The procurement of refinery projects is a high risk venture. Determining how to finance a refinery and manage typical risks in order to get sound economic returns is a major challenge. There are significant risks exposed in refinery business environment, for instance construction risk, demand risk, operation risk and especially price risk on both demand and supply sides. Availability and characteristics of types of crude oil supply and product derivatives can determine the choice of refinery types. Apart from buying crude oil in the spot market and selling its products on a similar basis it is necessary to create significant price certainty to ensure a robust cash flow is achieved. The supply contract and off-take contract can be used to create sufficient certainty of price, quantity and availability of both crude oil and sales of refined products, and thus ensure the financial viability of a refinery project. A mechanism for assessing the risks associated with procuring a refinery is presented and an evaluation of the economic parameters modelled in Visual Basic, Crystal Ball and Excel spreadsheet is illustrated.

Project finance requires that the risks identified during the project life cycle are mitigated before sanction of a refinery and sufficient revenues can be generated to service the debt and make an acceptable profit (Merna and Njiru 2002). Typically the financial instruments used in a project financing are debt, mezzanine (bonds) and equity. The higher-risk projects should normally take more equity to protect the interests of lenders and bond investors and lower-risk projects can accommodate more debt (Merna and Khu 2003).

A major risk in refinery operation is associated with the characteristics and quantities of the crude oil supply which can significantly influence refining margins. Refining low American Petroleum Institute (API) gravity crude oils requires more complex and expensive processing equipment, more processing stages and more energy, therefore costing more. The price difference between high-gravity and low-gravity crude oils reflects the refining cost difference. Investment in facilities to process heavier crude oils allows refiners to improve their profits by reducing the cost of their crude oils supply.

Each type of crude oil will produce different percentages of refined product. Buying cheaper heavy crude oil, for example, will have a high conversation cost to light products compared to buying expensive light crude oil which is cheaper to refine. Mixing a percentage of heavy crude with light crude oil is often used to refine at a lower cost. Therefore, the price difference between light and heavy crude oils and light and heavy products is among the most important variables affecting refinery margins. These differentials are incentives for installing expensive processing facilities in a refinery, including fluid catalytic cracking, hydrocracking, coking and other residual conversion facilities.

Modern petroleum refineries are designed to process a variety of indigenous and imported crude oils. Selecting supply contracts is crucial for companies as major costs are involved in purchasing raw materials (Bansal 2006). As the crude oil cost is about 90% of the refinery input cost, the selection of an optimum crude mix is extremely important to achieve higher margins. However, the number of options for buying crude oils under fluctuating prices and transporting them to refineries are huge, thus making evaluation of the crude oil mix extremely difficult.

Refineries normally purchase crude oil and sell its products on term contracts from forward and future markets and by spot purchases from the spot market. If, for example, a refinery depends on the spot market for supply, then its profit margin could be seriously affected by movements in market prices. Apart from buying crude oil in the spot market and selling its products on a similar basis it is necessary to create significant price certainty to ensure a robust cash flow is achieved. Using a project finance strategy, the refiner would be required to enter into supply contracts to reduce spot market risk. A typical supply contract and off-take contract is arranged in a petroleum refinery project procured through project finance, as Figure 15.2 illustrates.

Long-term supply and off-take contracts (forward contracts) can be employed in the bundling of supply contracts to determine the cost and price structure of the off-take contracts as illustrated in Figure 15.3. The principal aim of the supply and off-take contract is to create sufficient certainty of price, quantity and availability of both crude oil and sales of refined products, and thus ensure the financial viability of a refinery project.

Therefore, it is the supply contracts and off-take contracts for refined hydrocarbons that provide the guarantee on which a ‘project finance’ transaction is based (Elsey and Hurst 1996).

The refinery can refine five crude oils from suppliers located near to the refinery. In this mechanism Crystal Ball is employed to assess the crude oil history data which can be obtained from the EIA database against probability distribution by using one of several standard goodness-of-fit tests. The distribution with the highest ranking fit is chosen to represent crude oil data. Figure 15.4 shows that lognormal distribution fits Iran H crude spot market price. However, if a crude oil supply is purchased on a long-term basis and its products sold on a contract led basis, the crude oil price and refined product price are bounded. Thus the triangular distribution is assigned to this supply-off-take agreement. This is illustrated in Figure 15.5

The risks identified have direct impact on the cost of each activity in the model, for instance the change in construction would increase or decrease the distillation plant cost between 99.6% and 103% respectively as shown in Figure 15.6. The deterministic cost of each activity is calculated based on those ranges. However, the economic parameters with deterministic values do not reflect uncertainties in the refinery industry. Probabilistic analysis by means of Monte Carlo simulation can deal with this problem. Thus, both range and distribution can be assigned to those variables.

