Improved Model Support in Drilling Automation

To function properly, your car needs regular maintenance. But when was the last time you gave a computer program the same treatment? Drilling of oil and gas wells is a challenging operation. There will be large uncertainties around what you will meet as you drill down, or even what is going on in the well right now. That is why drillers today use real-time computer models, which can calculate what is happening or could happen in the well and pass this on to the driller. The same information may be shared with a drilling automation system, which together with the driller can help control the drilling process both faster and more accurately than the driller could have managed alone. Real-time computer models are therefore an important part of the drilling operation. But how do we know that the models work? An engine might be inspected regularly by a technician and you can even listen to the engine noise for signs of trouble. But just like an engine, a computer model's performance can also degrade, perhaps without anyone noticing. What it then tells the driller and the automation system may not correspond to reality. Perhaps the model was fed with incorrect or too little information about the well. Perhaps there is trouble with how it interacts with other machines and computer systems on board the drilling rig. Or maybe something dramatic is about to take place down in the well, something which neither man nor model has yet realized. One of the goals of this project is that the computer model receives the same level of monitoring and maintenance as the rest of the machinery on the drilling rig. With better model maintenance, wells will be drilled faster, safer and at less cost. More reliable models will also push the limits in terms of what drilling automation systems can accomplish.

The project has delivered a functional prototype of the SFM, which later was upgraded in the DEMO2000 project "Drilling Mud Process Control". Huisman is actively working to integrate it in the drilling rig control system. Both the technology and competence developed in the KPN has contributed to the development of additional related projects. These include the DEMO2000-projects "AutoViscosity" (project owner National Oilwell Varco) and "Digitized fluid transport" (project owner Kongsberg Digital). Further project initiatives based on the same technology and competence are under development.

As the oil and gas industry in Norway is facing more challenging areas, drilling automation will be a prerequisite to meet the challenges of longer, colder and deeper wells. The well is a dynamic system, which often requires automation to be enhanced by r eal-time wellbore models. These are computer programs that predict the behaviour of the well based on measurements. However, drilling with conventional drill strings only provide a few sparse measurements from the bottom of the well, making it difficult to detect if something is going wrong. In contrast, networked drill strings, also known as wired drill pipe, provide a broadband connection to the wellbore and enable measurements to be taken along the wellbore. The project aimes at improving the perfo rmance of drilling models by treating it as one would a physical machine. This includes monitoring of its performance and routine maintenance. In this case maintenence is equivalent to calibration, ans special attension will be put on hole cleaning perfor mance and topside handling of solids. The project raises questions such as "what does it take to drill a well safely and efficiently, despite limited and inaccurate information?" For automation, the answer depends on how the software reacts to inaccurate information. The project aims at detecting these reactions and correcting the situation, so that bad data do not jeopardize the drilling operation. If successful, the project will help deploy automation under more challenging conditions, which in turn enable safer and more cost-efficient drilling.