Risk management

Understanding Uncertainty and Risk in Capital Projects

Speakers at the recent SPE Asia Pacific Unconventional Resources Conference and Exhibition addressed the role of uncertainty and risk in sanctioning megaprojects.

jpt-2016-02-riskfig1.jpg
A wellsite generator from one of Arrow Energy’s coalbed methane development projects in the Surat Basin in Queensland. The high cost of conventional logging methods have forced the company to consider less expensive options that may limit the project’s economic value or increase risks in the operation. 
Photo courtesy of Arrow Energy.

The SPE Asia Pacific Unconventional Resources Conference and Exhibition held in Brisbane last November featured a special session, “Megaprojects: Past, Present, and Future,” that focused on some of the key issues facing megaprojects in the Asia Pacific region. Panel speakers discussed the main impediments to megaprojects, resource acquisition and management, and the project management skills and business climate in the region.

Another topic the speakers addressed was the role of uncertainty and risk in sanctioning megaprojects. In an environment where more than half of all projects are facing cost overruns, schedule delays, or both, the need to recognize problem areas prior to sanction is as great as ever (Ernst and Young 2014).

One way to improve decision making for megaprojects is to understand, quantify, and mitigate risk and uncertainty. To do that, exploration and production companies are spending money to gather and analyze the necessary data.

Steve Begg, a professor of petroleum engineering and management at the University of Adelaide and a panelist at the megaproject special session, said a company that is biased toward risk aversion or risk-seeking behaviors faces a worse long-term outcome than companies that take a risk-neutral, or an expected-value, approach to decision making.

“If you’ve got large degrees of uncertainty but you’re unbiased, then one project might fail but another one might be brilliant, and if you look at a collection of decision outcomes in a corporation, you’ll be well ahead. So uncertainty isn’t the problem. It’s bias,” he said.

The final investment decision often involves three elements: an owner’s goals for the project, the tasks an operator can realistically execute within a project, and the information an operator has about the project. Begg said a problem most companies face is that the information they have is skewed toward desirable outcomes such as lower costs, lower construction times, and higher production totals.

Begg said that faulty estimations are a measure of uncertainty, and risk is a possible consequence of uncertainty. A company that makes optimistic estimations prior to sanction is not engaging in risk-seeking behavior, and a company that makes cautious estimations is engaging in risk-averse behavior.

Uncertainty Modeling

To help reduce uncertainty in decision making, Begg proposed a holistic and probabilistic approach embedded in a decision support system. This approach, the scholastic integrated asset model (SIAM), includes a technological component that integrates a variety of evaluation and decision-making tools as well as a modeling philosophy that takes into account the magnitude of uncertainty in a decision.

The SIAM is designed to identify the uncertainties that impact decisions the most, value the acquisition of information, and encourage flexibility in forward plans to mitigate and exploit uncertainties (Begg 2013). It contains six elements:

  • Simplified component models for each part of the decision-making process
  • A Monte Carlo simulation engine
  • Modeling language for customization
  • The incorporation of interdependencies between components
  • The implementation of decision logic
  • The updating of information as a result of learning

By integrating the technical and business aspects of decision making, the SIAM establishes a value-driven focus in the work of multidisciplinary asset teams.
Petrobras recently proposed the development of an integrative method to forecasting production development projects (PDPs) that incorporates different uncertainty areas such as the association of the reservoir and activities schedule.

A typical PDP workflow involves parallel schedule-uncertainty analysis. On one track, companies analyze a set of probable schedules generated by a Monte Carlo simulation to determine a set of compliance dates for first oil and ramp-up activities. On the other track, they simulate representative reservoir probabilistic models based on a deterministic schedule and generate a probabilistic net present value (NPV). This leads to a second output of a range of probable outcomes and economic indicators reflecting the uncertainties in a deterministic schedule (OTC 26309).

Petrobras’ approach would merge the two tracks, providing a complete set of probabilistic production curves that encompass all possible schedules in every associated discipline.

The schedule-uncertainty analysis follows a nine-step sequence of activities. The first step is a qualitative analysis and documentation of scheduling risks. The second step is the development of a specific schedule based on the work breakdown structure highlighted in Fig. 1. In the third step, risks are matched with schedule activities, and in the fourth step probabilistic estimates for the duration of schedule activities are obtained through databases or interviews with subject matter experts. These estimates are set as the inputs in the model.

jpt-2016-02-riskfig2.jpg
Fig. 1—The development of a work breakdown structure is one of the steps in the schedule-uncertainty analysis Petrobras performs in its integrated production development project workflow.

 

The fifth step in the process is the attribution of each probabilistic estimate to a given schedule activity, and the definition of points of interest that are then set as outputs. The sixth step is the running of a schedule simulation with appropriate software. In the seventh step, a results evaluation through the output’s probabilistic distributions and sensitivity analysis helps define the inputs that could have the greatest impact on a project. In the eighth step, a risk assessment associated with the most impactful variables to first oil and ramp-up helps determine the most important risks within the project.

