f0nzie / zfactor.dl

Deep Learning applied to compressibility factor z. Correlation generated using neural networks

zFactor.DL

The goal of zFactor.DL is to create a correlation using neural networks.

Dataset

These observations have been taken from a digital scan of the 1942 Standing-Katz chart for hydrocarbon gases.

Original Standing-Katz dataset

The original data was gracefully provided by the authors of the paper [Kamyab, M., et al., Using artificial neural networks to estimate the z-factor for natural hydrocarbon gases, J. Pet. Sci.Eng. (2010), doi:10.1016/j.petrol.2010.07.006] (http://www.sciencedirect.com/science/article/pii/S0920410510001427?via%3Dihub).

The original dataset is an Excel file named SK_data.xls with Ppr, Tpr and z in columnar format with 2853 rows and 60 columns. It is under the folder [./inst/extdata] (https://github.com/f0nzie/zFactor.DL/tree/master/inst/extdata) in this repository. They are three columns or set per Tpr curve.

row    Ppr.0 Tpr.0 z.0    Ppr.1 Tpr.1 z.1  ...    Ppr.19 Tpr.19 z.19
 1
 . 
 .
 .
2853

Tidy Standing-Katz dataset

The tidy dataset contains 57060 observations of compressibility factors at different Pseudo-reduced Temperatures and Pseudo-reduced pressures. The original data was converted to tidy format using R. Refer to notebook 01-read_raw_data.Rmd for step by step process.

The original file has been converted to a tidy format file named tidy_SK.csv with 57,060 observations and 4 variables with the column names Tpr, Ppr, z and set number. This file is under the same folder [./inst/extdata] (https://github.com/f0nzie/zFactor.DL/tree/master/inst/extdata).

The tidy dataset format looks like this:

row     Tpr   Ppr   z   set
   1       
   2
   . 
   .
   .
57060

dim(narrow_dataset)
#> [1] 57060     4