Digital Transformation

E&P – AI adoption historical.

Over the years, AI has led to significant designing and computation optimizations in the global Petroleum Exploration and Production (E&P) industry, and its applications have only continued to grow with the advent of modern drilling and production technologies. In recent years, there has been a rapid increase in the number of artificial intelligence tools applications in petroleum industry due to greater availability of human expert and numerous publications of successful application case studies (Gharbi et. al. 2005). Artificial neural networks, fuzzy logic and evolutionary algorithms are the most commonly used AI techniques today in various petroleum engineering applications; oil and gas reservoir simulation, production, drilling and completion optimization, drilling automation and process control (Braswell 2013).

With the recent successful result of AI as intelligent systems tools in E&P industry, it is becoming explicit that E&P industry has realized the importance applicability of the intelligent model solving several disciplinary related problems. There is a great potential for these tools in exploration, production and management of hydrocarbons.

Current trends in the industry

Today, AI systems form the backbone of digital oil field (DOF) concepts and implementations. However, there is still great potential for new ways to optimize field development and production costs, prolong field life, and increase the recovery factor. The advancement of digital technologies, can leverage in to the E&P which allows industry to plan in the context of offset knowledge gained through operational recording and learning to deliver a continuous improvement cycle.

Plan a WELL in the current digital cognitive environment, will give an opportunity to allow the past data and inputs that provided, and then execute that plan through orchestration and automation, while constantly recording all the data. With the smart rigs have a sensor layout that records almost everything that takes place on the rig. That data then becomes integrated into the learning process.
When the next well is being designed, it can take advantage of the data from the last well because the cycle time to do this can be almost in real time with the latest digital technologies.

Importance of abstraction

Lets look at process of drilling a well in a much more holistic and disruptive way, by doing so we need to challenge the performance benchmarks, come out from traditional data silos mindset, (there isn’t always a single plan being used by everyone at the rig, hence different perceptions of risk and contingencies) and aligned with the ultimate objective.Look at it from a basic process of drilling a well, then identify the multi disciplinary abstract tasks for the ultimate objective. An abstract representation can be constructed from a concrete one by ignoring details and including only those aspects of primary importance.

Lets take a look at standard land rig operation, which consists of Crews(actors), tasks, process …etc. and see who are really contributing(abstract tasks) to our ultimate goal which is drilling a well. Once we have the tasks/process identified, and plotted in a paper against the actors(crews), then compose it as initial state, set of goal states, and a set of actions. Each action description includes set of pre-conditions and set of post conditions, then this search space can be modeled as a graph, where nodes corresponds to states and arcs corresponds to actions. So that we can construct a PLAN which is nothing but a path, that is a sequence of states together with the arcs linking them.

Abstraction in Planning

What is AI Planning ?

Planning is the explicit deliberation process that chooses and organizes actions by anticipating their outcomes and that aims at achieving some pre-stated objectives. Since planning is an important component of intelligent behavior of human, in-order to build intelligent entity such as robot or intelligent machines, we need to use (or build) a planning software for choosing and organizing actions.

Input to the planner is the Description of state which is an abstraction(which leaves many details about real world, which makes planning possible) of the real world.The controller will make sure “plan supervision” nothing but detect when observation differ from expected results. So it expects the world to be in a certain state as a result of an action. If it observe that it is not in the expected state, then it will do a plan revision. In the event of plan revision, it takes an existing plan and try to change it in some way to take into account the new state. This can be done by the controller for very simple cases or it has to be done by the planner for more complex cases.For more complex case the controller has to pass an execution status back to the planner, so that the planner can generate a new plan that again get passed to the controller. In the worst case, the planner will have to re-plan which means, it will have to create a completely new plan from scratch for the given problem. Dynamic planning then, closes the loop between the planner and execution by passing back the execution status to the planner for re-planning.

In planning, abstraction is often associated to a transformation of the problem representation that allows a problem to be solved more easily, i.e with a reduced computational effort. However, ignoring the wrong details can lead into building plans that do not work, or into costly backtracking and re-planning once overlooked inter-dependencies come to light.

What’s NEXT

In the coming post, lets take a look at standard land rig operation, which consists of Crews(actors), tasks, process …etc. to our ultimate goal which is drilling a well.

References: SPE-184320-MS – Application of Artificial Intelligence Techniques in Drilling System Design and Operations: A State of the Art Review and Future Research Pathways