Génie mécanique
Transcription
Génie mécanique
SECTION DE Génie mécanique Laboratoire de Gestion et Procédés de Production Study of a supply-chain concept based on a multi-criteria analysis and a risk evaluation of different scenarios Context Methodology In regards to an upcoming major railway project, the Bombardier Transportation production site in Villeneuve (CH) is confronted with a make-or-buy decision for its logistics strategy. A first estimation of the required storage space (based on benchmarking) forecasts a surface lack of several hundred square meters in the current central warehouse. Therefore, the management is considering externalization of warehousing, along with other logistics functions, in order to fulfill the surface deficit. Thus, two strategies are considered: Due to the complexity (number of finished goods, components and suppliers, interdependent elements) and the stochastic nature of the system (supply performance), computer simulation (using the SCOPE platform, developed at LGPP) has been chosen as a suitable resolution method. After the data gathering (through documentation and consultations with different departments in the company), a conceptual model has been set up, including all relevant information about material and information flows (train components, suppliers, assembly process, etc.). In particular, value stream mapping has been a useful tool (see figure below). Consultations Méthodes Achats - Passage commandes - Suivi commandes Advanced warehouse (3PL/4PL) F2 Planning / Contrôle production Commande ferme (selon contrat) LT EW LT EW RNC Lead-time de commande: - 11 jours (UE) - 7 jours (CH) Goals . . . M MPS (hebd.) OF (journ.) Total: 52 trains ≡ 252 véhicules (vhc) Niveau stocks RNC Contrôle réception M W LT L-T: 1 jour capacité "illimitée" Mise en kits Supermarché max 2 vhc 6 jours 60 jours FIFO capacité "illimitée" capacité "illimitée" MWLT: Mehrwegladungsträger = emb. réutilisable MWLT vides max 1 vhc C-T: L-T: 3 jours 9 jours cap. "illimitée" 1 jour 4 jours (max.) 12 jours (max.) 10% S 2.x C 30% For each scenario, gross storage surface requirements have been determined (including circulation, handling and stackability), based on the 95% highest inventory level values. Results show that the company estimation is realistic when a majority of components are delivered in A-level (S1.1-S1.4), while it appears to be under-estimated in case of lower supply reliability (S2.1S3.4). S 3.x Gross surface variation [%] 10.0% Total gross surface [m2] 1200 1070 m2: Estimation by Production Control 1000 800 600 400 200 0 0.0% Ivan KOSTAKEV Décision Docs OUTPUT Processus Documents de base des flux logistiques (Modèle conceptuel) BoM simplifiée Max. 24 jours (IR100) Due to the lack of history data concerning past orders, a proper supply-related risk analysis could not be performed; instead, an alternative approach has been proposed, defining 3 different supply reliability levels (A, B, C), characterized by normally distributed delivery delay patterns (see figure below). Total L-T: - 140 jours (IC/IR200) - 116 jours (IR100) 5 jours Delay 0 days = contractual delivery date 0.9 0.8 0.7 A B 0.6 C 0.5 0.4 0.3 0.2 0.1 0 -0.1 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Delivery delay [days] STRATEGY A – Annual logistic costs structure JIT Transport costs [CHF/year] Kitting costs 3PL [CHF/year] Kitting costs BTV [CHF/year] Storage costs 3PL [CHF/year] Storage costs BTV [CHF/year] scenario Globally, it appears that a complete externalization (B) is economically more profitable than a partial one (A). However, the results being close (8-11% difference), they strongly depend on the cost assumptions. In this scope, sensitivity analyses have allowed identifying threshold values. scenario 2.0% Author STRATEGY A STRATEGY B scenario 4.0% 3 4 Lot size [car Equivalents] x : Lot size (1, 2, 4, 6) 1400 6.0% 2 Docs INPUT Annual logistic costs comparison 8.0% 1 0 jours Max. 48 jours (IC / IR200) Based on the average surface during the simulation period, logistic costs (including storage, kitting and JIT deliveries from 3PL) have been calculated. Results have shown that a storage space externalization is only necessary with the eight most pessimistic scenarios (in terms of delivery performance). 