (1)Constraints of the distribution model (1)

• Transportation quantities are limited for each lane.

• Transportation quantities are limited for each lane and product.

Spltx sales quantity

Product-distribution location combination

, Dp l PL

Transportation lane: e EProduct-transportation

lane combination

, Dp e PE

transportation quantity sent via e with lead time zero

Dpetx

transportation quantity sent via e with lead time one

Dpetx

inventory level at location lDplty

min. and max. transportation quantity on e,D D

e eLT UT

,D Dplt pltLI UI min. and max. inventory

level for p at l in t

Sales locations

, D

D D De pet e

p e PE

LT x UT

,e E t T

D D Dpe pet peLT x UT , ,Dp e PE t T

Constraints of the distribution model (2)Constraints of the distribution model (2)

Decisions for p and t

transportation quantity received via e

Dpetx

Input for p,l and e

,D Dpe peLT UT min. and max. transportation

quantity for p on e

Spltx sales quantity

Product-distribution location combination

, Dp l PL

Transportation lane: e EProduct-transportation

lane combination

, Dp e PE

transportation quantity sent via e with lead time zero

Dpetx

transportation quantity sent via e with lead time one

Dpetx

inventory level at location lDplty

min. and max. transportation quantity on e,D D

e eLT UT

,D Dplt pltLI UI min. and max. inventory level

for p at l in t

Sales locations

• Limited inventory for each product and location

• Balancing incoming and outgoing material flows

D D Dpl plt plLI y UI , ,Dp l PL t T

, 1 , 1

0 1D D Dpl pl pl

D D D D Dplt pl t pet pet pe t

e R e S e S

y y x x x

, ,Dp l PL t T

• Sales and distribution balance

Dpl

D Spet plt

e S

x x

, ,Sp l PL t T

OutlineOutline


Sales model

Distribution model


Production model

Procurement model

Production siteProduction site

P4

process group 1 process group 2

Line 1

Period 1 Period 2

Line 2

carry-over process group

change-over processsequence-dependent

processes withinprocess group

P1

P5

P2 P3

P4 P6P7

P1

Idle

Operational conditions

• Equipment continuously operated except for planned maintenance

• Minimum utilization rate

• Raw material consumption depends on processing mode (throughput speed)

• Processes may have multiple outputs with flexible output rates

• At most one change-over between process groups per period

Production locations Decisions for r,l and t

quantity of process sPrlstx

operating time of group gPrlgtY

Prlsty operating time of process s

Input for r,l

operating hours in tPrltH

Distribution locations,P Prls rlsLQ UQ min. and max. output rate

Constraints of the production model (1)Constraints of the production model (1)related to resourcesrelated to resources

• Allocation of operating time among processes in a group

• Output variation of process groups

Pg

P Prlgt rlst

s S

Y y

, , , ,P Plr l RL r g RG t T

• Continuous operation of equipment

' '

','

Pl

P P P Prlgt rlg g rlg gt rlt

g Gr g RGg g

Y C H

, ,Pr l RL t T

P P P P Prls rlst rlst rls rlstLQ y x UQ y , , , ,P P

lr l RL r g RG t T

, Pr l RL

Resources , Plr g RG

Production locations Decisions for r,l,g and t

quantity of process groupPrlgtX

change-over from g to g’'Prlg gt

Prlgt setup state at the end of t

Input for r,l,g

Change-over time from g to g’'Prlg gC

Constraints of the production model (2)Constraints of the production model (2)related to change-oversrelated to change-overs

• At most one change-over between groups per period

• Start of campaign depends on setup state

• Setup state at the end of a period

, Pr l RL

Change-over , Plr g RG

'

, , ''

1P Pl l

rlg gt

r g RG r g RGg g

, ,Pr l RL t T

,

1Pl

Prlgt

r g RG

, ,Pr l RL t T

'

, ''

( )Pl

P P Prlgt rlgt rlg gt

r g RGg g

X B

, , , ,P Plr l RL r g RG t T

Raw material consumptionRaw material consumption

Raw material recipe function

Production utilization

Inputquantity

Raw material

Outputproduct

Outputquantity

Resource

RecipefactorMaximum capacity

Minimum utilization Linear recipe

Variable prod. costs

0

0.5

1.0

1.5

0% 20% 40% 60% 80% 100%Utilization

Production locations

Decisions for p,l and t

input quantityPinpltx

output quantityPoutpltx

Input for p,r,l,s

Constraints of the production model (3)Constraints of the production model (3)related to productsrelated to products

