Browsing by Author "Araz, OU"

Now showing 1 - 4 of 4
  • No Thumbnail Available
    Item
    Fuzzy demand-driven material requirements planning: a comprehensive analysis of fuzzy logic implementation in DDMRP
    Araz, OU; Ilgin, MA; Eski, O; Araz, C
    Demand-Driven Material Requirements Planning (DDMRP) is a new method of inventory control designed to address the challenges of today's complex and volatile supply chain environment. DDMRP replaces traditional Material Requirements Planning (MRP) with a more agile and responsive approach that is based on a set of colour-coded buffers. DDMRP ensures that the right inventory is in the right place at the right time, enabling companies to respond quickly and effectively to changing customer demand. DDMRP requires the setting of various parameters, such as variability factor (VF) and lead time factor (LTF), which can have a significant impact on systems' performance. Pre-assumed fixed values of DDMRP parameters are used in most of the existing studies. This may lead to either high stockout levels or excess inventory especially in environments involving high level of variability. In this study, we proposed a fuzzy logic-based approach which dynamically adjusts the values of VF and LTF parameters considering demand and lead time variability. The effectiveness of the proposed approach was investigated by comparing its performance with DDMRP based on several numerical experiments. The results showed that the proposed Fuzzy Demand-Driven Material Requirements Planning (FDDMRP) outperforms DDMRP in terms of backorder rate and total cost.
  • No Thumbnail Available
    Item
    A REACTIVE SCHEDULING APPROACH BASED ON FUZZY INFERENCE FOR HYBRID FLOWSHOP SYSTEMS
    Araz, OU; Eski, O; Araz, C
    Hybrid flowshops consist of multiple production stages each of which has multiple parallel machines. Scheduling of hybrid flowshops is a NP-hard even in its simplest form. The presence of uncertainty in real-world problems forces the decision makers to reconsider their scheduling decisions in reactive manner. In this study, we proposed a proactive-reactive scheduling approach which allows to be changed dispatching rule set applied in time. The methodology consists of three parts: Shop Floor Management system with a triggering mechanism based on fuzzy inference system, performance prediction of the alternative dispatching rule sets based on Taguchi design, simulation, artificial neural networks, and a multi-criteria decision making methodology for determining new scheduling dispatching rule set. The proposed approach is applied on a real world problem from literature and the results are compared with static approach.
  • No Thumbnail Available
    Item
    Modeling of the lyotropic cholesteric liquid crystal based toxic gas sensor using adaptive neuro-fuzzy inference systems
    Araz, OU; Kemiklioglu, E; Gurboga, B
    Detection of toxic gases is important in a variety of settings, including industrial facilities, laboratories, and even in homes. In these settings, toxic gas detection can help prevent accidents and protect the health and safety of workers, researchers, and others who may be exposed to these gases. This study evaluates an Adaptive NeuroFuzzy Inference System (ANFIS) models in predicting the machining responses in the detection of toxic gases vapor, such as toluene (T), phenol (P) and 1,2 dichloropropane (D) using lyotropic cholesteric crystal (CLC) have been shown to have potential as gas sensors due to their unique optical and liquid crystal (LC) properties, and the ANFIS model may be used to better understand and optimize these properties for toxic gas detection. Experiments were carefully carried out to gather data on the response of a lyotropic CLC toxic gas vapor sensor. The effectiveness of using ANFIS combined with Grid Partitioning (GP) was then carefully studied and evaluated in terms of modeling and predicting the responses of the sensor. The best ANFIS-GP model is chosen from these criteria; RSS, PCC, R2, RMSE, MSE, MAE, and MAPE. In addition, validation was performed between the model and experimental data using the LOOCV method. The results show that the ANFIS-GP5 model with 96 fuzzy inference systems (FIS) rules with high R2 values. According to the ANFIS-GP5 model, R2varied ranges from 0.77 to 1 for train, test, and total data of lyotropic CLC sensor exposed to toluene, phenol and 1,2 dichloropropane toxic gases vapors.
  • No Thumbnail Available
    Item
    Storage location assignment of steel coils in a manufacturing company: an integer linear programming model and a greedy randomized adaptive search procedure
    Edis, EB; Araz, OU; Eski, O; Edis, RS
    Warehouse operations have a significant role to survive in today's competitive world. Hence, companies introduce various solutions to improve efficiency of warehouses which is mainly affected by the performance of storage operations. This study deals with a real-life storage location and assignment problem encountered in a fastener company where several orders consisting steel coils are to be assigned into storage areas. Three individual objective functions; minimizing the number of lanes to be used, minimizing area usage, and maximizing volume utilization are considered. For the investigated problem, first, an integer linear programming (ILP) model is developed. Then, a greedy randomized adaptive search procedure (GRASP) which provides quick and efficient solutions is proposed. The proposed methods are applied to the real problem case and the results are compared with the current storage assignment. Moreover, through an extensive computational study, the performances of proposed methods are evaluated on a set of test problems with different range of characteristics. The computational results show that the ILP model proves optimality in most of the problem instances within reasonable computation times, while the GRASP gives quick solutions with small optimality gaps.