1600 Scénarios 0.x Scénarios1.x Scénarios 2.x Scénarios 3.x 12.0% C 50% Total gross storage surface (indoor) Sensitivity total gross storage surface (indoor & outdoor) 14.0% B 30% Value stream Mastersheet Mastersheet Suppliers map Process 1800 18.0% 16.0% A 20% 9 jours 60 jours 1 jour (max.) 5 jours Capacité "illimitée" The conceptual model has then been transposed into a simulation model, by applying different hypotheses and simplifications (SKU, production capacity, time unit, project phase, etc.). Then, various tests have been performed, so as to approve the model behavior compliance with predefined criteria. Coûts logistiques annuels [CHF] S 1.x B 30% Train Configuration IC/IR Mastersheet Components Logistic costs comparison The simulation runs have consisted in 19 scenarios, combining different lot-sizing policies (1, 2, 4 or 6 car equivalents) and delivery reliability levels (proportion of components with respectively A-, B- and C-level delivery). To each of the latter corresponds a given level of security stock, such that the production service level is maintained at 100%. A 40% CT: Max. 8 vhc (IC/IR200) Max. 4 vhc (IR100) Value-adding time: - 79 jours Fonction magasin Storage space requirement C Mise en rame, testing final & shipping Testing statique Production C-T: L-T: L-T: 4 jours retour MWLT vides EWLT: Einwegladungsträger = emballage perdu B 30% Contrat cadre: 1 à 1.5 vhc/sem. IC200: 17x IR200: 27x IR100: 8x ------------------------------------- - livraisons hebdomadaires - tailles de lot selon composants - MWLT ou EWLT MWLT ~40 Fournisseurs 176 composants distincts MPS (hebd.) OF (journ.) W LT LT MW Fn MPS (bimestr.) OF (journ.) Lead-time transport: - 1-2 jours LT EW Fi RNC RNC LT EW . . . Quantify more precisely the required storage space, by taking into account in a dynamical way both component needs and supply reliability. Perform an estimation of logistic costs for different supply scenarios and draw a comparison of both strategies A and B. Show the impact on storage space and costs in case of variations in the lot-sizing policy and the supply reliability. Provide a quantitative support in the decision-making process of the logistic concept elaboration. Client Ordre d’expédition produits finis Horizon prévisionnel: durée entière projet F4 The project purposes are the following: Demande d’achat (DA) Commandes à intervalle variable, quantité fixe (via plateforme web) F3 A 60% Ajout/suppr. Composants et autres infos Objet de l’étude F1 Advanced warehouse (3PL/4PL) Doc #6 (Méthodes) Probability density function BT Villeneuve Warehouse Suppliers Consultations Engineering Processus itératif Légende STRATEGY B BT Villeneuve Warehouse Suppliers Doc #3 (Achats) Consultations Achats Consultations Contrôle production Internal warehousing and externalization of the overhead inventory External warehousing and kits preparation for all main components STRATEGY A Analyse documentation générale projet Annual logistic costs [CHF] A: B: Kickoff Présentation 5 6 Sensitivity analyses of the storage space to lot-sizing policies have shown that the system is more sensitive to supply reliability (i.e. security stock) than lotsizing variations, given the current degrees of freedom of the system (i.e. the limited number of components whose lot-sizes are not yet set by the purchase contracts). Conclusions The MRP-based simulation model has allowed measuring different KPI, among which the inventory levels throughout the whole project period (2013-2019). Depending on the component (generic or specific) and the supply performance, security stock levels have been determined so as to reach a 100% production service level. Results have pointed out that a critical issue for both storage space and costs is the supply reliability, requiring therefore an even closer monitoring. Threshold values defining profitability limits between both logistic strategies have been provided, so as to bring quantitative figures to negotiations with 3PL/4PL providers. Acknowledgements Stéphane FAKERY Supervisor Prof. Rémy GLARDON Prof. Rémy GLARDON Olivier GOBET