Two-stage risk-averse stochastic programming approach for multi-item single source ordering problem: CVaR minimisation with transportation cost

Integrating inventory and transportation decisions is vital in supply chain management and can enable decision-makers to achieve competitive advantages. This study considers a multi-item replenishment problem (MIRP) with a piece-wise linear transportation cost under demand uncertainty, which usually...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:International journal of production research Ročník 61; číslo 7; s. 2129 - 2146
Hlavní autoři: Taghizadeh, Elham, Venkatachalam, Saravanan
Médium: Journal Article
Jazyk:angličtina
Vydáno: London Taylor & Francis 03.04.2023
Taylor & Francis LLC
Témata:
conditional-value-at-risk (CVaR)
Decision making
L-shaped algorithm
multi-cuts
Operating costs
Replenishment
Risk
sample average algorithm
Stochastic models
Stochastic programming
Supply chains
Transportation
ISSN:0020-7543, 1366-588X
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Integrating inventory and transportation decisions is vital in supply chain management and can enable decision-makers to achieve competitive advantages. This study considers a multi-item replenishment problem (MIRP) with a piece-wise linear transportation cost under demand uncertainty, which usually occurs both in retail and production environment when several items must be ordered from a single supplier. Conventionally, two-stage stochastic programming formulation is risk-neutral, and it lacks robustness in the presence of high data variability. Hence, we introduce the Conditional Value at Risk (CVaR) approach for MIRP. Additionally, we deploy both single and multi-cut L-shaped and the sample average approximation method to circumvent the computational complexity to solve large-scale instances. The data-driven simulation study is used to benchmark the results from deterministic, risk-neutral, and risk-averse stochastic models. The results indicate that under higher data variations, the risk-averse model provides better perspectives for a decision-maker. The results show a 40-50% reduction in lost sales with marginal growth in total cost while considering CVaR instead of a risk-neutral approach.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0020-7543
1366-588X
DOI:10.1080/00207543.2022.2060770