Vehicle Routing Optimization
April-May 2020
About the company
Quick is a logistics digital platform specialized in Inventory, Transportation and Distribution. Outside of Colombia they operate in Brazil, Chile and Mexico and cover 78 cities in Latin America. Some of their clients are major corporations such as L'Oréal, Unilever, Johnson & Johnson, Nutresa, Homecenter, etc.
The Challenge
This Challenge is based on the case of one of the most important clients of Quick. The objective is to develop a software that optimizes Quick's vehicles routing to deliver more than 200 customers from 15 different warehouses in Bogotá. The software has to automatically decide for each vehicle which warehouses and customers to visit and when, in order to minimize costs and maximize service level.
+80%
-20%
100%
improvement in
Service Level
less vehicles required
compliance with customers' schedules
3
new insightful performance indicators
Winners of the Challenge
Juan José Jaramillo
His software finds - in less than one second - a solution that reduces by 37% the number of non attended clients. When left running for about 15 minutes, the reduction reaches 84%. Juan José designed also 2 new insightful performance indicators and made some recommendations that he supported by simulations with his software.
Daniel Fernando Vizcaino
Daniel's software aims at balancing as much as possible the two most important criteria for the company: runtime execution and the quality of the solution. By maximizing the use of the vehicles, his software manages to decrease by more than 60% the number of non attended clients in less than 5 minutes.
Mateo Cepeda
Jorge Enrique Barrera
Mateo and Jorge's software finds a good solution very fast. In less than 20 seconds the number of non attended clients is reduced by 15%. They also made a thorough analysis of the company's business case and proposed some modifications in the organization that could lead to 25% improvement.
A word from the professor
The Vehicle Routing Problem is known to be very complex to optimize. For instance, a problem with only 60 clients leads to more than one icosipentillion combinations (imagine a number with 75 zeros !). To solve this problem, our students combined many optimization algorithms: state of the art metaheuristics and heuristics of their own design. They implemented hybrid techniques such as mutli-stage and high-level relay hybrid. To be less technical, this means that they created a synergy between algorithms that are fast but not so accurate and algorithms that are accurate but rather slow. In the world of Operations Research, this is one of our magic tricks behind finding a very good solution in the shortest time possible.
Prof. Rabie Nait-Abdallah
Leader of the Challenge
Javeriana University