Page 278 - PVT Property Correlations

244 PVT Property Correlations

ANN models consist of many components, with each component having
several alternatives for performing calculations. These countless combina-
tions and permutations offer investigators a multitude of optimization oppor-
tunities. In addition, the development of AI techniques is currently receiving
considerable attention. Alternatives and options will continue to appear, and
general advancements in AI are expected to yield improved models for PVT
properties prediction.
The main disadvantage of ANN models remains the lack of available
details for models developed. In general, investigators do not publish the
ANN weights, and therefore the results cannot be reproduced. PVT correla-
tions results, however, can be easily reproduced.

NOMENCLATURE
a weighted summation
AI artificial intelligence
b input for the output node
B bias node
c output after the transformation function
C threshold
F activation function
G transformation function
h activated weighted summation
I input node
equilibrium constant for component (i)
K i
O output node
PSO particle swarm optimization
SOM self-organizing maps
W weight value
X input value
Y output value

REFERENCES
Adeeyo, Y.A. 2016. Artificial neural network modelling of bubblepoint pressure and formation
volume factor at bubblepoint pressure of Nigerian crude oil. In: Paper SPE-184378-MS
Presented at the SPE Nigeria Annual International Conference and Exhibition, 2 4 August,
Lagos, Nigeria. https://doi.org/10.2118/184378-MS.
Ahmadi, M.A., Pournik, M., Shadizadeh, S.R., 2015. Toward connectionist model for predicting
bubble point pressure of crude oils: application of artificial intelligence. Petroleum 1,
307 317. Available from: https://doi.org/10.1016/j.petlm.2015.08.003.
Al-Gathe, A.A., Abd-El Fattah, K.A., El-Banbi, A.H., El-Metwally, K.A., 2015. A hybrid neuro-
fuzzy approach for black oil viscosity prediction. Int. J. Innov. Appl. Stud. 13 (4), 946 957.
Alimadadi, F., Fakhri, A., Farooghi, D., Sadati, H., 2011. Using a committee machine with artifi-
cial neural networks to predict PVT properties of Iran crude oil. SPEREE 14. Available
from: https://doi.org/10.2118/141165-PA.