Apr-2010

Crude oil selection: optimisation by weight or by volume?

The LP model should optimise crude and products in their trading units for accurate assessment of crude oil worth and maximisation of gains from product blending

M D Pawde and Sachin Singh
Hindustan Petroleum Corporation Ltd (HPCL)

Article Summary

Linear programming (LP) is a technique used widely for optimisation in petroleum refineries. LP models of refineries are used for capital investment decisions, the evaluation of term contracts for crude oil, spot crude oil purchases, production planning and scheduling, and supply chain optimisation. Robustness and precision in the LP model are critical to the profitability of the refinery. A good refinery LP model accurately captures unit operation yields and properties, stream blending, the extent of constraints on product specifications, flexibility on cargo sizes of crudes and finished products, and other relevant data.

The methodology of crude oil evaluation using an LP model is as critical as a good LP model itself. Each refiner has its unique requirements of LP that depend upon the environment in which the refinery operates and the market to which it caters. The methodology of evaluation should address, among other concerns, the time horizon of optimisation, unit of measurement of crude and product quantities and prices, the option to import finished products, and inventory pricing and accumulation/depletion. Deve-loping an accurate assessment methodology, which closely reflects the realities faced by a particular refinery, is an interesting process.

This article is based on the authors’ experiences in setting up a LP-based crude oil evaluation system for Hindustan Petroleum Corporation (HPCL). It discusses how units of measurement of crude and product quantities and prices (the choice between weight-based and volume-based optimisation) can give conflicting recommendations. The paper also describes options available in Aspen Process Industry Modelling System (PIMS) to specify correctly the desired methodology of optimisation.

Optimisation of crude oil trades
The world over, crude oil is traded by volume; it is priced in dollars per barrel. Volume-based LP models, which capture crude oil yields in volume per cent (vol%) and buy/sell crudes by volume, assess them accurately. However, weight-based models need to incorporate proper conversion and comparison methods for the correct evaluation of crude oils.

Let us consider two hypothetical crude oils, A and B, one heavy (API 35) and the other light (API 45), but with the same weight per cent (wt%) and vol% yields. Yields in weight and volume will be the same if the individual ratios of stream density to crude density are identical for both the crudes (see Table 1).

Product prices and standard densities used for sample calculations are shown in Table 2. Conversion factors used for naphtha, kerosene, gas oil and resid are from Platts. For liquified petroleum gas (LPG) and vacuum gas oil (VGO), typically produced densities have been used.

Using $/MT prices from Table 2, and wt% yields, product value in $/MT is determined and divided by a bbl/MT conversion factor for the respective crudes to arrive at their $/bbl value. It can be concluded from Table 3a that both crudes have the same value in $/MT; the conclusion is also apparent from the fact that both crudes have the same wt% yields. One hundred tonnes of either of the crudes will provide the same revenue to the refinery; their break-even prices per barrel are respectively $62.3 and $58.7. In other words, the refiner can pay 3.6 $/bbl more for crude oil A than for crude oil B to make the same profit. Let us call this Conclusion 1.

Product values are now recalculated using vol% yields and $/bbl prices from Table 2. Both crudes have the same vol% yields and, hence, are expected to have the same value per barrel.

The calculation in Table 3b shows that the refiner cannot pay any more per barrel for crude oil A than they are paying for crude oil B. Let us call this Conclusion 2.

Which of these conclusions is correct? Conclusion 1 is the recommendation of optimisation by weight, whereas Conclusion 2 is the result of volume optimisation. Optimisation by weight ascribes higher values to heavier barrels than to lighter barrels of crude oils (A and B respectively). On the other hand, optimisation by volume values heavier and lighter crudes at par.

A mixed weight-volume optimisation model buys, sells, transfers and stores crudes and products in their actual trading units. Such a model gives due credit to heavier and lighter products, and is an accurate way of evaluating crude oils.

Let us rework the gross product value for mixed yields, starting first from weight and then from volume yields. Let us assume that the refiner sells gas oil, kerosene and VGO by volume and the rest of the products by weight. Starting with weight yields, mixed yields are calculated using actual product densities of products sold by volume. Starting from volume yields, the mixed yields are calculated using actual densities of products sold by weight. $/bbl prices for gas oil, kerosene and VGO, and $/MT for LPG, naphtha and vacuum resid are derived from Table 2.

 It is evident from Tables 4a and 4b that both methods draw the same conclusion: crude oil A - crude oil B = 0.9 $/bbl. That is, the refiner can shell out 0.9 $/bbl more for crude oil A than they can for crude oil B to make the same margin. Table 5 summarises the results.

The value of crude oil A decreases when it is priced by volume compared with when it is priced by weight. On the other hand, the value of crude oil B increases when it is priced by volume compared with when it is priced by weight. Value in mixed units lies between weight and volume values for crude oil A, but is higher than either of the two for crude oil B. These observations are explained below.
Notations used for crude properties are: