Metaheurıstıc approaches for maxımum blood collectıon problem

Thanks to recent technological and medical advances, blood components can now be extracted from whole blood after a donation. One of such components is the platelet, which has a wide range of uses in medical fields, including cancer treatment and other surgical procedures. Due to the perishable nature of platelets, it is recommended that the separation occurs within six hours after the donation. Moreover, platelets constitute less than one percent of the whole blood volume, yet they are highly demanded. Thus, it becomes apparent that there is a need for an effective platelet supply chain that meets patient needs. Given the importance of platelets in healthcare, their perishability, and their limited supply, an effective platelet supply chain leans on well-managed whole blood collection operations. In this study, we consider a blood collection problem (BCP) focusing on the collection of whole blood donations from the blood donation sites (BDS). Different from the basic form of BCP, we consider processing time limit of blood and arbitrary donation patterns of donors as well as relaxing the assumption of assigning each blood collection vehicle (BCV) to a set of BDSs. Therefore, we define the non-clustered maximum blood collection problem (NC-MBCP) as a variant of BCP. In this problem, the goal is to maximize the total platelets collected from numerous BDSs utilizing a set of BCVs that collect blood from those BDSs and transfer it to a central processing facility before it becomes non-usable for platelet production. In this study, we examine routing decisions for platelet collections while relaxing the clustering requirement from the BDSs, resulting in a significant increase in the complexity of the problem. In order to solve the problem, we propose a hybrid genetic algorithm and an invasive weed optimization algorithm that provide considerable improvements over the best solution in the literature for the clustered variant of the problem and outperform it by 9.40% improvement by the hybrid genetic algorithm and 9.14% improvement by the invasive weed optimization algorithm on average.