Aquifer Strength and initial reservoir pressure

[Aries]

Hello to everyone,

I'm using MULTPV in Eclipse to simulate aquifer on my model. Initially I used a 4000 multiplier for my pore volume. Found the reservoir pressure is very high (compared to observed) and model water breakthrough is early. In addition to that the model water production is higher also than the observe. So I started tuning in my MULTPV to bring down my pressure and control water production. My question is, why my initial pressure is not changing as I changed my multipliers for pore volume. I only see the lowering of pressure after about 5 years of production . What I mean to say is that for the first 5 years of production, the pressure is unaffected by lowering my porevol multipliers. Could the model be carrying too much oil? Pressure at datum is a given and I trust it. Second question, can I fix first my kv/kh ratio to tackle the early water break and excessive water production before playing around my porevolume multipliers?

I initially used analytic aquifer but it turned out to slow down my model a lot so I gave up on it. Can't find the reason so I gave up on it.

Thanks,

Aries

[DAH7542]

Aries, many things here, i would recommend to do the following:

1. Check that your model initializes under static-equilibrium-condition, i.e. run your model for several time steps (2 to 3 years) without wells and verify that it doesn't have saturation or pressure variations. It's easy and should always be done before proceding with the history matching process.
2. Pressure history match: should be done before saturation matching, so forget about early water breakthrough by now and concetrate on pressure history-matching; pressure variations are two-fold: 2a. Spatial pressure gradient and 2b. Temporal pressure gradient, by looking at the diffusivity equation you can infer what parameters affect each one. Remember to compare simulation-solution and measured pressures at the same datum. It is also advisable for the pressure matching step to run your model under reservoir-volume control (I think you're using liquid rate control now). Variations in aquifer size must affect the temporal pressure gradient, so try to find why not in your case, some ideas:
- Check again that you correctly defined your aquifer (I infer you are using a numerical aquifer based on your comments, and numerical aquifers should be defined in the GRID section under AQUNUM keyword, analytical aquifers are defined in a different section). Open FloViz or Petrel and take a look at the cell and check that is properly defined: the 3D property "AQUIFERN" should give a value less than 0 for that cell (-1 if you only have one numerical aquifer).
- Use material balance (excel or any comercial software) to have an initial guess of the aquifer size.
- Use HMMLAQUN for numerical aquifer modifications instead of the MULTPV keyword.... somethings mistakes can occur when defining the boxes for MULTPV modifications.
- Make drastic changes to the aquifer size (e.g zero pore vol multiplier on the numerical aquifer cell) just to see if it has an effect on pressure, once again, It must have.
- Besides the numerical aquifer cell you defined in AQUNUM, are there too many cells under the oil-water contact?, they also contribute to the aquifer volume.
3. Saturation history match: try it after you have an acceptable pressure match.

Good luck

[Aries]

Thanks DAH7542 for your in-depth reply. Some of the things you suggested to look at had been implemented i.e stability of the model, rates in reservoir volumes, making sure pressure is at datum during comparison. Right now all what I am doing is having big pore vol on the edge of the reservoir(changed by MULTPV). The analytical aquifer that I started initially was abandoned due to run time issue. But I will go back to it again based on your suggestion. Thanks for your time and effort to respond. Many thanks again and will update you shortly.

Aries

[brhm]

also make sure you are comparing the correct reservoir pressures. Pore volume weighted average might be different than the measurement you obtained from the volumes in communication.

ps as you have experienced, multpv multiplications can result convergence errors...