Refinery planning : a mature technology, yes but…

Reinventing refinery planning


Refinery planning has been among the first applications of mathematical programming in the industry. Linear programming has been successfully applied to both long term and short term refinery planning.

Fundamentally, linear programming is about calculating a set of activities under the condition that no activity should be used if a more profitable activity or combination of activities was available. In addition, the method implies values to be placed on scarce resources (marginal costs) which define the opportunities open to the company.

The practice of linear programming by oil refiners has evolved over time in two, somewhat competing directions

  • Running a large number of cases in order to quickly evaluate various crude oils available in the market from a trader’s perspective
  • A short term perspective on refinery planning viewed as a preparation step to refinery scheduling: this view on refinery planning requires a more accurate representation of both the logistics and stream movements (hence a generalized use of pooling) and of the process units performance (hence the use of non-linear process models)

Another important factor which has impacted the way linear programming models were used in recent years has been the arrival of a younger generation of users subject to a higher turnover. In the past, a certain level of expertise in linear programming model building and in-depth interpretation of results was achieved after several years of practice. By contrast, today’s users are more interested in a quick overview of a large number of solutions.


The new generation refinery planning systems must therefore address the following requirements

Easy to use

Provide simple and easy to learn modeling environment, data structures and user interface


Enable the user to quickly build alternative scenarios through very simple manipulations


Provide summary solutions which can be quickly grasped and compared among each other


Ensure data consistency between refinery planning and refinery scheduling
Request an online demo

Contact us

to know more about our refinery planning solutions

Princeps’ refinery planning solution: 


Princeps’ refinery planning system offers a flexible and powerful modeling environment, for single or multi-period, single or multi-refinery planning, with pooling and distributive recursion, and non-linear process modeling (simulator interface) capabilities… A comprehensive distribution network optimization capability is part of system as well.

In addition to these capabilities, below are some of the distinctive characteristics of Princeps’ refinery planning system:

  • Multi user achitecture

    models, scenarios and solutions are stored in the central database, and can be shared based on authorisations and roles

  • Scenarios and alternates

    Scenarios are organized in a tree structure, which can be either generated automatically or (partially) built manually

    Alternates are a convenient way of building new scenarios. An alternate is a set of modifications with a name, which can be applied to one or several scenarios in order to generate new scenarios

  • Parametrisation and sensitivity analysis

    which considerably simplifies alternative scenarios definition, execution and reporting. With its novel capabilities allowing virtually limitless combinations of parametric loops, the generation of hundreds of scenarios is just a few clicks away.

  • Planning/scheduling models and scenarios

    Simplified and unified model building and maintenance: Process models are built once and shared by both planning and scheduling.

    Planning directives such as operating parameters or blending recipes can be directly retrieved and used in scheduling

  • Solution database

    All planning runs results are archived in the output database, and can be retrieved and shared among planners

    Individual or multicase reports can be built based on a selection of archived cases

    Fully configurable reports

  • Other advanced features

    Full versioning and archiving of models and cases

    Parallel processing: Multiple scenarios can be solved in parallel using multi-core CPU