JUNE 2008 — Barcelona (SPAIN)

At the 2008 International Logistics Fair (SIL) held in Barcelona from 3-6 June, RENFE Mercancías presented the algorithm developed by GOAL SYSTEMS for optimising spaces on the train decks, included as part of the system for selling containers that RENFE Mercancías will place at the disposal of its customers from 1 July 2008.

At the 2008 International Logistics Fair (SIL) held in Barcelona from 3-6 June, RENFE Mercancías presented the algorithm developed by GOAL SYSTEMS for optimising spaces on the train decks, included as part of the system for selling containers that RENFE Mercancías will place at the disposal of its customers from 1 July 2008.

The object of the ALGORITHM is to maximise the spaces offered on Multiclient Intermodal Transport, through the optimum location of containers on the train decks, intervening in the entire process from the booking made by the client through to load distribution on the train at the terminal.

At RENFE Mercancías a very high volume of data is handled:

— 16 CPUs calculating at the same time.
— 30-day log file of operations.
— More than 1,100,000 spaces put on sale, distributed among more than 6,000 retail offers.
— More than 6,000 daily calculations.
— Built into the OMNIUM platform of RENFE Mercancías.

The algorithm has been designed with some technical features that respond to this high volume as well as the critical nature of the system for RENFE Mercancías:
— Role-based configurable security that can be integrated with any security system (MS-AD, LDAP,
WebServices, etc.).
— System that can be totally parametrised and configurable.
— Integration with other systems through WS/*, SOAP, HTML and XML standards.
— Tolerance against hardware and software faults.
— High scalability.
— Ease of deployment and solution maintenance.

The optimisation system of GOAL SYSTEMS receives a series of input data from the algorithm: Configuration of the retail offer, types of freight wagons, UTIs and customers’ bookings, and has a user interface where the company can:

— Enter the structural data required for an optimisation problem (supply, wagons, bookings, etc.).
— Configure the objective function to be optimised.
— Carry out booking and cargo simulations with different business objectives, validating the solutions
obtained.
— Configure the end objective function to be used within the production environment.
— Parametrise system sections and data logs to be saved in production.
The business rules are parametrised in the algorithm by the central user in charge of the business, and respond to different questions that are put forward such as:

Can UTIs be moved at terminals?
— Whether or not to allow movement of UTIs within the same wagon, at terminals.
— Whether or not to allow movement of UTIs between wagons of a single train, at terminals.

What combinations of UTIs can be loaded onto a wagon?
— Maximum load per wagon and per line category.
— Maximum load per train and per line category.
— Ratio between axle weights.
— Ratio between bogie weights.
— Linear load limit per line category.
— Maximum load per axle, per stretch and per line category.

Can I load a UTI on the train?
— Asymmetric rules.
— Semitrailer rules.
— Hazardous goods rules.
— Fastening rules.
— Stacking rules.

Always ensuring the legality of the train’s cargo at all times, both in the booking made by the client as well as loading at the terminal.

The business objectives are weighted through parametrisable costs that have been included in the algorithm:
— Cost of leaving wagons empty. Targeted at improving the operation and maximising last-minute sales.
— Cost of leaving wagons empty at the end of each batch. Targeted at improving the operation and minimising
manoeuvres.
— Cost through equal distribution of the load on each wagon. Targeted at improving the operation and
minimising wagon maintenance.
— Cost for minimising manoeuvres. Targeted at improving the operation and minimising the number of
manoeuvres.
— Cost through separating incompatible hazardous goods. Targeted at maximising transport safety.

The parametrisation of costs enables the definition of the objective function, whose mission is to classify solutions based on business defined criteria. This cost function is a linear combination of all costs, whose weighting and prioritising values are defined by the user, enabling clear capacity for simulation and help in taking decisions.

Optimisation is achieved by analysing all of the possible combinations that satisfy the restrictions established and by minimising the defined objective function. Solutions are obtained and are classified in accordance with economic criteria and solution quality, as penalty criteria can be entered against the deviations over desired values.

Goal Systems