Bottom Hole Pressure (BHP) Prediction Model

This project implements a machine learning model to predict Bottom Hole Pressure (BHP) in oil wells based on various well parameters.

Features Used

Qo: Oil production rate (STB/day)
GOR: Gas-Oil Ratio (scf/STB)
THT: Tubing Head Temperature (°F)
Pwh: Wellhead Pressure (psi)
WCT: Water Cut (%)
Depth: Well depth (ft)

Derived Features

The model uses these engineered features:

Fluid gradient: (WCT/100)*0.433 + (1-(WCT/100))*0.273
Ph (Hydrostatic Pressure): Fluid gradient * Depth

Author

Kwadwo Fosu Adom

Model Usage

import pickle
import pandas as pd

# Load your test data
test_df = pd.read_csv('your_test_data.csv')  # or other source

# Calculate derived features
test_df['Fluid gradient'] = (test_df['WCT']/100)*0.433 + (1-(test_df['WCT']/100))*0.273
test_df['Ph'] = test_df['Fluid gradient'] * test_df['Depth']

# Features to scale (must match training)
scaled_features = ['Qo', 'GOR', 'THT', 'Pwh(psi)', 'Ph', 'Depth']

# Load model and scaler
with open('modelBIGDATA5US1P57.pkl', 'rb') as file:
    saved_data = pickle.load(file)
    model = saved_data['model']
    scaler = saved_data['scaler']

# Make predictions
X_test_scaled = scaler.transform(test_df[scaled_features])
test_df['Predicted_BHP'] = model.predict(X_test_scaled)