Global Supply Chain Sustainability Optimization in ...

2 downloads 0 Views 426KB Size Report
for supply chain network design. Presentation at the. 27th European Conference on Operational Research (EURO). Matthias Kannegiesser, Hans-Otto Günther, ...
Time-to-Sustainability as optimization strategy for supply chain network design

Matthias Kannegiesser, Hans-Otto Günther, Niels Autenrieb

Presentation at the 27th European Conference on Operational Research (EURO)

Glasgow, 14 July, 2015 Technical University Berlin Department of Production Management

Seoul National University Dept. of Industrial Engineering

Agenda

 Problem introduction: Sustainability in network design  Time-to-Sustainability (TTS) optimization strategy  Numerical results

/15/2015 5:40 PM

 Summary

2

Sustainability in network design aims for a long-term triple bottom line in the supply chain network. Sustainability in network design

Illustrative example

Sustainability triple bottom line meets ...

...supply chain network design

Economic Raw materials

Part Supplier

Manufacturer

x

x

Retailer

End consumer

Import

x Sustainability

Domestic

Demand

/15/2015 5:40 PM

Social

Environment

• Triple-bottom-line perspective on corporate performance • Measured with economic, social and environmental key performance indicators • Setting long-term targets e.g. to reduce CO2e emissions, maintaining cost competitiveness and improve social conditions CO2e: Carbon dioxide equivalents

Manufacturing

Logistics

Retail

Transport lane

• Long-term decision on network structures • Focus on locations and lanes decisions driven by future demand vs. anticipated costs • Discrete open/close vs. continuous capacity increase/ decrease decisions 3

Various research areas can be related to sustainable supply chain and network design. Literature review

Min. cost or max profit (after tax) of supply chain network with projected demand Sustainable supply chain - Seuring (2013) - Brandenburg et al. (2014) - Taticchi et al. (2014) Optimizing multiple sustainability objective trade-off incl. weighting factors

/15/2015 5:40 PM

Goal Programming (GP) - Charnes et al. (1955) - Charnes and Cooper (1961) - Schniederjans (1995)

Research Gaps • Economic objectives only • Single objective function • Max. profit/min. cost paradigm, vs. sustainability balance • Assuming sustainability to be conflicting trade-offs • Multi-objective requires subjective weighting factors • Single-period focus, aspect of time & transition missing • Economic objectives only • Single objective function • Single-period focus, aspect of time & transition missing

Illustration multi-objective trade-off optimization

economic value

Conventional SC network design - Meixell and Gargeya (2005) - Goetschalckx and Fleischmann (2008) - Melo et al. (2010) - Corominas et al. (2015)

Selection

Potential optimum situation, with trade-off line chosen by society

Actual situation

total environmental value

Flexible method, multiple economic goals can be modeled, delta to target optimized

SC: Supply Chain

4

How can decisions for supply chain network design towards sustainability can be effectively supported?

/15/2015 5:40 PM

Research questions

1

How to support the aspect of time in sustainability transformation in long-term network design?

2

How to deal with multiple incompatible sustainability objectives and indicators in quantitative optimization strategies?

3

How to deliver insights to decision makers to answer, if and how sustainability objectives can be achieved?

5

Agenda

 Problem introduction: Sustainability in network design  Time-to-Sustainability (TTS) optimization strategy  Numerical results

/15/2015 5:40 PM

 Summary

6

Sustainability optimization models for network design can be structured along an optimization framework. Sustainable supply chain optimization framework and strategies Alternative optimization strategies • Min. Costs/max. profit with sustainability constraints • Multi-objective trade-off optimization • Minimize time-to-sustainability

Framework

Sustainability key performance indicators

Optimization strategies in objective function

External scenario drivers Demographics Globalization

/15/2015 5:40 PM

Resources & Environment

1. Value Chain

External scenario drivers Consumer patterns

Integrated value chain model

Process Model

Transport Model

Network definition

Productin-use model

Approach

Technology Regulations & Activism

LP-Model

2. KPIs 3. Targets & Time 4. SubSystem 5. Opti.Strategy 6. Results

Demarcation of the relevant value chain and definition of the aims of the investigation Identification of the factors of interest and definition of the KPIs Setting of targets and timeframe Modeling of the relevant sub-system incl. data Selection of the optimization strategy Detailed analysis of the obtained results

KPI: Key Performance Indicator Model details in M. Kannegiesser, H.-O. Günther: Sustainable development of global supply chains – part 1: sustainability optimization framework. Flexible Services and Manufacturing, No. 1-2 (2014), 24-47; External scenario drivers based on Laudicina (2005)

7

TTS minimizes the time to reach sustainability targets in all KPIs steady-state in the supply chain network. Minimize Time-to-Sustainability (TTS) principle Illustrative example Sustainability KPIs Sustainability target met Environmental targets e.g. CO2 emissions sustainable corridor

Social targets e.g. job redundancies

Actual emissions

 Actual job redundancies

Actual costs Economic targets e.g. total costs



sustainable corridor Sustainability target not met

0%

20% 0%

Sustainability steady state 110%



sustainable corridor

0%

Time

Minimize time to sustainability /15/2015 5:40 PM

40%

Minimum TTS (steady state) KPI: Key Performance Indicator Note: corridor boundary values expressed as ratio to baseline value in percent

8

The standard TTS checks for each KPI if a period is already sustainable steady state (=1) or not (=0). Time-to-sustainability variants (1) – Standard TTS Parameters Basek Baseline (reference) value of KPI k Targetk Target value of KPI k as percentage of the

Equations Sustainability targets for all KPIs and periods

M  kt  KPI kt  Targetk  Basek

k  K , t  T

baseline value

UBk Upper bound of KPI k as percentage of the

baseline value

A period is sustainable if all succeeding periods are also sustainable

 kt   k ,t 1 Decision variables KPI kt

/15/2015 5:40 PM

 kt

Actual value of KPI k in period t =1, if the target for KPI k has not been achieved by period t (0, else)

zk

Time to sustainability for KPI k

Z

Overall time to sustainability

Time-to-sustainability

zk   kt

k  K

tT

Minimize the overall time-to-sustainability

min Z

TTS: Time-to-Sustainability

k  K , t T : t  1 with  k1  1

s.t. Z  zk

k  K

9

Characteristically for TTS is the check of periods to reach sustainability targets steady state.

period sustainability [0=Yes, 1=No]

Standard TTS result – Illustrative example

1

0

period 1

2

3

/15/2015 5:40 PM

CO2 emissions

TTS: Time-to-Sustainability

4

5 costs

6

7

8

9

10

labor dismissals

10

The total TTS reduces integers by focusing on total sustainability of a period across all KPIs. Time-to-sustainability variants (2) – Total TTS Parameters Basek Baseline (reference) value of KPI k Targetk Target value of KPI k as percentage of the baseline value

UBk Upper bound of KPI k as percentage of the

baseline value

Equations Sustainability targets for all KPIs and periods

 kt  KPI kt  Targetk  Basek

k  K , t  T

M 

k  K , t  T

tot t

  kt

A period is sustainable if all succeeding periods are also sustainable

 ttot   ttot1 Decision variables KPI kt

/15/2015 5:40 PM



Actual value of KPI k in period t

 t  T : t  1 with 1tot  1

Time-to-sustainability

Z   ttot tT

tot t

 kt Z

=1, if the entire set of sustainability targets for KPI k has not been achieved by period t (0, else)

Minimize the overall time-to-sustainability

Total overflow with respect to KPI k in period t

min Z Overall time to sustainability

TTS: Time-to-Sustainability

11

The overflow TTS minimizes penalized period overflows compared to target values without integers. Time-to-sustainability variants (3) – Overflow TTS Parameters

Basek

Baseline (reference) value of KPI k

Targetk Target value of KPI k as percentage of the baseline value

UBk



Upper bound of KPI k as percentage of the baseline value Penalty parameter

Decision variables KPI kt

 kt /15/2015 5:40 PM

Z

 kt

Actual value of KPI k in period t Normalized total overflow with respect to KPI k in period t Overall time to sustainability