• Sales and distribution balance

• Recipe for consumption of input material

• Aggregation of output from different processes

, Pr l RL

Products , ,Pin Poutp l PL PL

raw material recipe function,P Pprls prlsa b

Distribution locationsPprls output yield in process s


Dpetx

, Pl

Pout P Pplt prls rlst

r s RS

X x

, ,outp l PL t T

, ,P P Pl g l

Pin P P P Pplt prls rlst prls rlgt

r g RG s S r g RG

X a x b Y

, ,Pinp l PL t T

Ppl

D Poutpet plt

e S

x x

, ,Pp l PL t T

OutlineOutline


Sales model

Distribution model


Production model

Procurement model

Spot and contract procurementSpot and contract procurement

spot offer quantity

Contract offer quantity

800 €

600 €

Contract offer cost rate

spot offer cost rate spot procure-

ment quantity

Totalprocurement

quantity

800 €

600 €

Offer Procurement

750 €

Contract

Spot

Average procure-ment costs


Decisions for p,l and t

spot procure-ment quantity

Qspotpltx

Input for p,l and t

Constraints of the procurement modelConstraints of the procurement model

• Balance of procurement quantities and production input

, Qp l PL

contract quantityQconpltq

Distribution locations,Q Qplt pltLS US min. and max. spot quantity

Pinpltx

• Boundaries for spot procurement quantities

Q Qspot Qspotplt plt pltLS x US { , } ,Qp l PL t T

, Pin

Qspot QconPinplt plt plt

p l PL

x x q

, ,Qp l PL t T

Production locations

Objective functionObjective function


Production location

Production location

Sales locationsDistribution locations

maxt T

S

Ss Ssplt plt

pl PL

x p

, Q

Qspot Qspotplt plt

p l PL

c x

, P Pl g

P Prst rs

r g RG s S

x c

, D

D DIplt pl

p l PL

y c

, D

D DTpet pe

p e PE

x c

OutlineOutline


Sales model

Distribution model


Production model

Procurement model

Basis elements Scale

PeriodsProducts

129

Locations Product-location combinationsTransportation lanesTransportation lane – product

comb.

24713467

Production resourcesProduction processes

1415

ConstraintsVariablesCPU time (MIP gap 0.5%)

14,35317,2113.0 sec

Model implementation using ILOG OPL Studio 3.71 and CPLEX 9.1

Test case and implementation Test case and implementation

Test case: excerpt from a real industrial value chain

Profit

Distribution costs

Production costs

Procurement costs

Turnover

High volatility of profit, sales and procurement costs, but stable production and distribution costs

Optimization run with company dataOptimization run with company data

No full capacity utilization and spot demand coverage

OutlineOutline

Introduction







Teaching SNP & APSTeaching SNP & APS

Development of teaching software for SCMusing Advanced Planning Systems

Joint project by:B. Fleischmann, Augsburg, Germany,M. Grunow, Copenhagen, Denmark,H.-O. Günther, Berlin, Germany,H. Meyr, Darmstadt, Germany,H. Stadtler, Hamburg, Germany

Supported by:SAP AG, Walldorf, Germany,DATANGO AG, Berlin, Germany

Output:DVD with teaching software (Phase 1)Accompanying text book (Phase 2)

19 products made to stock 3 production sites with two lines each 3 warehouses 60 customers

Ludwigshafen

Augsburg

Magdeburg

Case study: Soft drink manufacturerCase study: Soft drink manufacturer

short-term

medium-term

long-term

PurchasingWorkbench

Procurement Production Distribution Sales

Strategic Network Planning

Modules of mySAP SCMModules of mySAP SCM

Supply Network Planning (SNP)

DeploymentDeploymentProductionPlanning

&Detailed

Scheduling(PP/DS)

ProductionPlanning

&Detailed

Scheduling(PP/DS)

DemandPlanning

(DP)

DemandPlanning

(DP)

TransportationPlanning & VSTransportationPlanning & VS

ATP & CTPATP & CTP

for

Thank you

attention! your

m. grunow technical university of denmark, copenhagen national tsing hua university, hsinchu, taiwan...

Documents

supply network planning

scheduling case

global network design

enterprise advanced

logistics network

supply chain management

value chain management

chemical industry case