scholarly journals Initial User Experience With an Artificial Intelligence Program for the Preliminary Design of Centrifugal Compressors

Author(s):  
Carol J. Russo ◽  
Dennis J. Nicklaus ◽  
Siu S. Tong

A new approach is evaluated for the design of turbomachinery components using existing analysis codes coupled to a generic Artificial Intelligence (AI) software framework called ENGINEOUS. This AI framework uses intelligent search techniques with a small set of basic component design rules to iterate to an optimized solution and to quantify parameter trade-offs. Initial experience with ENGINEOUS indicates that it is a powerful design tool which quickly identifies non-obvious solutions balanced for conflicting multiple goals in a small number of iterations which vary linearly with the number of variables. The solution path and driving logic are easily visible to the designer and a parameter study option can rapidly quantify potential design trade-offs which together allow a critique of the selected design to balance performance against development risks. Because this AI design approach fosters intelligent interface with the designer and is generic, the potential application areas and productivity benefits appear enormous.

Author(s):  
А.И. Гайкович ◽  
С.И. Лукин ◽  
О.Я. Тимофеев

Процесс создания проекта судна или корабля рассматривается как преобразование информации, содержащейся в техническом задании на проектирование, нормативных документах и знаниях проектанта, в информацию, объем которой позволяет реализовать проект. Проектирование может быть представлено как поиск решения в пространстве задач. Построение цепочки последовательно решаемых задач составляет методику проектирования. Проектные задачи могут быть разбиты на две группы. Первая группа ‒ это полностью формализуемые задачи, для решения которых есть известные алгоритмы. Например, построение теоретического чертежа по известным главным размерениям и коэффициентам формы. Ко второй группе задач можно отнести трудно формализуемые или неформализуемые задачи. Например, к задачам этого типа можно отнести разработку общего расположения корабля. Важнейшим инструментом проектирования современного корабля или судна является система ав­томатизированного проектирования (САПР). Решение САПР задач первой группы не представляет проблемы. Введение в состав САПР задач второй группы подразумевает разработку специального ма­тематического аппарата, базой для которого, которым является искусственный интеллект, использующий теорию нечетких множеств. Однако, настройка искусственных нейронных сетей, создание шкал для функций принадлежности элементов нечетких множеств и функций предпочтений лица принимающего решения, требует участие человека. Таким образом, указанные элементы искусственного интеллекта фиксируют качества проек­танта как специалиста и создают его виртуальный портрет. The process of design a project of a ship is considered as the transformation of information contained in the design specification, regulatory documents and the designer's knowledge into information, the volume of which allows the project to be implemented. Designing can be represented as a search for a solution in the space of problems. The construction of a chain of sequentially solved tasks constitutes the design methodology. Design problems can be divided into two groups. The first group is completely formalizable tasks, for the solution of which there are known algorithms. For example, the construction of ship's surface by known main dimensions and shape coefficients. Tasks of the second group may in­clude those which are difficult to formalize or non-formalizable. For example, tasks of this type can include develop­ment of general arrangement of a ship. The most important design tool of a modern ship or vessel is a computer-aided design system (CAD). The solu­tion of CAD problems of the first group is not a problem. Introduction of tasks of the second group into CAD implies development of a special mathematical apparatus, the basis for which is artificial intelligence, which uses the theory of fuzzy sets. However, the adjustment of artificial neural networks, the creation of scales for membership functions of fuzzy sets elements and functions of preferences of decision maker, requires human participation. Thus, the above elements of artificial intelligence fix the qualities of the designer as a specialist and create his virtual portrait.


2021 ◽  
Author(s):  
Jon Gustav Vabø ◽  
Evan Thomas Delaney ◽  
Tom Savel ◽  
Norbert Dolle

Abstract This paper describes the transformational application of Artificial Intelligence (AI) in Equinor's annual well planning and maturation process. Well planning is a complex decision-making process, like many other processes in the industry. There are thousands of choices, conflicting business drivers, lots of uncertainty, and hidden bias. These complexities all add up, which makes good decision making very hard. In this application, AI has been used for automated and unbiased evaluation of the full solution space, with the objective to optimize the selection of drilling campaigns while taking into account complex issues such as anti-collision with existing wells, drilling hazards and trade-offs between cost, value and risk. Designing drillable well trajectories involves a sequence of decisions, which makes the process very suitable for AI algorithms. Different solver architectures, or algorithms, can be used to play this game. This is similar to how companies such as Google-owned DeepMind develop customized solvers for games such as Go and StarCraft. The chosen method is a Tree Search algorithm with an evolutionary layer on top, providing a good balance in terms of performance (i.e., speed) vs. exploration capability (i.e., it looks "wide" in the option space). The algorithm has been deployed in a full stack web-based application that allows users to follow an end-2-end workflow: from defining well trajectory design rules and constraints to running the AI engine and evaluating results to the optimization of multi-well drilling campaigns based on risk, value and cost objectives. The full-size paper describes different Norwegian Continental Shelf (NCS) use cases of this AI assisted well trajectory planning. Results to-date indicate significant CAPEX savings potential and step-change improvements in decision speed (months to days) compared to routine manual workflows. There are very limited real transformative examples of Artificial Intelligence in multi- disciplinary workflows. This paper therefore gives a unique insight how a combination of data science, domain expertise and end user feedback can lead to powerful and transformative AI solutions – implemented at scale within an existing organization.


2018 ◽  
Vol 9 (2) ◽  
pp. 1569-1578 ◽  
Author(s):  
Khaled Z. Abdelgawad ◽  
Mahmoud Elzenary ◽  
Salaheldin Elkatatny ◽  
Mohamed Mahmoud ◽  
Abdulazeez Abdulraheem ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document