| dc.description.abstract | This dissertation investigates dynamics modeling, optimization, and control methodologies of the down-hole drilling system, which can enable a more accurate and reliable automated tracking of drilling trajectory, mitigating drilling vibration, improving the drilling rate, etc. Unlike many existing works, which only consider drilling control in the torsional dimension, the proposed research aims to address the drilling dynamics modeling and control considering both coupled axial and torsional drill string dynamics. The dissertation will first address optimization and control for vertical drilling, and then resolve critical modeling and control challenges for the directional drilling process.
In Chapter 2, a customized Dynamic Programming (DP) method is proposed to enable a computationally efficient optimization for the vertical down-hole drilling process. The method is enabled by a new customized DP searching scheme based on a partial inversion of the dynamics model. Through extensive simulation, the method is proved to be effective in searching for an optimal drilling control solution. This method can generate an open-loop optimal control solution, which can be used as a guide for drilling control or in a driller-assist system.
In Chapter 3, to enable a closed-loop control solution for the vertical drilling, a neutral- delay differential equations (NDDEs) model based control approach is proposed, specifically to address an axial-torsional coupled vertical drilling dynamics capturing more transient dynamics behaviors through the NDDE. An equivalent input disturbance (EID) approach is used to control the NDDEs model by constructing the Lyapunov-Krasovskii functional (LKF) and formulating them into a linear matrix inequality (LMI). The control gains can be obtained to effectively mitigate the undesired vibrations and maintain accurate trajectory tracking performance under different control references.
The works on Chapter 2 and Chapter 3 are mostly for vertical drilling, and the remaining of the dissertation will focus on modeling and control for directional drilling. Chapter 4 proposes a dual heuristic programming (DHP) approach for automated directional drilling control. By approximating the derivative of the cost-to-go function using a neural network (NN), the DHP approach solves the “curse of dimensionality” associated with the traditional DP. The result shows that the proposed controller is robust, computationally efficient, and effective for the directional drilling system.
To validate the DHP based control method using a high-fidelity directional drilling model, a hybrid drilling dynamics model is proposed in Chapter 5. The philosophy of the proposed modeling approach is to use the finite element method (FEM) to describe curved sections in the drill string and use the transfer matrix method (TMM) to model straight sections in the drill string. By integrating different methods, we can achieve both modeling accuracy and computational efficiency for a geometrically complex structure. Compared to existing directional drilling models used for off-line analysis, this model can be used for real-time testbeds such as software-in-the-loop (SIL) system and hardware-in-the-loop (HIL) system.
Finally, a software-in-the-loop real-time simulation testbed is built to test the designed DHP based controller in Chapter 6. A higher-order hybrid model of directional drilling is implemented in the SIL. The SIL results demonstrate that the designed DHP based controller can effectively mitigate harmful vibrations and accurately track the desired references. | en |
