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