Combination approach improves investment decisions for capital projects

Dec. 8, 2003
When evaluating project economics, plant owners that use a combination of three methods will improve understanding of the capital project and make better decisions whether to invest in a given project. The three methods include benchmarking, Monte Carlo risk analysis, and a historical return on investment correlation.

When evaluating project economics, plant owners that use a combination of three methods will improve understanding of the capital project and make better decisions whether to invest in a given project. The three methods include benchmarking, Monte Carlo risk analysis, and a historical return on investment correlation.

Benchmarking is a tool that ensures that overall refinery performance does not deteriorate; it can also lead to engineering and construction (E&C) contractor performance measures. Monte Carlo risk analysis allows one to view the cumulative impact of risk and eliminates the need to make perfect forecasts. Analyses using historical return on investment (ROI) correlations show the impact of past performance on a new project.

All refiners face the critical decisions of capital deployment. The recent surge in capital spending to meet low-sulfur gasoline and diesel specifications has many refiners facing key capital decisions.

It is important, therefore, that refiners use all the data and analysis tools available when making these critical decisions. A capital effectiveness review combines three traditional methods and leads to a better understanding of project economics and better decisions.

Why projects fail

Promising projects fail to meet financial expectations for many reasons:

  • The strategic concept that many decision makers use may be flawed as it was in the 1970s when the US petroleum industry attempted to develop large-scale oil shale projects based on unrealistic crude price forecasts.
  • An analysis may fail to discover the meaning of a key variable, such as the rush by US Gulf Coast refiners to build cokers in the 1980s. Most forecasts did not account for the impact of a UK coalminer's strike or the possibility of building too many cokers at one time.
  • A project's execution is often less than perfect. Many projects exceed budgets. Other times, it takes several attempts to get a new piece of equipment running consistently.
Click here to enlarge image

Table 1 lists some of the more common reasons that capital projects fail to meet expectations and groups them into the categories of concept, analysis, and execution.

In addition to the items in Table 1, refiners must consider market changes. The downstream petroleum industry tends to move with a "herd mentality" when the market exposes a profit opportunity. One must consider this fact when examining capital projects.

Review

A capital effectiveness review creates three different views of the same project using:

  • Pro forma design benchmarks.
  • Cash flow risk analysis.
  • Historical returns.

Examining a project in different ways increases one's knowledge of the pros and cons of that project. The result is fewer unpleasant surprises.

To illustrate the methods used in an effectiveness review, we will use a simple example of a refinery that must add a diesel hydrodesulfurization (HDS) unit to its existing processing scheme to meet ultralow-sulfur diesel specifications

Click here to enlarge image

Table 2 summarizes the example project key variables and economics. Many of the parameters are held constant to simplify the example. The example project has a simple 28% ROI.

Pro forma design benchmarking

Even the best design and engineering companies can lose their way during the project if clear, measurable objectives are not established.

Often, the E&C company performs a project to meet three objectives: on time, under budget, and done safely. Objectives such as operating cost reduction, energy efficiency, yield optimization, and integration into the existing plant are not considered further down the E&C priority list.

The way to ensure that all objectives are met is to benchmark the pro forma design vs. similar operations already in service. During the benchmarking exercise, one must develop performance measures and include these measures in the engineering and constructing contract so that performance of the refinery with the new diesel HDS does not deteriorate.

For example, one real-world refiner had completed the technology selection and design phases of a similar project and was well into the engineering stage. The refiner included a performance measure in the E&C contract such that the energy efficiency of the entire refinery was to remain constant or improve, as measured by the Solomon energy intensity index (EII; see box for definition).

When the E&C contractor was ready to start field construction, the refiner finally focused on the impact on energy performance. A pro forma design benchmarking analysis showed that the refinery's EII had deteriorated significantly with the addition of the new process unit. The pro forma analysis exposed three problems:

  • The technology was not the most energy-efficient design. The refiner chose better yields at the expense of energy efficiency. Whereas both the refinery and E&C teams understood this trade-off, senior refinery management was surprised.
  • The integration of the new process unit into the existing refinery steam system was not optimized, which resulted in excess low-pressure steam.
  • After close investigation, the E&C contractor realized that its internal team objectives had not emphasized energy efficiency. The E&C team followed objectives similar to the "on time, under budget, safe" priorities.

The design and engineering process has many opportunities to optimize energy use, operating costs, capital costs, and yields. Engineers are left largely on their own to make each of these decisions; therefore, clear and balanced objectives are needed.

The pro forma design benchmarking had a successful outcome. The refiner and E&C company negotiated an adjustment to the EII target in the E&C contract to account for the technology selected.

The E&C firm modified the integration of the new process unit into the existing refinery by looking beyond the new unit boundary limit. Each E&C engineer had to identify energy saving options and the cost of each option so that the refinery could make the right economic decision.

The result was a new pro forma design EII lower than the contract target and an improved understanding of the project's impact on refinery energy efficiency.

Cash flow risk analysis

The power of desktop computers and the introduction of easy-to-use software make risk analysis practical for most organizations. The most common assessment technique used to analyze risk, beyond a simple sensitivity analysis, is the Monte Carlo simulation.

Monte Carlo simulation is a stochastic technique based on the use of random numbers and probability statistics to identify potential problems. Applying Monte Carlo simulations to risk allows one to examine more complex projects.

Monte Carlo simulations sample the forces affecting the ROI in a number of random iterations and measure the risk of the project as a whole.

