Stay Ahead of the Game by Routing
In today’s competitive world, customers require efficient, cost-effective, and timely delivery of goods and services. Efficient design and operation of distribution systems therefore becomes more critical in helping companies stay competitive in their markets, and improve customer service levels. One of the biggest costs for companies is distribution costs: approximately 15-20% of product costs consists of distribution costs according to current research. Distribution planning is also a time-consuming, labor-intensive process, and when we consider the tight competition among companies, reducing the percentage that is distribution costs becomes more important for distribution processes.
The modelling of distribution networks and planning of effective routes are achieved by solving vehicle routing problems (VRP), a combinatorial approach to determining optimal routes for vehicles. There are several types of VRPs because constraints and objectives differ for different companies. The objectives of VRPs, generally, are to determine the routes which have minimum cost, maximize customer satisfaction, minimize CO2 emissions, balance the vehicles service loads, etc. In real life applications, frequently encountered VRP types are homogeneous or heterogeneous fleet VRP, VRP with time windows, pickup and delivery VRP, and single or multi-depot VRP.
Although decreasing these costs is crucial , some companies still don’t use the optimization approach, and try to solve this problem using basic heuristics based on their past experiences. However, VRPs are NP-hard (non-deterministic polynomial-time hard) problems, so they often require too long CPU times to reach optimal solutions by using mathematical programming approaches, even for small or medium size problems. They may even fail to find the optimal solution. As part of the banking analytics solution to deal with these obstacles, we have developed effective hybrid approaches where mathematical modeling is empowered with metaheuristics, such as genetic algorithm(GA), tabu search(TS), and simulated annealing(SA). P1M1 is a specialist in various kinds of VRPs which solve cash replenishment and cash-in-transit (CIT) problems. Our model creates the optimal set of routes for a fleet of CIT vehicles for banks, with dynamic adjustments to meet new requests during operations. While reducing the transportation, replenishment, removal, and interest costs, the model fulfils all relevant key performance indicators to improve service quality. Through the help of a two-phase hybrid approach, our customers can now solve their CIT problems in a much shorter space of time, and lower their costs, resulting in higher customer satisfaction, and helping them stay ahead of the game.