The final step is a review of the risk-response plans made during the qualitative risk analysis for maintenance, monitoring, and control, thus restarting the evaluation cycle for a simulation of mixed scenarios.

Petrobras ran 5,000 simulations for an unnamed PDP using the Monte Carlo method, producing 5,000 probable schedules to first oil. To determine which schedule on this spectrum was most likely to occur, the company integrated each one into a reservoir model to develop a probable distribution of startup dates for every well. To account for the limitations that commercial reservoir simulators have in manipulating schedules, it ran each simulation in smaller time increments.

The workflow was then implemented in an uncertainty analysis software program, after which the experimental design was modeled with the simulated schedules covering the maximum variable spectrum. From that, Petrobras could predict the NPV of each schedule from the values of the variables considered.

Costs and Uncertainty

Value creation will become an even greater need for operators as megaprojects continue to grow in scope and cost. In 2014, Ernst and Young estimated that 68% of projects in the Asia Pacific region scheduled for first oil between 2014 and 2035 will face cost overruns, and the average project in the region is expected to run 57% over budget.

The rising cost of megaprojects may lead to more risk-averse decision making from owners. While Begg cautioned against making such a conclusion—only the individuals present during the final investment decision process can really determine if an organization is being risk-averse—he said that high capital costs increase the chances of bad outcomes, and with that increased chance comes an increased aversion to risk-taking action.

“I would say, from what I know and understand, given the fact that if you’ve got a megaproject, it’s likely to be a very significant part of your portfolio, and if it fails it has a very big impact on the corporation, that those are the conditions that would lead to risk aversion,” Begg said.

Risk aversion is not always a suboptimal course of action for owners when handling megaprojects, especially the case in riskier environments. Begg said it can be a reasonable behavior if an owner faces severe consequences for a bad decision made prior to sanction.

“It depends on the size of your portfolio. If you had 50 situations, or even 10 situations that were highly risky, but you took a risk-neutral approach, then maybe over the course of 10 outcomes you would be ahead on average. So it depends on whether you’re looking at an individual event or an individual decision in its outcome, or a collection of decisions and their outcomes,” he said.

In discussing risk aversion, Begg differentiated between decision making, where risk aversion applies, and the estimation of outcomes. He said overconfidence, or a greater belief in the outcome of an event than the evidence suggests, is a strong motivator within the industry.

Risk and opportunity are the two facets of uncertainty: Because their outcomes are known, sure decisions carry no risk and no opportunity for additional gains. Overconfidence lowers the range of uncertainty: If a decision maker is overconfident, the outcome of the decision will look less risky than it is in reality.

Begg said optimism, or the tendency to skew estimates toward a desired outcome, is a subset of overconfidence.

“Overconfidence relates to the range of uncertainty,” Begg said. “I would like costs to be lower, so I tend to predict a lower range of costs. I would like the time to startup to be lower, so I tend to predict that as lower. I would like production to be higher, so my uncertainty estimates can be biased toward the high side. The optimism relates to where the central value (of an estimate) is: Are things generally low or are things generally high?”

In addition, Begg said groupthink narrows the range of uncertainty in decision making, though that narrow range may increase the chance of a faulty decision. Without groupthink, a company would have a wide range of opinions to consider when making decisions, thus mitigating the impact of cognitive bias.

“If you have groupthink combined with a motivational bias, then you might get both optimism in terms of the typical outcome being skewed toward the desired outcome, plus the overconfidence since there isn’t even a wide range of outcomes, because everybody’s thinking the same,” Begg said.

China: Technological Advances

Uncertainty in decision making is not solely a concern for the owners of megaprojects. As unconventional resources take up a greater share of the industry, owners and operators are looking to efficient ways to increase production, especially in newer markets. In some regions, countries are still deciding whether unconventionals are a viable proposition at all.

At the conference, representatives from PetroChina and Schlumberger released the early findings of an integrated project examining the feasibility of certain technologies in the first integrated shale block in the Sichuan Basin, the Longmaxi formation. China is in the early stages of shale gas development, so companies investing in the area must balance the need for capital investment with potentially low economic return.

Longmaxi is one of the more promising shale gas reserves in the country. The formation is a constant marine-deposited layer with an average thickness of 120–200 m characterized by a lower section of calcareous-siliceous shale. It is a thick formation, but recent exploration data have shown its most promising gas-bearing interval is at a depth of 20–30 m (SPE 176861). Total porosity in that interval ranges from 2% to 6%. The total organic carbon (TOC) also ranges from 2% to 6%.

