Article 4: What’s your preferred method for building saturation models?

Date of publication: August 14^th 2018, by Matt Bowyer

A while ago I wrote an article talking about the challenges of saturation modelling and the workflow I usually go through, starting with interpretation of well logs and building onto saturation height functions. I would like to say a big thank you to those of you read it and commented!

This time I'd like to discuss, again quite broadly, a couple of methods I use for building a saturation model.

I appreciate I'm describing this workflow the wrong way round - I should start the workflow by applying environmental corrections, interpreting my logs and then building a saturation height function, but watch this space… I'll be talking about that soon!

In terms of building a saturation model I generally start with a log based approach. It's a pragmatic and usually (although not always) a simple place to start. I already have my spreadsheets setup to perform the calculations. This means I can take log data directly from whatever application it resides in and, with a touch of manipulation, get cracking with doing some saturation modelling (the data manipulation usually takes the form of using a zone log to crop the data to my zones of interest).

My two preferred methods for building a saturation function are the Cuddy FOIL Function and the function proposed by Skelt and Harrison. I know these two methods use well log data, well Skelt-Harrison uses Pc data as well, but I like the simplicity of the log based approaches.

When using the FOIL function, porosity and water saturation logs are the key inputs. I'll ensure that bed boundary effects due to the resistivity tool resolution have been removed and importantly that I'm using only Net reservoir data. Then, it's a case of multiplying the porosity and saturation logs together to get bulk volume of water. I plot this against height above contact on a log-log plot to get my function. It's a great approach, easy to apply in regions where permeability is hard to map and doesn't require porosity banding. Other than checking against the Archie log, what QC steps am I missing here? Also, are there any pitfalls - does the function fit better in a particular facies, for example?

I like the Skelt-Harrison function also - I will freely admit that so far I have only used it with log data. It is a practical function, which usually provides a close fit to the saturation - height data. The main issue that arises is me getting all of the brackets in the right place! However, I am not entirely sure I should be using the function on log data alone … I should probably be including Pc data in there and reconciling the two. I'll get back to you on this one, I need to update my spreadsheet. It does take a bit of effort to step up, and derive, the four constants.

In either case, I can then take the function(s), write them up in a little 'text' file and use this as a macro in Petr…, sorry, reservoir modelling software. In the case of the FOIL function, I get a BVW property first, divide back by porosity and voilà! A saturation model. For Skelt, I can apply the function as is.

There are usually a few QC steps along the way… checking the derived saturations against fluid samples, well test data and so forth. I also like to use my cap curve spreadsheet to build a J function, model this, and then compare back to my logs. Is comparing my log interpretation and my modelled saturations the only way? I did once hear of other QC approaches, back in my big oil days, but never used them. I have little chance of remembering them now.

I'm interested to know whether people have preferred methods for building a saturation model?! What are the key things that we should be considering? Thinking of which, personally I have worked very little in carbonate reservoirs… I'd be very interested to hear how folk handle not just saturation modelling in carbonate reservoirs, but carbonate petrophysics in general!

Given there is a lot curve fitting here, and I'm discussing pragmatic approaches in this article, can I just fit any curve I wish to my data? Talking of which, Matthew Meyer at PDS Ava tested out different saturation height functions and the impact of each function on the hydrocarbon volume calculation. It’s a very interesting read.

I'm very lucky doing what I'm doing at the moment… I have a job I enjoy and I get to do some research along the way. So stay tuned, because next time I'm going to present you with a little case study and actually show some data. I have a great data set (capillary pressure data included!) and I'm going to test a few things out! But, I don't want to just stick to my usual workflows, I really want to make this case-study a multidimensional one, where I incorporate your suggestions too.

Thank you ever so much for reading!

All comments, suggestions, questions, gripes are very welcome.