### Inventory Control: Multi-Echelon Optimization in a spare parts network

```FB 10 – Institut für Mathematik
Algorithmische Algebra und
Diskrete Mathematik
Prof. Dr. Andreas Bley
OBERSEMINAR
ALGORITHMISCHE ALGEBRA UND DISKRETE MATHEMATIK
Inventory Control:
Multi-Echelon Optimization in a spare parts network
Referent:
Herr Christopher Grob (Externer Doktorand, Volkswagen AG)
Termin:
Dienstag, 12. April 2016, 17.15 Uhr
Ort:
Raum 1403, Heinrich-Plett-Str. 40, AVZ,
Kassel-Oberzwehren
Abstract:
The Volkswagen After Sales network is supplying customers all over the world with
spare parts for vehicles of the Volkswagen Group. Volkswagen wants to ensure customers get spare parts in a timely manner. The main challenge in this network is to
cope with the stochastic and highly volatile demand.
We consider the problem of optimizing the distribution of stock in a 2-level network
with one central warehouse and n non-identical local warehouses, which order using
an (R,Q)-policy. We assume that fill-rate targets and order quantities for each local
warehouse are given, orders are fully backordered, partial deliveries are not allowed,
and orders are served on a first come – first serve basis. Lead time is stochastic and
lead time demand is assumed to follow a negative binomial distribution. Mean and variance of the wait time due to stock-outs at the central depot are approximated based
on the Kiesmuller et al. (2004) method.
gez. Prof. Dr. Andreas Bley
The main challenge in this optimization problem is the computation of reorder points
for given fill rate targets, which is computationally very expensive. In order to reduce
the number of these function evaluations, we propose a method that adaptively refines
two approximate models which under- and overestimate the actually needed stock at
local warehouses. In each iteration we resolve and strengthen the corresponding integer linear programs in the proximity of the current approximate solution. Thus, we are
able to give performance bounds for the exact problem and solve it even to optimality.
We report on numerical results with real world data, which show the eﬀectiveness of
this approach.
Tee/Kaffee ab 16.45 Uhr im Raum 1404