The talk focuses on two main areas of application for LP in the oil industry: mixed integer LP to evaluate tenders for additives used in the manufacture of lubricants and multi-period models used for Refinery Planning.
For the lubricants additives tenders, formulating the mathematical model was the easy part, while the difficult part concerned data management issues, and the softer skills needed to tease out unstated assumptions by suppliers about unfair use of the quantity discounts offered and the art of structuring the call for tenders so as to keep suppliers happy by setting breakpoints for prices at the right points in the volume ranges. For a small OR department (6 analysts), with a diverse project portfolio and high turnover of staff moving to other jobs within the company, a key management issue was how to sustain the requisite combination of technical expertise, familiarity with the software, and knowledge of the internal customer's business.
The discussion of Refinery Planning LPs focuses on the integrated management of issues pertaining to the regular (monthly) use of a multi-period model to update a rolling three monthly plan. Why use a multi-period model instead of a single period model for the first month of the three (the part of the model whose solution actually becomes a firm plan)? Should end of period closing stocks be fixed or flexible - or is there some way of getting the best of both worlds? How can the model draw on process control history to ensure that it uses the most accurate and current data - and how can this be done in a user-friendly way? How should the non-linear behaviour of processing units be modelled - and if this is done via modes with linear yields (the standard industry approach), what is the scope for non-linear interpolation between modes - and can it be done in a user-friendly way? How can a planning model, which takes no account of time sequencing within months, be linked to a scheduling system with time sequencing within individual days?
The talk concludes with some observations on communications between refinery planners and four very different groups of users: manufacturing management, (crude and product) traders, marketers and politicians.
Current 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996
