When you are setting up the case data for your monthly plan, what price should you use for the crude or other materials that you have already bought?   This interesting question came up at the December session of the Oxford Princeton Refining and Trading course last December, when I came in to help out with a demo of a GRTMPS.   Quite a few people thought the answer was obvious, but they did not all come up with the same answer.
Replacement Cost Zero Cost
When you use crude you need to buy more to keep the refinery running next month.  This might be more or less than you paid for the current inventory depending on whether the market price is rising or falling. Money already spent is a sunk cost – you can’t optimize it.  You should just let the model use it as profitably as possible to meet your product demands.
I can see merit in both views, but also issues for the quality of the plan generated.
If you put zero prices on the existing feed stocks they will all have the same cost, regardless of their characteristics.   The optimization will make the products as cheaply as possible, using up good quality and leaving poor quality feeds in the tanks.   Next month, you will might make less money than you could have if you had more of the better material, as you need to blend this off.   On the other hand, replacement costs could also distort the solution.  If prices are rising, then the crude might appear too expensive and not be used at all.  You would not be getting any benefit from what you have already spent.  You have working capital and storage space tied up reducing your available resources and flexibility going forward.

An underlying cause for these issues is that planning cases are often set up with no view of the future.  The world just simply ends when the model duration does; so there is no influence from any longer term consequences.   It might be enough to put in some simple rules based on your experience of refinery operations – or from a longer term overview such as annual budget cases.  These rules would be about the maximum amount of crude of various types to be carried forward, or the minimum that has to be used each month to make space for the next cargo.   To leave the decision more within the realm of the optimization rather than assumption, however, it would be useful to do multi-period cases, extending the time horizon of the model with at least a general description of the situation expected in the next month(s) and inventory to pass crudes into the future.   That would allow the quality needs of the future to influence current usage to make a better solution overall.   However, multiple periods do mean larger models with more data and more complicated output -  and that clearly has some disadvantages itself.   As a compromise, you could continue with traditional one period modelling, but use inventory, rather than simple purchases, to model the existing crude stocks, so you can put some value and constraints on what is left.   For example, you could constrain the average sulphur or API of the crude left to be within certain ranges.

Adding some information about logistical constraints and scheduling requirements to the case might also help drive plans in more sensible directions.    For example, if ullage is needed in order to receive the next crude delivery, then there are clearly costs associated with not using the crude you already have    You might end up paying demurrage while a ship waits to off-load.     If you are modelling the crude stocks via inventory, you can build this possibility into a holding cost.  I am inclined to favour the zero cost for near term planning cases.  If I am at the refinery, ready to process crudes and needing to produce products, then I have to work with what is there and use it as efficiently as possible.  If there are considerations related to next month’s operations, then I would expect that these should have been determined earlier on, when the crude was being evaluated for purchase, and its arrival scheduled.  These might translate into constraints forcing the use or storage of certain materials.  A little inventory modelling with some quality constraints could be used to provide some flexibility.

I am, therefore, expecting, that the evaluation cases run before the crudes were bought did have a long enough view to such guidelines on what should be used in the current period or stored.  The duration of such cases would need to be long enough to cover the whole time period in which a crude is likely to be used and then its replacement by the next cargo.  This is quite possibly more than a month, and may well require multi-period modelling.   These runs would have some opening stocks in them as well – and here maybe the cost is relevant – a way of comparing using the existing materials to buying new ones.   It is also important to note that replacement does not have to mean like for like, the same crude again, or even one crude for another.  Replacement might mean a mixture that provides a similar quality.

Of course, not all refineries operate the same way.  Perhaps things look different if your crudes arrive by pipeline, instead of by ship.  Maybe you have very flexible feed arrangements.  Maybe you have very limited tankage.  Maybe your crude selection and operational planning are more tightly integrated.   Are you a “zero-coster” or a  “replacement-pricer”?
 From Kathy's Desk 12th March 2017.

Comments and suggestions gratefully received via the usual e-mail addresses or here.
You can also use this form to ask to be added to the mailing list to be notified when new articles are published.