The same principle can be used in other variables such as refining cost and refining margins. A triangular distribution is commonly used in the model where variable distributions are not well known but can be bounded, such as construction cost, transport costs, power and operating costs.

Computing refining margin varies from refinery to refinery. To simplify the computation gross product worth (GPW), crude oil prices and their GPW are imported directly from the EIA database into the model.

The bundle of crude oil supply contracts and respective off-take contracts can be determined by the analyst. Figure 15.7 illustrates examples of decision variables and corresponding constraints for lower and upper bounds.

Figure 15.9 illustrates the results of a sensitivity analysis. Curves with steep slopes, positive or negative, indicate that those variables have a large effect on the project’s financial viability, whilst curves that are almost horizontal have little or no effect on the project’s financial viability. Although the Daqing supply contracts and its off-take products contracts are in place it was found that the project is still very sensitive to the crude price risk, supply default risk, supply delay risk, construction risk and GPW risk and less sensitive to changes in refining the Daqing crude oil and design risk.

Table 15.2 and Figure 15.10 indicate the economic parameters and cumulative cash flow when assessed on a stochastic basis. The IRR for the base case is 16.7% and for the best case 26%; however, for the worst case it is only 4%. Clearly the project is risky with wide variation between the worst and best case cash flow in the operation period as illustrated in Figure 15.10, for a single Daqing crude oil supply.

A similar stochastic simulation process is also applied to four other possible crude oil supplies. Table 15.3 shows the economic parameters for each single supply. Xinjiang crude, for example, has competitive advantages such as low purchase price and low transport cost because of its location and availability to the refinery, resulting in less supply risk and price risk than other crude supply contracts.

Sensitivity analysis results show that most single crude oil supplies are sensitive to changes in supply, crude price, demand and GPW. The results of probability analyses of other single crude oil supplies are shown in Table 15.4. Clearly there is greater financial uncertainty accompanying the project if the refinery takes a single crude oil supply. Thus, apart from the single Xinjiang supply, the other crude oil supply would be unattractive to investors when such risks were taken into account.

The same testing process is employed to test two types of crude oil supplies (Test 2), three crude oil supplies (Test 3), four crude oil suppliers (Test 4) and five crude oil suppliers (Test 5). The decision in these tests is to determine the percentage of each crude oil the refinery should take to maximise the IRR and return whilst maintaining an acceptable level of risk. The constraint in those tests limits the total crude oil procured per day at no more than the refinery capacity of 220 000 b/d. Investors expect the maximum mean IRR for the minimum risk. Thus, the objective of a bundle of crude oil supplies is set to maximise the mean IRR with a standard deviation between 0.030 and 0.039.

In Test 4 when the fourth crude supply (Iran H) was then added to the bundle and tested in the model the bundle became more attractive than other solutions. This is because the risks associated with the fourth crude oil supply balanced the total risks on both supply and off-take sides and thus overall supply risk.

Efficient frontier analysis is then employed to consider the balance between return and risk in selecting the optimal crude supply contract bundle based on the risks identified.

Table 15.5 illustrates mean return and standard deviation for combinations of crude supply contracts of a bundle of the five crude oil supplies tested. Under certain risk levels different bundles will generate different returns.

Figure 15.11 shows the efficient frontier for the bundle of five crude oil supplies. The efficient frontier is the intersection of the set of bundles with minimum variance (risk) and the set of bundles providing the maximum return. For example, a bundle of 59% Daqing, 0% Iran L, 0% Saudi L, 7.0% Xinjiang 34% Iran H crude oils is more efficient than the bundle of 61% Daqing/39% Iran H in Test 2 because it has a higher IRR and NPV although both of them are exposed to a similar risk level.

The analyses show that the project is exposed to different levels of risks and different economic returns. After risk management, Solution 5 still has the highest return with highest risk; Solution 1 has lowest economic return but lowest risk. Investors choosing Solution 1 would seek a large amount of debt; whereas investors choosing Solution 5 wouldrequire more equity as risk finance.

The assessment clearly illustrates the bundles’ best, worst and base case economic parameters with impact of both supply risk and typical refinery risks.

The assessment offers a detailed method for determining the crude oils to be purchased and their percentage within a bundle of crude oil supply contracts.

The assessment can aid stakeholders in the decision-making process regarding the type and quantity of crude oil supply contracts based on identified risks.

Investors in refinery projects can assess specific risks affecting crude oil supply in relation to the overall project economic parameters.

There are numerous combinations of crude oil supply bundles. The risks associated with supply and off-take are extremely complex. From the tests shown in the analyses the refinery economic viability is very sensitive to the crude oil supply and off-take. The choice of a bundle of crude oil supplies is paramount to the commercial viability of a refinery thus making risk management an integral part of refinery procurement and operation.

The authors wish to thank Dr Anthony Merna and Mr Yang Chu for allowing them to use this amended version of their paper.