Equations Normalized sustainability achievement for all KPIs and periods

 kt 

KPI kt  Targetk  Basek UBk  Targetk   Basek

k  K , t  T

Linear vs. exponential penalty linear

 kt    t

exponential

 kt   t

k  K , t  T

Time-to-sustainability

Z    kt   kt kK tT

Minimize the overall time-to-sustainability

min Z

Overflow penalty with respect to KPI k in period t

TTS: Time-to-Sustainability

12

The overflow TTS outperforms the other variants specifically on solving times. Time-to-sustainability variants – Comparison Solving times by TTS variants Comparison of TTS variants Criteria

Method flexibility

CO2 reduction targets

Standard TTS

Total TTS

Overflow TTS

-50%

High

High

High

-40% -30%

Validity of results

Low

Low

High -25%

Objectivity of results

High

High

High

Solving time

Long

Medium

Short

Dependency on data complexity

High

High

Low

-20% Overflow TTS exp. Overflow TTS lin. Total TTS Standard TTS

-10% -5%

/15/2015 5:40 PM

0

1.000

2.000

3.000

4.000

5.000

solving time [sec.]

TTS: Time-to-Sustainability

13

Agenda

 Problem introduction: Sustainability in network design  Time-to-Sustainability (TTS) optimization strategy  Numerical results

/15/2015 5:40 PM

 Summary

14

We evaluated the TTS variants with a numerical data basis from the automotive industry. Overview numerical data basis Illustrative excerpt



End-to-end to modeling of the automotive industry supply chain network



Focus on the long-term sustainable development of the industry supply chain towards 2030 under the paradigm of powertrain electrification



Key geographic focus on network in Europe/Germany and China

 



/15/2015 5:40 PM



Three types of cars: small, compact and premium

Industry supply chain network illustration



Five powertrain options andReverse three sizes of cars supply chain forward flow reverse flow Recycler

Five powertrain options incl. old/new ICE, EV, PHEV, HEV

Data from industry sources e.g. for projected demand, productivity, technology, energy data Focus on economic, environmental and social sustainability performance: total costs, CO2e emissions and avoided job dismissals

Waste disposal Scrap Yard

Forward supply chain Parts suppliers

Raw material supplier

Use Customers Export

Component suppliers

OEMs

Distributors

Customers

ICE: Internal Combustion Engine, EV: Electric Vehicle, PHEV: Plug-in Hybrid Electric Vehicle, HEV: Hybrid Electric Vehicle For more details on the case refer to: M. Kannegiesser, H.-O. Günther, O. Gylfason, Sustainable development of global supply chains - part 2: investigation of the European automotive industry. Flexible Services and Manufacturing, No. 1-2 (2014), 48-68. and H.-O. Günther, M. Kannegiesser, N. Autenrieb, The role of electric vehicles for supply chain sustainability in the automotive industry. Journal of 15 Cleaner Production, 90 (2015).

Overflow TTS comes closest to sustainability targets outperforming other variants. Results by TTS variants average emissions [tonne CO2eq/vehicle] 24

Total make + use Emissions per vehicle by TTS variants - S1: 25% emission reduction target -

23 22

Upper limit Sustainability target Standard TTS Total TTS Overflow TTS Linear Overflow TTS Exp

Minimize overflow

21 20 19 18

Minimize TTS

17 16

15 14 2

3

4

5

6

7

8

9

10

11

12

13

14

15

/15/2015 5:40 PM

Period

16

Looking at different KPIs in parallel, emission reduction is the critical objectives. Overflow TTS results – CO2 vs. Costs vs. Jobs comparison

CO2 emissions per new vehicle (t)

25

20.000

Costs per new vehicle (EUR)

2.500

Job dismissals (1.000)

18.000 2.000

20 16.000 15

12.000

10

Emissions per Vehicle S1 Emissions per Vehicle S2 Max. Level 25%-Target (S1: Scen. 1) 35%-Target (S2: Scen. 2)

5

0 2

3

4

5

6

7

8

9 10 11 12 13 14 15

Period

/15/2015 5:40 PM

1.500

14.000

1.000 Max. Level

10.000 8.000

Costs per New Vehicle EUR (S1) Costs per Vehicle EUR (S2)

500

Job dismissals (S1)

Job dismissals (S2)

