Reservoir Engineering – Pitfall Series

In this new series, we will be exploring the common difficulties, hazards and drawbacks that can occur during project lifecycles and how to successfully navigate them. We will be looking at this from various angles including reservoir engineering, geology and geophysics. As we go through, we hope to get you to think about the challenges you have faced and how you dealt, or would deal with, the issues raised. We invite you to share your thoughts with us and the wider geoscience community on LinkedIn.

Continue reading this article for the first in the series, where our reservoir engineer Ben Adillah takes a look at navigating reservoir model initiation challenges.


Pitfalls in Dynamic Model Initialisation

While seemingly straightforward, the initialisation phase of reservoir modelling can introduce significant errors if not approached with meticulous attention.

Pitfall #1: The Rush to Initialise

Why the rush? Should one proceed initialising even when input is not QC-ed?

Imagine a scenario where a crucial reservoir modelling decision hinged on predicted production forecasts. Yet, those forecasts were skewed due to inaccurate initialisation data. Unfortunately, this exemplifies the real-world consequences of rushing through the initialisation phase.

In the fast-paced environment of any project, the pressure to expedite work processes are understandable. However, succumbing to this pressure and neglecting proper quality control (QC) of dynamic input data can lead to significant issues later on. Inaccuracies in pressure, production rates, and fluid properties can propagate through the model, resulting in erroneous predictions and flawed decision-making. As we all know, garbage-in, garbage-out.

Taking the time to meticulously QC dynamic input data is crucial. This involves verifying data consistency, correcting discrepancies, and validating sources. While it may seem time-consuming upfront, investing in proper QC is a wise long-term strategy. It can save substantial time and resources by preventing costly rework and ensuring data integrity throughout the modelling process.

Pitfall #2: Choosing an Accurate Initialisation Method

Do I know which type of initialisation method to use?

Selecting the appropriate initialisation method is critical and depends heavily on the specific characteristics of the reservoir. Different methods offer varying advantages and limitations. Familiarising yourself with the various options, such as:

  • Static initialisation: Uses static reservoir properties (e.g., saturation, formation volume factor) throughout the reservoir volume.
  • Equilibration: The simulator computes initial saturation and pressure using the fluid model and saturation function as defined. It assumes that gas, oil, and water are in hydrostatic equilibrium.
  • Enumeration: The user explicitly assigns each initial time-dependent property on each grid block based on their different phases. As each primary variable are defined, there are chances that the model might not be in equilibrium, thus a stability check is highly suggested.

The choice should be tailored to the reservoir’s conditions and the objectives of the modelling exercise. Consulting with subject matter experts or conducting sensitivity analyses can help determine the most suitable approach. This ensures that the initialisation method aligns with the reservoir’s unique properties and the goals of the modelling project.

Pitfall #3: Maintaining Model Stability

How stable is the model?

Model stability is essential for reliable simulation results. An unstable model can produce unrealistic predictions, undermining the validity of the entire modelling effort. It’s crucial to perform an initialisation stability test at the start to evaluate the model under a simulation with no wells for a few years. This helps identify issues like:

  • Large and unrealistic changes in pressure or saturation over time.
  • Unphysical flow behaviour.

Identifying and mitigating sources of instability—such as grid resolution issues, incorrect boundary conditions, or overly large timesteps—is key. Continuous monitoring and necessary adjustments throughout the modelling process help maintain model stability, providing a solid foundation for accurate reservoir simulations.

Pitfall #4: Water Saturation Modelling Woes

What kind of Sw modelling should be performed in the initialisation?

An example of a water saturation model in Petrel.

Water saturation (Sw) modelling is a fundamental aspect of reservoir initialisation. Its accuracy can significantly impact the reliability of the entire model. Different approaches, like the Leverett J-function, or empirical correlations, offer various benefits and challenges.

Choosing the right Sw modelling technique requires a thorough understanding of the reservoir’s petrophysical properties. Conducting sensitivity analyses can be invaluable. This helps assess the impact of different Sw models on the reservoir’s performance, enabling you to select the method that provides the most accurate representation of water saturation in your specific reservoir context.