Monte Carlo simulations allow one to substitute a distribution representing the variability in a specific variable instead of making a "single-point" assumption. Monte Carlo is a better representation of the process variability that happens in real life.

Instead of a perfect forecast, one can directly simulate real-world variability of process variables. Monte Carlo simulation also allows one to analyze the cumulative effect of the variability in several variables. The end result is a distribution showing the probability that an ROI estimate will result.

The Monte Carlo risk analysis model unfolds in several steps:

  • Develop a cash flow spreadsheet to model project key drivers and ROI.
  • For key ROI drivers, develop probability distributions to represent the variability of this variable.
  • Add the Monte Carlo simulation software to your spreadsheet model. The software lies on top of the spreadsheet.
  • Run the simulation and review the distribution of possible ROIs.

The Monte Carlo simulation randomly samples each of the distributions for each key variable and calculates the ROI for that set of variables. It continues to select data randomly from each of the distributions and calculates new ROIs.

This iterative process continues for a sufficient number of passes (usually 100 iterations). The result is a distribution of possible ROI for the project.

Click here to enlarge image

For the diesel HDS example, we focused on four key ROI drivers:

  • Low sulfur-high sulfur diesel price differential.
  • Cash operating cost for the refinery with a new diesel HDS.
  • Capital cost for the new diesel HDS.
  • Refinery utilization stream factor.

Table 3 shows the simple, single-period cash flow spreadsheet model for the example diesel HDS project. Other models will probably be a 5-10 year discounted cash flow with much more detail.

Click here to enlarge image

Table 3 shows the four key variables in red. For each one, a probability distribution was developed from industry data (Table 4).

Click here to enlarge image

Figs. 1-4 show each of the four distributions.

Click here to enlarge image

Fig. 5 shows the resulting ROI distribution. It shows that there is a 63% chance that the actual project ROI will be less than the 28% calculated initially.

Click here to enlarge image

Effective use of this graph for discussing a project or a list of potential projects requires an internal process to focus on ROI at two or more probability points. A typical analysis process, for example, would examine projects at the 0.2 and 0.8 probability points.

Click here to enlarge image

The diesel HDS project would be a "12-40 project," meaning the project has an 80% chance of exceeding a 12% ROI and a 20% chance of exceeding a 40% ROI.

Historical ROI regression

Actual ROI is usually lower than expected. There are many reasons this is typical in the refining industry.

Click here to enlarge image

One reason is the generally optimistic nature of people. Engineers tend to be overly optimistic when evaluating projects.

Click here to enlarge image

A second reason is the previously mentioned herd mentality. If a new type of process unit looks profitable to one refiner, it probably looks profitable to others. The industry builds the same process units and forces the market to change, erasing the original opportunity.

To help view capital projects correctly, we developed an ROI predictor based on a historical regression of ROI components from industry data.

A specific refiner can develop a predictor tailored to past performance of its company by analyzing the company's data.

Click here to enlarge image

In developing a regression of ROI components, we first had to modify the ROI equation into a form that is readily analyzed. Equation 1 (see equation box) is a standard form of the ROI calculation. After several attempts at combinations of terms, Equation 2 showed the best regression coefficient (R2) with the industry data we had available.

Equation 2 shows a modified form of the ROI equation, when correlated against our data, yielded an R2 > 0.95.

This means that 95% of the variation in ROI is due to the variation in gross margin, cash operating cost, and capital intensity.

The historical ROI regression analysis has some shortcomings. First, the ROI prediction is for the entire refinery and not for a specific project being considered.

So one must calculate the historical ROI before and after the addition of the new capital project and back-calculate the ROI of the specific project (Equation 3).

Second, the regression uses refinery replacement values that we calculated. While this provides a consistent estimate for refinery replacement values across all the refineries in the database, it does not match the actual capital used by any specific refinery.

Third, the rule of "big numbers" must be considered. The back-calculation shown in Equation 3 involves dividing a big number (refinery replacement value) with a relatively small number (incremental investment for our specific project).

Any time this happens, the level of precision comes into question.

Even with these drawbacks, historical ROI is still useful. It encompasses all the reasons projects do not live up to expectations. The historical regression includes the data every time:

  • A project runs over budget or starts up behind schedule.
  • The market moves to close off an opportunity.
  • A new process unit has reliability problems.
  • The regulations change the rules of the game.
  • Another refinery bottleneck blocks the full potential of the new project.
Click here to enlarge image

Using the regression factors in our example diesel HDS project gives a historical ROI of 19% (Table 5). This more closely reflects the ROI one should expect if history "repeats itself."

Capital intensity measures the "velocity" of moving barrels through the capital invested in a refinery. Not all refineries are the same—some have a large capacity and low complexity whereas others are smaller and more complex.

The modified ROI regression equation shows that increasing capital intensity (velocity) only improves ROI when the gross margin is positive.

Pushing barrels through a refinery only to produce low value products, therefore, does not improve refinery ROI.

This may seem obvious but many refiners try to expand their way into first quartile performance by simply focusing on volume.

Click here to enlarge image

The author
Michael J. Hileman is a vice-president with Solomon Associates, a division of HSB Solomon LLC, Hartford, Conn. Before joining Solomon, Hileman was vice-president of supply, distribution, and sales for a large independent refiner-marketer. He holds a BS in chemical engineering from Rose-Hulman Institute of Technology, Terre Haute, Ind.

Based on a presentation to the NPRA Clean Fuels Challenge, Aug. 12-13, 2003, Houston.