Sust. Cost Level (5% Baseline) 6.000

Max. Level

0 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Period

2 3 4 5 6 7 8 9 10 11 12 13 14 15

Period

17

TTS allows to analyze the impact on different reduction targets on reduction paths. Analyzing CO2 emissions reduction scenarios – make vs. use CO2 per vehicle delta to target [t] 5,0

S1: 25% CO2 reduction target - absolute savings to target -

4,0

3,0

3,0

2,0

2,0

1,0

1,0

0,0

0,0 3

4

CO2 delta to target % 50%

5

6

7

8

40%

30%

30%

20%

20%

10%

10%

0%

0% 3

4

5

6

7

8

9 10 11 12 13 14 15

3

4

5

CO2 delta to target % 50%

40%

2

- absolute savings to target -

Make CO2-Emissions Use CO2-Emissions

2

9 10 11 12 13 14 15

- relative savings to target

S2: 30% CO2 reduction target

5,0

4,0

2

/15/2015 5:40 PM

CO2 per vehicle delta to target [t]

6

7

8

9 10 11 12 13 14 15

- relative savings to target -

2

3

4

5

6

7

8

9 10 11 12 13 14 15 18

Underlying supply chain structures can be analyzed in more details Analyzing impact on supply chain structures – excerpt by location Development of process units in China Mio. process units 400

Development of process units in Germany Mio. process units

Overflow TTS exp. 400

Overflow TTS lin. 350

Overflow TTS exp. Overflow TTS lin.

Total TTS

350

Total TTS

Standard TTS 300

250

250

200

200

150

150 100

100 2

3

4

5

6

7

8

9

Periods /15/2015 5:40 PM

Standard TTS

300

10 11 12 13 14 15

2

3

4

5

6

7

8

9

10 11 12 13 14 15

Periods

19

There is not much result difference between linear or exponential penalties in the specific case. Analyzing impact of linear vs. exponential penalties on overflow TTS results Penalty factor log. scale

CO2 TTS overflow exp. vs. lin. %

10000

105%

1000

100

100%

10

linear penalty factors expon. penalty factors CO2 - TTS overflow exp. vs lin.

1

95% 2

3

4

5

6

7

8

9

10

11

12

13

14

15

/15/2015 5:40 PM

Periods

Potential area for further research to understand effect of penalties as „sustainability interest rates“ on results 20

Agenda

 Problem introduction: Sustainability in network design  Time-to-Sustainability (TTS) optimization strategy  Numerical results

/15/2015 5:40 PM

 Summary

21

TTS provides new insights for sustainability-driven decision support in supply chain network design. Summary Research questions

/15/2015 5:40 PM

1

How to support the aspect of time in sustainability transformation in long-term network design?

2

How to deal with multiple incompatible sustainability objectives and indicators in quantitative optimization strategies?

3

How to deliver insights to decision makers to answer, if and how sustainability objectives can be achieved?

Findings and areas for further research • TTS can explicitly model the time aspect in network design and the transition towards reaching sustainability targets. • TTS can model multiple sustainability objectives without subjective weighting with the same approach. • Thanks to Overflow TTS, the model is LP and hence reaches fast solution times even with multiple KPIs and industry data sets. • TTS allows to simulate various sustainability target value scenarios and reveals, if and when these targets are all feasible steady-state. • Underlying restructuring transitions in the supply chain network can be analyzed in all scenarios.

TTS approach applicable to all types of network design problems. 22

Related publications: • M. Kannegiesser, H.-O. Günther: Sustainable development of global supply chains – part 1: sustainability optimization framework. Flexible Services and Manufacturing, No. 1-2 (2014), 24-47.

 

• M. Kannegiesser, H.-O. Günther, O. Gylfason: Sustainable development of global supply chains - part 2: investigation of the European automotive industry. Flexible Services and Manufacturing, No. 1-2 (2014), 48-68.



• H.-O. Günther, M. Kannegiesser, N. Autenrieb: The role of electric vehicles for supply chain sustainability in the automotive industry. Journal of Cleaner Production. No. 90 (2015), 220-233.

/15/2015 5:40 PM

Minimum TTS (steady state)

• M. Kannegiesser, H.-O. Günther, N. Autenrieb: The time-tosustainability optimization strategy for sustainable supply network design. Journal of Cleaner Production, (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.030,.

23