In late 2013, PetroChina started its first batch drilling project in the Zhaotong block, located in the southwestern edge of the Sichuan Basin. The first phase of the project involved the deployment of five platforms with four to eight laterals in each pad. One pilot well was drilled for each platform and studied in detail to give the company a better sense of the formation properties near each operation area. By investigating the microseismic data and production rate from each lateral, the company was able to identify key production drivers and determine the most economic technologies to use on the development.

The company utilized nuclear magnetic resonance, borehole imagery, and induced gamma spectroscopy for conventional wireline logging in its vertical wells, revealing the reservoir and completion quality of Longmaxi and its adjacent formations. It identified two zones within a sublayer that had a TOC content of around 2.8% to 4.0% and a porosity of 2.9% to 5.0%. For logging of its horizontal wells, the company used logging-while-drilling (LWD) tools such as LWD spectroscopy, 3D borehole calipers, and gamma ray imagery.

Typically, operators working in Longmaxi presume that the fracture would propagate upward and downward with low contrast within the section, so they would complete an exploration offset well with the lateral placed in the middle of the Longmaxi organic shale section. However, Liang et al. state that this idea has not been proven by accurate fracture simulation, and previous testing had not considered the possibility of fracture misalignment and excess fluid loss between the interface of the bedding layers of the shale formation. Misalignment and fluid loss could lead to fracture height restriction.

With that in mind, PetroChina incorporated a more integrated drilling and fracturing strategy, evaluating the fracture geometry with two fracturing simulation models. The models accounted for extra leakage and pressure loss through misaligned layers and could be used to understand the geometry of complex fractures by simulating the interaction between hydraulic fractures and natural fractures. This modeling led to the discovery of two optimal landing points from which the company could base its well placement strategy and ultimately execute fracturing treatment.

The first round of batch drilling was successful enough to encourage further production of shale gas in China. PetroChina found that the integrated approach allowed it to develop the reservoir economically. The initial production of its H1 well was 60% higher than a typical offset well with similar wellhead pressure. Wells H2 and H3, which were drilled after H1, improved production totals by an additional 20% to 50%.

A construction program is underway to deliver more natural gas from Longmaxi, and PetroChina will perform further analysis of its production mechanisms as it gathers more correlation data from the field.

Australia: Coalbed Methane

The burgeoning unconventional resources market in Australia is facing a similar technology risk. Arrow Energy’s coalbed methane (CBM) development projects in Queensland are designed to handle the drilling and evaluation of approximately 1,000 wells over a 10-year period. However, high logging costs have forced the company to look into more cost-effective ways to evaluate formations. This included either early data coverage (which limits the project’s economic value) or restricted logging (which increases the project’s risk).

The company’s study, which was presented at the conference, focused on a CBM project in the eastern part of the Surat Basin, an area of approximately 22 000 km. It needed geophysical logs as part of the data acquisition program, particularly during the stages of early development. The mining logging technology consists of a basic logging suite. In order to perform advanced logging in exploration and appraisal wells, the company sought a replacement for the conventional logging tool that would ensure holistic subsurface characterizations and front-end engineering work for future project decision making (SPE 176823).

The study compared three logging technologies: a mining drillpipe push logging tool, where a composite collar is inserted inside the drilling borehole assembly before the production section is lifted; a slimline mining wireline logging tool; and a conventional wireline logging tool, which is the most common practice for CBM formation evaluation.

A comparison of the three technologies can be seen in Table 1. While mining stackable logging tools are more cost-efficient and leave a smaller footprint, they are also less reliable in performing local standard operating procedures.

jpt-2016-02-riskfig3.jpg

For Further Reading

OTC 26309 Reservoir-Schedule Coupled Uncertainty Analysis for PD Projects: Optimization Opportunities and Improvements for More Robust Production Forecasts by V.C. Silva and J.W. Pinto, Petrobras.

SPE 176823 Formation Evaluation Logoff Results Comparing New Generation Mining-Style Logging Tools to Conventional Oil and Gas Logging Tools for Application in Coalbed Methane (CBM) Field Development by T. Gan, B. Balmain, A. Sibgatullin et al., Arrow Energy; E. Murphy and L. Cook, Shell International.

SPE 176861 Technology Feasibility and Production Driver Study in the First Integrated Shale Gas Block in Sichuan Basin by X. Liang, J. Yajun, G. Wang et al., PetroChina Zhejiang Oilfield Company; X. Zhou, Y. Luo et al., Schlumberger.

Begg, S.H. 2013. Some Reflections on Uncertainty, Decisions, Models, and People. In Proceedings: The Second AusIMM International Geometallurgy Conference 2013, 3–6. Melbourne, Australia: The Australasian Institute of Mining and Metallurgy.

Ernst and Young. 2014. Spotlight on Oil and Gas Megaprojects. Internal Report, Ernst and Young Global Limited, London, UK (accessed 28 December 2015).