● AI · Completions ANNBP ANFIS JTMGB Vol.4 No.1 · 2013

AI for post-hydraulic
fracturing gas rate
prediction

Neural network backpropagation (NNBP) and adaptive neuro-fuzzy inference system (ANFIS) applied to predict incremental gas rate following hydraulic fracturing — across 11 input parameters, 20 actual datasets, validated against December 2011 field actuals.

11
Input parameters
20
Actual datasets
0.989
ANFIS R² · 3 test pts
0.970
R² · Dec 2011 validation

Rock, wellbore, and operational inputs

The study selected 11 parameters as inputs, representing three domains of hydraulic fracturing performance. All data was calibrated by nodal analysis to a common datum (VLP production system) to make datasets from different wells comparable.

/ Reservoir
Thickness (h)
/ Reservoir
Water saturation (Sw)
/ Reservoir
Porosity (φ)
/ Reservoir
Permeability (k)
/ Well productivity
Pressure gradient (ppg)
/ Well productivity
Skin factor (s)
/ Well productivity
Productivity index (PI)
/ Well productivity
Critical rate factor (Cri)
/ Well productivity
Cement bonding quality (CBL)
/ Operational
Gel volume (Gel)
/ Operational
Proppant mass (Prop)
/ Output
Incremental gas rate (%)

ANNBP architecture

Feedforward backpropagation network — 11 inputs, 1 hidden layer with 1 unit, 1 output. Activation functions: logsig and purelin. Dataset split: 12 training, 7 no-CBL, 1 December 2010 testing.

  • Pattern Recognition — map input-output relationships
  • Identifying Performance Drivers — delta R² sensitivity per parameter
  • Performance Prediction — generalized network with validation set

ANFIS architecture

Sugeno FIS with subtractive clustering (cluster radius determines rule number = number of datasets). Hybrid optimization: backpropagation + least squares. Validation dataset: 4% of total data.

  • FIS method: Sugeno
  • Initialization: Subtractive clustering
  • Optimization: Backpropagation + least squares
  • Rule number = 12 (= number of datasets)

ANFIS outperforms NNBP when training data is reduced

Both methods achieved R² approaching 1 in training. The decisive difference appeared in testing scenarios — when training data was progressively reduced and testing data increased:

Method 1 test point 2 test points 3 test points
NNBP 0.990 0.988 0.935
ANFIS 0.999 0.989 0.990

ANFIS combines fuzzy inference system (FIS) with backpropagation neural networks — this combination provides better generalization with fewer training points than standard NNBP.

Proppant mass, water saturation, skin factor, porosity

Sensitivity analysis using the trained networks (running without each input parameter and calculating delta R²), combined with simple linear regression for confirmation, identified four parameters with average influence exceeding 10%:

  • Proppant mass (~30%): Directly represents fracture volume and dimensionless fracture conductivity (CfD) — increasing proppant increases effective wellbore radius (rw')
  • Water saturation (~23%): Reservoir characteristic related to hydrocarbon volume and fluid-rock interaction; influences capillary pressure behavior in low-permeability formation
  • Skin factor (~23%): Near-wellbore damage descriptor; proportional to post-treatment productivity index per Rae et al. (1999) skin bypass treatment equation
  • Porosity (~10%): Controls hydrocarbon volume and fracturing fluid efficiency; higher porosity increases fluid loss, reducing fracture propagation and potential incremental rate

A reduced 4-parameter ANFIS model (using only these four drivers) achieved R² = 0.97537 (validation) and R² = 0.99992 (testing) — demonstrating that four parameters can explain the majority of hydraulic fracturing performance variance.

Excel application tested against live field data

The final ANFIS network was deployed as an Excel macro application for gas rate prediction with additional proppant mass sensitivity analysis. The model was tested against the most recent actual data from a job performed in December 2011 — data the model had never seen during training.

The application predicted 0.400 Mmcfd; the actual job result was 0.470 Mmcfd. R² = 0.96985 — excellent match with actual field data within operating tolerance.

/ Publication
Rachmat, S. & Hadad, A.F. (2013).
Successful Application on Artificial Intelligence Methods for Prediction of Post-Hydraulic Fracturing Gas Rate.
Jurnal Teknologi Minyak dan Gas Bumi (JTMGB), Vol. 4, No. 1, April 2013: 13–23.
ISSN 2088-7590. Institut Teknologi Bandung / Vico Indonesia.