Case Study 2 Providing decision support for the

Case Study 2 Providing decision support for the location of a new ABS facility producing electric cars Group 3 Alexandru Olteanu Massimo Gurrieri Florian Schnetzer Sebastian Langton July, 9 th 2010 Thomas Veneziano Yann Bouchery

Outline (1) Definition of the problem: Country or site selection? (2) Problem structuring: from the decision context to the value tree (3) MCDA method (4) Results (5) Conclusions

Definition of the problem • Identification of ABS’ main objective: • • Facility location problem literature review: • • Selection of the “best” site for a new facility for the manufacturing of electric cars Clear distinction between country and site selection Corollary • In the case study, the selected sites have been proposed after identifying suitable country “candidates”

Problem structuring: ideal value tree

Problem structuring: Data critique • Data-related challenges • • Partly irrelevant, redundant, missing, incomplete data • • Great quantities of heterogeneous data Mainly country-oriented, little site-oriented data Preference-related challenges • Eliciting preferences of the decision maker is not possible! Formulation of working hypothesis

Problem Structuring: adapted value tree “Adapted value tree” and the relevant criteria with respect to available data and preferences:

Problem Structuring CO 1 CO 2 HR 1 HR 2 HR 3 OE 1 OE 2 OE 3 OE 4 OE 5 OE 6 Paris Lavenec 811 749 6445 6286 NA 4700 4. 93 1. 75 NA 2. 06 3. 29 3. 84 5. 00 4. 26 5. 00 4. 28 NA 8. 40 5. 00 4. 50 NA 2. 94 Gora Malinek Tortel Domsod Liska Lisboa 749 716 730 765 811 6170 6340 6387 6335 6312 6462 5000 4050 2888 2000 NA 1. 75 1. 82 4. 16 1. 70 1. 54 2. 16 0. 90 1. 68 2. 38 2. 28 NA 3. 84 3. 65 3. 12 2. 11 1. 40 4. 26 2. 35 2. 76 4. 92 5. 00 4. 28 16. 78 4. 01 0. 50 4. 20 0. 80 4. 20 3. 60 3. 99 3. 20 5. 00 NA 4. 50 4. 00 5. 00 4. 84 2. 69 4. 70 3. 45 NA CO 1: Investment cost CO 2: Production cost HR 1: Workforce Education Index HR 2: Workforce Skill Index HR 3: Unionization Index OE 1: Taxation Index OE 2: Automotive industry Index OE 3: Stability Index OE 4: Transport OE 5: Industrial infrastructure OE 6: Living environment

Chosen MCDA-method • Outranking methods preferred to MAVT-method • Application of “RUBIS” for solving the decision problem Reasons for our choice: • Dealing with impreciseness • Providing recommendations and potentially bad choices Approach: • Development of 4 different scenarios for weighting the criteria

Results Scenario 1 2 3 4 Short description Equal weights on objectives and equal importance on subordinated criteria Equal weigths on objectives and different importance on subordinated criteria Different weigths on objectives and equal importance on subordinated criteria Different weigths on objectives and different importance on subordinated criteria Potentially best choice Domsod (HU) Gora (CZ) Lavenec (CZ) Tortel (HU) Paris (F) Lavenec (CZ) Gora (CZ) Domsod (HU) Lavenec (CZ) Gora (CZ) Potentially bad choices Lisboa (P) Paris (F) Lisboa (P) Liska (PL) Lisboa (P) Paris (F)

Conclusions 1. Which solution we can recommend: - Gora and Lavenec appear in three out of four scenarios; Paris, which is the best solution for one scenario is definitely too expensive - Further analysis and complementing research on suggested sites 2. What could be improved for a better site selection: - Real life interaction with the decision maker during the whole process - Extension of the set of alternatives - Improve the quality and the completeness of the provided data

THANK you