Scheduling Drilling Processes With Petri Nets

Abstract

Safety issues in drilling are related to two facts: Wells are becoming more complex, and manually piloting a drilling rig is a difficult task which requires highly skilled personnel. Consequently, improving safety is conditioned on a better anticipation of the well behaviour, and an easier way of operating drilling rigs. On top of safety issues comes the drilling industry vision of autonomous drilling control systems. This vision aims at realizing a drilling control system, which not only is capable of executing a drilling program but can also automatically respond to incidents. However, we need to overcome a number of challenges before the autonomous drilling vision comes true. In this thesis we aim to address the following challenges: First, we need to provide a system component which guaranties a safe control of the rig. That is, any control operation that can be performed has to be legal. Such a component is called control supervisor. Second, we need to provide a capability for handling incidents. This includes processes that can monitor the well dynamics and trigger actions to cope with eventual incidents. Such processes are called reactive processes. Third, because reactive processes could trigger conflicting actions, we need a mechanism to coordinate them. We call such a mechanism reactive process scheduler. Realizing a control supervisor has its foundation in the Discrete Event System (DES) paradigm. The main problem we address in that context is to provide a DES model that captures the dynamics of the rig, and which can be checked for correctness. Realizing a reactive process scheduler is related to obtaining an emergent behaviour out of basic ones. A basic behaviour is associated to every reactive process, and the task of a reactive process scheduler is to coordinate those reactive processes in order to obtain a satisfactory behaviour of the overall system. The main contributions of this thesis are: 1. Bringing to light some hidden challenges related to drilling control systems. 2. Including two system components to the existing drilling control system architecture: A control supervisor, and a reactive process scheduler. 3. A Petri net class to model the control supervisor which properties can be fully analysed. 4. A theoretical approach for modelling reactive processes and their scheduling. All in all, this thesis aims to not only ease the development of safer drilling systems, but also to take a step towards the more ambitious vision of autonomous drilling.