The quality of a history match depends on many things of which the geological model (geostatistical or otherwise) is only one. For example, in a field with relatively few wells and no downhole monitoring it is often possible to achieve very good history matches with significantly different geological realisations. However, if the problem is more highly specified, that is where more there is more data, achieving any acceptable history match may be a challenge.

It is worth remembering that the aim of the history matching process is not to achieve a "perfect" history match. It is to produce a model which reliably predicts the future and allows you to investigate how you might optimize that future.

My opinion is No. I agree with Ann. And the same conclusion is true for every model.

If you are testing some feature of a model (let's say production), accuracy in the prediction of that feature does not guarantee accuracy in all it's other features.

Probably the best method for you to test your geostatistical reservoir model (seismic inversion, Acoustic Impedance, Elastic Impedance, Porosity, Lithologies and so on...) is to test it against blind data (because that data is real).

Ann and Pedro,

You are right about the particular case of few wells with short production history. However, the case that I have is very complex. I have giant oil field with 60 production and injection wells and approximately 60 years of production history.

i have generated multiple realizations for the optimization work. These different realizations.have led to significant reservoir flow response during the prediction period.

My opinion in this case, if the obtained perfect history matching does not reflect the accuracy of the geostatistical model, there is no feasibility to spend time and efforts to build high quality geological model.

Watheq,

I suggest starting from the position that a simulation model is always, always WRONG. There is no such thing as a perfect model. We simply do not have the data. Even when we build simulation models of a core flood, we are still simplifying the system.

So, if you have generated multiple realizations, then you have multiple wrong models. However, this is not a wasted effort, because the predictions that these models give, starts to give you an idea of the possible future for your field. The weight you put on each of these futures should in some way be determined by how wrong their history matches are. So from a geological perspective you may have mapped out the probability space.

If you haven't already done so, don't forget to consider other uncertainties such as fluid properties, relative permeabilities & well damage.