Duygu Aghazadeh

Dr. Duygu Aghazadeh

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications

24 June 2024 in Central European Journal of Operations Research
Verified
Journal Article
Open access
PDF file
DADuygu Aghazadeh
Duygu Aghazadeh
KEKadir Ertogral
Kadir Ertogral
Logistics related costs constitute a major part in total cost of a product in general. Considering a company that delivers goods to its customers using its owned fleet, fleet ownership and operational costs together with the inventory costs compose the total logistics costs. In this study, we suggest an approximate Dynamic Programming algorithm, with a look ahead strategy, that uses the fix and optimize method as the imbedded heuristic for solving integrated fleet composition and replenishment planning problem. The total annual distribution cost factors considered in the problem are vehicle ownership costs, approximate routing costs, and inventory related costs. In this problem, we aim to minimize the total logistic cost by optimizing the fleet composition, replenishment patterns, and customers assigned to each vehicle in the fleet. We produced a set of reasonably large instances randomly and showed the efficacy of the suggested solution method.

1 January 2024 in Journal of Industrial and Management Optimization
Verified
Journal Article
Closed access
PDF file
DADuygu Aghazadeh
Duygu Aghazadeh
KEKadir Ertogral
Kadir Ertogral
The aim of the current study is to put forward a new metaheuristic method for the problem of vehicle fleet dimensioning and shipment schedule planning. The problem includes making decisions on both the configuration of a fleet for serving customers with a single product and the replenishment schedules according to a set of given candidate delivery frequencies. The problem combines two significant logistical issues specifically fleet sizing and replenishment planning. The principal idea of the model is to minimize all the relevant costs which covers the costs related to vehicle ownership, inventory, and routing. We bring forward two versions of a Problem Space Search metaheuristic that apply fix and optimize heuristic repeatedly. The metaheuristics rely on both perturbing data and the sequence of subproblems generated in the fix and optimize algorithm. The novel metaheuristic techniques developed in the current study exhibit significant efficacy compared to existing studies on a group of non-small randomly generated instances. Results show that there is a noticeable decrease in the gaps from optimal/ best bounds taken from CPLEX and solution duration compared to previously employed solution technique.

28 February 2025 in Havacılık ve Uzay Çalışmaları Dergisi
Verified
Journal Article
Closed access
DADuygu Aghazadeh
Duygu Aghazadeh
The study delves into the utilization of machine learning to predict and reduce inflight food waste, improving sustainability in aviation logistics. Inflight food waste, a major environmental problem, is determined by passenger choices, flight parameters, and catering services. The research presents two efficient machine learning algorithms, that are, Multiple Linear and Random Forest Regression to perform food waste prediction during the flights. The models are trained using a synthetically created dataset of 10,000 records and 15 features, which include factors such as meal type, waste weight, and passenger number. The study undertakes considerable feature engineering, including the development of new features such as "Waste per Passenger" and "Meal Efficiency" to increase forecast accuracy. A correlation analysis is also used to determine the most influential characteristics. The models' performance is assessed in a Python-based computational environment, with MLR concentrating on linear links between food waste and predictors and RFR on non-linear interactions. The results show that both models can effectively forecast inflight food waste, with RFR being more adaptable to complicated patterns. The research concludes with recommendations for airline managers to apply data-driven waste reduction techniques that correspond with overall sustainability goals in aviation logistics. The models created are a useful tool for optimizing inflight food, lowering environmental impact, and contributing to the industry's sustainability initiatives.

1 January 2017
Verified
Conference Paper
Closed access
BIBurcu Cansu Inanc
Burcu Cansu Inanc
DADuygu Aghazadeh
Duygu Aghazadeh
EKEmre Anıl Kakillioğlu
Emre Anıl Kakillioğlu
NÜNilgün Fescioğlu Ünver
Nilgün Fescioğlu Ünver

10 September 2026 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Fatih KasimogluFatih Kasimoglu
Fatih Kasimoglu
DADuygu Aghazadeh
Duygu Aghazadeh
DUDURDU HAKAN UTKU
DURDU HAKAN UTKU
This study investigates a sustainable vehicle-routing problem in which a heterogeneous fleet serves geographically dispersed customer demands from a central distribution facility. The problem simultaneously minimizes transportation costs and CO2 emissions, with deliveries performed by either in-house or externally rented vehicles. A bi-objective mixed-integer programming (MIP) model is formulated, and two lexicographic anchor solutions are generated using opposite objective-priority orderings. A tailored Problem Space Search (PSS) metaheuristic is evaluated on five application-informed simulated datasets containing 10–50 nodes. Six parameter configurations combining m ∈ {10, 20} and β ∈ {0.15, 0.20, 0.25} are evaluated using 30 random seeds. For cases where CPLEX certifies primary-objective optimality, the mean PSS deviation ranges from 0.00% to 7.81%, while the best PSS run remains within 3.20% of the optimum in every case. On the 50-node instance, each PSS run improves the time-limited CPLEX primary incumbent under both priority orderings, although unresolved CPLEX gaps preclude near-optimality claims. PSS also improves the embedded heuristic in most cases, while increasing m generally improves solution quality at additional computational cost. The results demonstrate the computational effectiveness of PSS for the sustainable fleet-assignment and routing instances examined.