Two stochastic versions of the Arps curve decline

Christian Paroissin

2024-11-04

Abstract:Based on the Arps equation, we propose two stochastic models for curve decline useful in oil engineering context. Theoretical properties and simulations of these models are provided. The first passage time distribution of these stochastic models to a constant level is then studied. In conclusion, we discuss about statistical inference of the parameters from the observations of the oil production cumulative rate.

Probability

What problem does this paper attempt to address?

This paper attempts to solve the problems of modeling and analyzing the production decline curves of oil wells in the field of petroleum engineering. Specifically, the author proposes two stochastic models based on the Arps equation to describe the decline of oil well production over time. These models aim to more accurately reflect the uncertainties in oil well production and provide theoretical properties and simulation results. In addition, the paper also studies the time distribution when these stochastic models first drop to a certain fixed level, and discusses the method of parameter statistical inference from the observed cumulative oil production rate.

### Main Problems and Solutions

1. **Introduction of Stochastic Models**:
- The author proposes two models based on stochastic differential equations (SDEs) so that the expected values conform to the classical Arps equation. This helps to more precisely describe the decline process of oil well production under the premise of considering uncertainties.

2. **Study of First - Passage - Time Distribution**:
- The paper studies the time distribution when these stochastic models first drop to a certain fixed level (i.e., the first - passage - time distribution), which is of great significance for predicting when the oil well production will be below a certain threshold.

3. **Parameter Estimation and Statistical Inference**:
- Finally, the paper discusses how to perform parameter estimation from the observed cumulative oil production rate data, which is crucial for model calibration in practical applications.

### Specific Contents

- **Arps Equation**: The classical Arps equation has the following form:
\[
q_t=\frac{q_0}{(1 + b d_0 t)^{1/b}}
\]
where \( q_t \) is the oil production rate at time \( t \), \( q_0 \) is the initial oil production rate, and \( b \) and \( d_0 \) are the shape parameter and the initial flow constant respectively.

- **Stochastic Models**:
- The first model assumes a constant volatility \( \sigma \), and the corresponding SDE is:
\[
dQ_t = -\frac{d_0 Q_t}{1 + b d_0 t} dt+\sigma dB_t
\]
- The second model assumes a linearly increasing volatility \( \alpha(x)=\sigma x \), and the corresponding SDE is:
\[
dQ_t = -\frac{d_0 Q_t}{1 + b d_0 t} dt+\sigma Q_t dB_t
\]

- **First - Passage - Time Distribution**:
- For these two models, the paper studies their time distribution when they first drop to a certain fixed level and analyzes them using the relevant theories of Brownian motion.

- **Parameter Estimation**:
- Since in practical applications, usually only the cumulative oil production rate can be observed, the paper discusses how to perform parameter estimation from these observed data.

In conclusion, by introducing stochastic models and studying the first - passage - time distribution, this paper aims to more precisely describe and predict the decline process of oil well production and provides relevant theoretical and numerical simulation results.