From Professor Khalid Aziz, Stanford University
– “Ten Golden Rules for Simulation Engineers”, JPT – November 1989, 1157
1. Understand Your Problem and Define Your Objectives
Before you do any simulation:
• Understand characteristics of reservoir
• Understand fluids
• Objective of study clearly stated on paper
• Ask yourself if the objective are realistic
2. Keep it Simple
Start and end with simplest model
Consistent with
• the nature of the reservoir
• Objective of study
• Availability of data
Most sophisticated model available – may not serve your needs
Understand model limitations and capabilities
3. Understand Interaction Between Different Parts
Reservoir not an isolated entity
Connected to
• Aquifers
• Surface facilities
Separation into different components
• may be inappropriate – neglects interactions
• when appropriate – can lead to substantial savings
4. Don’t Assume Bigger is Always Better
• Always question size of a study that is limited by the computer
resources and/or budget
• Greater number of blocks and components do not automatically
translate into greater accuracy and reliability (reverse sometimes true)
5. Know Your Limitations and Trust Your Judgments
• Remember simulation is not an exact science – more of an art
• Trust your judgment – based on analysis of the field or lab observations
• Do simple material balance to check simulation results
6. Be Reasonable in Your Expectations
• Don’t try to get from the simulator what it is incapable of producing
• Remember – if you exclude a mechanism during model development
– cannot study its effect with that model
7. Question Data Adjustments for History Matching
• Remember HM process does not have a unique solution
• A good HM with inappropriate adjustments to the data will lead to poor predictions
• Pay close attention to physical and geological reasonableness
8. Don’t Smooth Extremes
• Pay attention to extremes in permeability (barriers and channels)
• Be careful in the process of averaging
• Never average out extremes
9. Pay Attention to the Measurement and Use Scales
Measurement values at the core scale may not directly apply at larger block
scale – do influence values at other scales
• Permeability may be a scalar at some small scale – and a tensor at larger scale
• Dispersive terms in our equations are a result of process of averaging
10. Don’t Skimp on Necessary Laboratory Work
Models do not replace lab experiments
• designed to understand the nature of the process
• Or measure essential parameters of the equations being solved
Plan lab work with its end use in mind