(340f) Optimization of the Utilities Production in TOTAL Normandy Refinery Using Ariane Software

Refining has significantly improved its energy performance in the last 20 years. However, refineries are still working on improving the efficiency of refining processes, to be more competitive. In this context, Total, one of the largest integrated international oil and gas companies, is making efforts to improve its refineries’ energy efficiency. The Normandy refinery, a world class facility with crude distillation capacity of 200,000 barrels/day representing over 10% of Total's refining capacity in Europe, has been selected to implement a utilities management and power plant optimization system.

Optimization of the utilities production in Total's Normandy Refinery using Ariane software will be presented. Ariane is a decision support tool to operate the utility system at a minimum operating cost while still satisfying all of the refinery’s needs and constraints. This modeling and non-linear optimization tool is dedicated to power plants. This paper discusses how the various tools of the Arianne software were implemented and are now used by the operating staff at the Normandy Refinery for:

-        Accounting of utilities production/consumption

-        On-line optimization

-        Prediction of power plant operation

-        Preparation of reports.

Although cost savings of such projects are difficult to quantify with precision (process and/or external changes occur during the project, no initial point (base line) clearly defined, etc), the main results will be presented:

-        a cost reduction of at least 2.5% of the steam and electricity invoice of the refinery,

-        over 1.2 EEI (Energy Intensity Index) performance improvement,

-        since its implementation, the system implemented earned the confidence of the site operators by making efficient and realistic recommendations.

Real time optimization of a full site’s utility system is always complex.  The main steps of this successful on-line implementation project will be presented.

The numerical modeling and optimization issues encountered during this project together with lessons learned will also be detailed.