Go / No-Go Decisions

The Issue

E&P projects are often related to execution decision gates as a means of de-risking a project. The most typical example is an Explore-Appraise-Develop execution path, where an exploration well is drilled, and if that well reveals hydrocarbons, a decision is made to proceed with Appraisal (Go) or exit (No-Go). Similarly with Appraisal: if the appraisal program results in a certain volume of hydrocarbons, the decision is made to proceed with Development or to stop.

explore appraise develop graphic

Figure 1: Typical Decision Gates for Explore, Appraise and Develop

More complex examples also exist. Complex multiple pilot programs are a good example, whereby a path to a large development scale-up might be made through several pilot programs that progressively get bigger (and more expensive) as they proceed.

Often, the question in this type of development is one of value of information, whether the risk reduction of the series of pilots is worth the cost, and so often, a number of alternatives need to be investigated using different combinations of possible pilots.

Typically these types of questions are addressed by decision trees. The challenge here is they can become very large, especially when an integrated approach is required (and it usually is) that accounts for all of the other uncertainties surrounding the decision, in addition to the one decision being investigated by the pilot testing. In such a case, a decision tree approach can become very cumbersome due to the large number of end nodes (thousands or tens of thousands) leaving the analyst to revert to shortcuts, which can detract from a full understanding of the full range of possible outcomes.

explore appraise develop graphic test development appraisal success exploration results

Figure 2: Simple Decision Tree with only eight end nodes

The PetroVR Solution

PetroVR is a discrete event simulator with a powerful built-in scheduling capability that allows for modeling exactly how a developer would envisage developing the project over time. With this, PetroVR helps take a more natural and intuitive approach to modeling decisions rather than the traditional decision tree methodology. The decisions are modeled right into the schedule at the point in time when they are to be made. Logic can be attributed to each decision event based on how it would actually be made, given the information value measure that is available at that time.

Logic can be attributed to each decision event right in the model, just as
the decision would be made in real time

Let's take the simple example of a Decision Tree shown above and show how easy it is to conceptualize and model in PetroVR. The first activity on the schedule is Exploration. After that, a decision is made and modeled right into the schedule itself. The decision to be made is: if the exploration well is a success (hydrocarbons present) then proceed with the next activity (Appraisal), and if not stop the project.

In the next activity, Appraisal, a second decision is modeled into the schedule after that activity. In this case: if reserves > minimum economic cutoff volume then proceed with next activity, if not, stop the project. The next activity is an extended production test, and the decision after that is based on the outcome of the test, build small, medium or large facilities.

As can be seen by this workflow, in PetroVR it is simple and easy to build any number of decisions right into the schedule and let the PetroVR simulator take care of the outcomes and decision paths through it's powerful combination of a time event simulator together with uncertainty characterization via Monte Carlo simulation. The analyst must only capture a relatively few number of decisions, and PetroVR calculated all possible outcomes, rather than in a decision tree world where every outcome (many) must be calculated through manually. This saves the analyst a significant amount of time, and is much less error prone.

Here is how it might look in PetroVR:

petrovr decision exploration success decision based outcomes small medium large facilities

Figure 3: Decisions modeled into a PetroVR Schedule

How It Works

  1. The range of uncertainty is expressed by the analyst for each of the inputs.
  2. This uncertainty range is sampled by the Monte Carlo simulation and the decision tests are performed. If the sample passes the test, the next phase is executed, if the sample fails the test, then the project is stopped and there is an exit. An example of a fully successful exploration, appraisal and piloting program, as well as a couple of failure cases that lead to project exit, are shown below.
  3. An Expected Monetary Value (EMV) for the test program is established by repeated simulation runs. In each run, the appropriate decision path is followed on the basis of the sampled values of all model uncertainties.
exploration drill exploration success graphic yellow pink green build option
Figure 4A
exploration drill exploration success graphic yellow pink
Figure 4b
exploration drill exploration success appraise graphic
Figure 4c

Simulation Schedules (Figs. 4a - 4c) reveal one sample that passes all decisions, a second that fails at the first decision gate, and a third that fails at the second decision gate.

The Results

Compared to decision trees, a discrete event simulator can save a significant amount of modeling time and is much less error prone. PetroVR takes a natural and intuitive approach to modeling decisions, where decisions are modeled directly into the schedule at the point in time when they are to be made. Logic is attributed to each decision event based on how it would actually be made, given the information value measure available at that time.