Initial water saturation (Swi) distribution typically comes from log analysis where the height is dependent on porosity, permeability, and facies. Incorporating capillary pressure (Pc) scaling can further enhance the accuracy of Sw modelling. This involves adjusting Pc curves based on reservoir conditions to better match observed data, improving model fidelity.

An additional approach to consider is SWATINIT. This method leverages the expertise of the petrophysicist and geomodeller, who provide an initial water saturation property within the static model. The reservoir engineer then uses this pre-defined Sw distribution for initialisation. SWATINIT offers the advantage of incorporating observed data directly into the model.

Pitfall #5: Balancing the Volumes (or Volume Reconciliation)

Do my Static and Dynamic volume match?

An example of a static model porosity property grid from Petrel.

A common pitfall in reservoir modelling is the mismatch between dynamic and static volumes. Ensuring these volumes are consistent is crucial for accurate reserve and resource estimation and production forecasting. When starting a simulation study, we usually have a volume calculated by the geologist. The engineer should assess this volume calculation and evaluate any differences with the initial dynamic output.

Here’s how to address volume discrepancies:

  • Geologist vs. Reservoir Engineer Considerations: The geologist’s volume calculation typically includes all cells above the oil-water contact, with near-contact volumes determined by triangulation. The simulation engineer might use all active cells with specific considerations for transition zones.
  • Reconciliation Techniques:
    • Excluding inactive cells: Removing cells below the oil-water contact can improve pore volume accuracy for oil-bearing zones.
    • Fine grid equilibration: Running the initialisation with a refined grid near the oil-water contact can improve the accuracy of volume calculations in this critical zone.
    • Quiescence option: Reservoir simulation software generally offers a “quiescence” option. This method essentially slices the cells near the contact into smaller sections, allowing for a more precise calculation of the volume near the oil-water contact.
    • Fluid property reconciliation: Differences in how fluid properties (e.g., Bo, Bg, Rs) are defined between the static model and the initialisation process can also lead to volume mismatches. Here, you can:
      • Extract the initialisation simulation results for fluid properties like Bg, Bo, and Rs.
      • Use these extracted properties as input for the static model volume calculation. By implementing these techniques, you can reconcile volume discrepancies and ensure consistency between the static and dynamic models.

Pitfall #6: Checklists are Your Best Friend

Do you have a checklist to ensure all crucial initialisation steps are covered?

Overlooking a critical step during initialisation can have significant consequences downstream in the modelling process. Developing and adhering to a comprehensive checklist can help mitigate this risk. Here are some key items to consider including in your checklist:

  • Data quality control (QC) of dynamic input data: Verify data consistency, correct discrepancies, and validate sources.
  • Selection of the appropriate initialisation method: Consider reservoir characteristics and modelling objectives.
  • Initialisation stability test: Identify and address potential instability issues.
  • Water saturation (Sw) modelling: Choose a method suited to the reservoir and conduct sensitivity analyses if necessary.
  • Volume reconciliation: Ensure consistency between static and dynamic volume calculations.
  • Review for error messages: Address any error messages that arise during initialisation.
  • Consistency check between static and dynamic assumptions: Verify alignment between the assumptions used in both the static and dynamic models.
  • Saturation review: Pay close attention to saturation distributions, especially if different methods were used in the static and dynamic models.

By following a well-defined checklist, you can systematically address these critical steps and minimize the risk of errors during initialisation.

Conclusion: Avoiding the Pitfalls

Initialising a reservoir model is a complex yet crucial process that demands meticulous attention to detail and a deep understanding of the reservoir’s characteristics. By addressing these common pitfalls and continuously asking the right questions, reservoir engineers and geoscientists can develop more accurate and reliable models, leading to better decision-making and improved reservoir management.

#pitfall Share your experiences with reservoir modelling pitfalls! What challenges have you encountered during the initialisation phase?

Look out for other articles in the series covering other #pitfall topics.