Modeling for Automatic Control and Estimation of Influx and Loss During Drilling Operations

Abstract

In the modern energy driven world, oil and gas will be important resources, at least in the foreseeable future. Wells with challenging pressure windows are more com-monly drilled, and in recent years the drop in oil prices has lead to an industry focus on reduction of non productive time. Managed Pressure Drilling, considered an un-conventional drilling method, tackles a lot of current industry challenges. With man-aged pressure drilling one has more control over the bottom hole pressure. This lead to the possibility of drilling what would be considered conventionally un-drillable wells, an increase in safety and reduced non productive time.

Managed pressure drilling has been forecasted to grow in the next years, fuelled by an increase in the level of automation. Drilling for oil and gas is a complex process, involving pumping of fluid through kilometres of fluid conduit, leading to wave propagation phenomena that becomes apparent at timescales relevant for automatic control. There has been a significant research effort in academia for both modelling, control design and observer designs in recent years. A lot of the work done in academia has not taken the step from university and out into the field.

Gas influx, and the detection of this has had a surge in research after the Macondo Disaster. The design of kick detection and well control strategies require mathemat-ical models of the same dynamics as control design for managed pressure drilling. It can be said that in terms of kick detection, especially for conventional drilling, it is not the algorithms on how this can be done that is missing, but the sensors required for them. Through the work on this thesis, a possible alternative measurement principle has been studied, using a subcritical venturi flume.

Through the work of this PhD, the topic of modelling, at different detail levels has been studied. This has been performed in parallel with the development of a full MPD control system at Kelda Drilling Controls. This system is now in operation, and has numerous successful wells drilled. For design and upgrades of this control system, extensive testing is performed on a high fidelity PDE model capturing the dynamics of the drilling process. Development of the models and their accuracy is covered in three of the attached papers. For control design, simpler models are usually required. There has been numerous control and estimator designs based on simpler models in literature, but there has been limited publications on real life use of these designs.

The control system developed by Kelda Drilling Controls through the last 5 years is based on the most commonly used simplified model, and has been found have good performance in real drilling operations for both single phase and multiphase cases. Perhaps one of the main success criteria in this has been the extensive testing of different solutions on the high fidelity models coming from this work.

Managed pressure drilling opens up a lot of possibilities when it comes to both detection of unwanted reservoir influx, and circulating the influx out. MPD systems have less regulatory requirements and lower pressure rating than conventional well control equipment. Due to this, performing what would normally be considered a well control operation using MPD equipment should be done with care. The Influx Management Envelope (IME) helps to deal with this. Although wanted by industry, and very likely introduced as a planning tool in the official guidelines by the International Association of Drilling Contractors (IACD) for MPD operations world wide, no systematic description of the IME has existed in peer-reviewed journals. This is covered in the included paper on the IME.