The project focuses on analyzing oil and gas well production data across multiple reservoirs in Africa, applying data cleaning, empirical analysis, and machine learning techniques to derive insights into reservoir performance and well behavior.
# π Reservoir & Well Performance Analysis
## π Overview
This project presents a **comprehensive analysis of oil and gas well performance** using a combination of **data science techniques and petroleum engineering principles**.
The workflow integrates:
- Data cleaning and preprocessing
- Engineering-based feature extraction
- Exploratory data analysis (EDA)
- Machine learning (supervised & unsupervised)
---
## π― Objectives
- Identify the **reservoir for each well**
- Classify reservoirs as **Saturated or Undersaturated**
- Determine **flow mechanism** (Natural Flow vs Gas Lift)
- Analyze **production stability**
- Extract meaningful insights from well production data
---
## π Dataset
- **Total Records:** 7,955
- **Wells:** 20
- **Key Features:**
- Bottom-hole pressure (BHP)
- Wellhead pressure & temperature
- Choke size
- Cumulative oil, gas, and water production
- On-stream hours
---
## βοΈ Workflow
### 1. Data Cleaning & Preprocessing
- Renamed columns for consistency
- Converted date column to datetime format
- Removed commas from numeric columns
- Converted object types to float
- Verified:
- β
No missing values
- β
No duplicates
- β
Consistent well identifiers
---
### 2. Feature Engineering (Domain-Based)
#### Reservoir Identification
- Based on:
- Maximum bottom-hole pressure
- Reservoir pressure constraints (β€ 200 psi differential)
#### Reservoir Type Classification
- **Saturated:** Initial Pressure β€ Bubble Point
- **Undersaturated:** Initial Pressure > Bubble Point
#### Flow Mechanism
- **Natural Flow:** Formation Gas = Total Gas
- **Gas Lift:** Otherwise
#### Production Rates
- Derived from cumulative values using differencing
---
### 3. Well Performance Analysis
#### Production Stability
- Wells classified as:
- **Unsteady** β >50% oil drop within 3β6 months
- **Steady** β otherwise
#### Gas-Oil Ratio (GOR)
- Computed and compared with solution GOR
#### Water Cut Trends
- Evaluated water production behavior over time
---
### 4. Explor β¦