Please select:

GRILO plan to validate the functionality of current GRID and Peer-to-Peer technology as a foundation for advanced optimization algorithms applied to logistic and transportation optimization problem, aiming at dramatically reduce the elaboration/computing costs. Main objectives of this experiment are: assess whether recent advances in open GRID stacks reached a level of interest for conventional industrial applications; deploy an efficient graph and matrix level algebraic formalization of real industry logistic problem, (i.e. considering large instances, real road map, normative and operational constraints such as finite capacity of human and machinery resources, time windows, deadlines to be met, etc) and integrate the industrial Enterprise Resource Planning systems (ERP) with efficient scheduler/planner.

Fact Sheet

What is the experiment about?

Logistic and Scheduling companies face computationally intensive problems to manage logistics and many times are limited by heuristic solutions chosen only on the basis that no more computing power was available to search for possibly better solutions. Recent advances on heuristic and metaheuristic solution approaches has led to the development of very effective methodologies, several of them inspired by natural phenomena. These methods can usually get high quality solutions, but at significant computing cost. The distribution of data and processing at the heart of GRID computing seems thus synergetic with their high computing requirements.

Why this experiment? (objective)

Main objectives of GRILO are: assess whether recent advances in open GRID stacks reached a level of interest for conventional industrial applications; deploy an efficient graph and matrix level algebraic formalization of real industry logistic problem, (i.e. considering large instances, real road map, normative and operational constraints such as finite capacity of human and machinery resources, time windows, deadlines to be met, etc) and integrate the industrial Enterprise Resource Planning systems (ERP) with an efficient scheduler/planner.

Industrial input/added-value of this experiment

Once deployed as ‘service’, the manufacturing companies could benefit by common services like scheduling and planning systems based on GRID technology that could allow much better response and more detailed simulation of production, transport, logistics. Therefore the possibility to use fast response tools can give more quality added value to company services (like dispatch planning, timing, transportation routes, etc) and increase their competitiveness. On the other hand, the services provided should be consistent with the industry real needs and be able to match the flexibility required by such kind of companies. This understanding can be achieved involving pilot users and getting more and more inputs from logistic stakeholders as well as companies extensively using transportation facilities.

How is this experiment going to be handled?

It is a current trend, following the saturation niche market standard-information services, the growing interest towards value-added integrations (add-on) to the companies ERP. Among these, optimization of logistic functions, both internal and external ones, is a foremost option given the impact of logistics cost on global companies expenses and the need of this optimization basically in any manufacturing sector. However, effective logistic optimization, with special reference to transportation costs, is highly computationally intensive. Moreover, companies distributing to many customers need to populate data structures containing millions of shortest time / shortest distance entries on detailed road maps. This poses severe storage requirements, which coupled to wallclock time constraints for getting a solution, rule out the possibility of effectively supporting large-scale SME dispatchers. A GRID infrastructure could provide a means to give these companies access to:
•     Significant reduction in distribution costs;
•     Increased central control on distribution activities and relative costs;
•     Formalization and replicability of planning skills;
•     Analysis of cost structure and support to distribution scenario definition;
•     Geomarketing and price policies for new customers and customer areas;
•     Balancing of working load among different drivers;
•     Balancing of working load among different depots;
•     Company vehicle fleet optimization.
Of course one of the main focus for the main partner (SOGEA) is also to become an application service provider; planning functionalities could thus be also offered by receiving data, taking care of GRID computing, and returning the results in dynamic pages for customers. This approach will be driven also by (hopefully) successful results during the exploitation task and after a mid term evaluation of the so-called early business plan. At this preliminary stage a prototype application is expected but a massive validation on methodology and technology platform (i.e. middleware) is mandatory to match the experiment quality criteria and boost further developments and alliances among partners.