[PC15a] On a robust inventory problem
Conférence Internationale avec comité de lecture :
PGMO 2015,
October 2015,
pp.7 p.,
Palaiseau,
France,
Mots clés: Robust discrete optimization, inventory problem, dynamic programming.
Résumé:
In this paper we address the problem of optimizing multi-period inventory in the
case of uncertain demands. At each time period, the company produces a certain
quantity of goods which is used to serve a client demand. The unit selling price
was fixed in advance by contract for the time horizon and an expected value of the
quantity to deliver at each time period is known. In case of overproduction, goods
are added to the stock. In case of underproduction the missing goods are either
taken from the stock or bought on the international market. In addition at each
time period the manager can decide to buy more goods and add them to the stock
or to sell a part of the goods in stock, on the international market. The unit price
of purchase (or sale) of these goods on the international market are estimated in
advance for every period, according to the previous years.
But in fact, the demand and the purchasing costs on the international market
are uncertain and may differ from their expected values. Following the Betsimas
and Thiele approach, we assume that there is no known probabilistic distribution
of these values, but each one may vary in a given interval. We also assume that the
variation of the purchasing or selling costs is small, while the real demand can be
far enough from its expected value. Then the prices on the international market can
be approximate in the following way: the unit purchasing cost is set to its expected
value plus the maximum possible gap and the unit selling cost is set to its expected
value minus the maximum possible gap. Doing so, we guarantee a lower bound on
the profits.
The manager takes decisions in two stages: first he before discovering the actual
value taken by the demand, second once uncertainty has been revealed.
In this paper we address the problem of optimizing multi-period inventory in
the case of uncertain demands. We consider a wholesaler who purchases goods
on the international market and stocks them in a warehouse before selling them to
local customers. To serve the demand he can either demand at each time period,
the manager decides the quantity to buy His decisions are made in two stages: first
before discovering the actual value taken by the demand, second once uncertainty
has been revealed.