Introduction To The Perspective
Approach
|
|
|
To Design, plan and execute an AEC project requires the expertise of hundreds, and sometimes thousands of AEC professionals. While the end result is one project, these professionals have task-specific expertise, and they conceptualize the project in ways that are suited to their specific tasks. Therefore the work of these professionals must somehow be integrated as the project unfolds through time. Understanding the characteristics of AEC projects is essential to any effort to integrate information on these projects. This research presents industry test cases that illustrate that architects, engineers, contractors, fabricators, and other AEC professionals use task-specific views to design, plan, and execute AEC projects. These views contain the information they need, structured in a way that is suited to their task. Today, they often construct these views based on information in other engineers’ views. These dependent views often become source views of other dependent views: A network of implicit dependencies between task-specific views emerges as the design progresses. The process is iterative, when source views are modified, dependent views must often be re-integrated. This research focuses on the construction and integration of geometric views, because geometry is a primary language of communication and coordination on AEC projects today. Traditionally, engineers have constructed and integrated these geometric views manually, using pencils and more recently computer aided drafting (CAD). Manually constructing and integrating views is often difficult, time-consuming, and error-prone. Newer approaches develop central or federated project models that predetermine the potential views for a project or for an entire industry; but these approaches are proving difficult to implement on multi-disciplinary, constructive, iterative, and unique AEC projects. While points of departure, current project modeling approaches lack simple formal methods that enable engineers to construct a new dependent geometric view from information in one or many source geometric views and control the integration of these views as the project progresses. As a result, engineers must manually construct and integrate many of the geometric views they require today. To address this need, this research asks three related research questions: 1. What is a simple, generic geometric engineering view
that formalizes its dependency on other geometric views? Engineers need task-specific views to do their work, and these views depend on information in other views. CAD tools today contain generic geometric views, such as layers (a collection of features), or assemblies (collections of features) that enable engineers to construct task-specific geometric views. However, these views do not explicitly formalize the existence, status, and nature of their dependency on other engineer’s views. Therefore the integration between views cannot be automated or formally managed. Parametric CAD tools enable engineers to specify the dependency between concepts in a model. For example an engineer can specify that a slab be placed with respect to a particular beam that supports it. However, current parametric approaches are not designed to explicitly formalize the dependency between views. Engineers could benefit from a simple generic geometric view that specifies the existence, status, and nature of its dependencies on other engineering views. 2. What is a simple generic method that engineers can
use to construct a dependent geometric view from source views? Because AEC projects are unique, engineers need to be able to specify the nature of the dependencies between views. These engineers are not computer programmers; they need a simple, generic approach that they can use to formulate the reasoning that embodies the nature of this dependency, such that this reasoning constructs the dependent view from source views. Query languages contain some numerical and textual reasoning that engineers can use to construct dependent views, however these languages lack many of the geometric operators that engineers need, and are not very intuitive for engineers to use. Parametric approaches provide tools that enable engineers to specify and control the nature of the geometric dependency between features in a model. However current parametric approaches do not contain explicit mechanisms that enable an engineer to formally construct a view from other views. 3. What is a simple generic process that engineers
can use to control the iterative construction of views and their
dependencies? Because AEC projects are iterative, engineers need simple, generic, methods to control the integration of their views with respect to changing source views. These methods should enable an engineer to know the status of a view with respect to its source views, and integrate a view with source views as needed. These processes need to be designed to work with the representation and reasoning identified in the first two research questions. This thesis addresses these research questions by formalizing the Perspective Approach, which empowers engineers to formally construct and integrate a task-specific view, called a Perspective by easily specifying its dependency on other Perspectives. A project model emerges as a directed acyclic graph of Perspectives and dependencies. The Perspective Approach is designed to provide an intuitive yet formal way to conceptualize and construct a multi-disciplinary design that maps closely to the way engineers on AEC projects work today, and provide these engineers with tools to engage in formal, automated, and integrated design and analysis. Chapters 2, 3 and 4 together motivate, formalize and validate the Perspective Approach. Chapter 2: Engineering Test Cases To Motivate The Formalization Of An AEC Project Model As A Directed Acyclic Graph Of Views And Dependencies, motivates the Perspective Approach by presenting industry test cases that illustrate the multi-disciplinary, constructive, iterative, and unique character of AEC projects. These test cases show that, to perform their tasks on these projects, AEC engineers construct task-specific engineering views from information in other engineers’ views. These engineers have difficulty constructing and integrating task-specific views on these projects today. Based on these observations, the chapter proposes that engineers could benefit from an approach that provides them with simple, formal methods to iteratively construct a task-specific view from other views as needed, and control the integration of these views as the project progresses. A project model emerges as a directed acyclic graph (d.a.g.) of task-specific views and dependencies (Figure 1). The chapter defines requirements for such an approach, and discusses current project modeling approaches in terms of these requirements, concluding that, while points of departure, current AEC project modeling approaches do not explicitly formalize a project model in this way.
Chapter 2 argues that AEC project information models could be formalized as a directed acyclic graph of views and dependencies. The lines are dashed because, as practiced today, the dependencies between views are often implicit. Figure 1: The dependency between views. Chapter 3: The Perspective Approach: Enabling Engineers to Construct and Integrate Geometric Views and Generate an Evolving Project Model, formalizes the Perspective Approach, in which engineers from multiple disciplines can iteratively construct geometric engineering views, called geometric Perspectives, from other Perspectives and control the integration of this multi-disciplinary, evolving Project Model as the project progresses. Geometric Perspectives are like most CAD layers that engineers construct today; they contain Features that describe the project for a specific task. Features can be named, and they can contain geometric data types and relationships to other Features in other Perspectives. What makes a Perspective different from a CAD layer is that it formalizes the existence, status, and nature of its dependency on other Perspectives. The chapter also formalizes simple Management Processes that enable engineers to easily control the integration of their Perspectives with the Perspectives on which they depend. The chapter describes an implementation of a prototype of the Perspective Approach, called PerspectorApp, and retrospectively validates this approach on two test cases from the design and construction of the Walt Disney Concert Hall. The validation shows how the Perspective Approach could enable engineers from multiple disciplines to construct and control the integration of task-specific Perspectives on an evolving AEC project more quickly, accurately, and simply than current practice allows. Perspectives address the first research question and the Management Processes address the third research question. Chapter 4: Perspectors: Composable, Reusable Reasoning Modules To Construct A Geometric Engineering View From Other Geometric Engineering Views, addresses the second research question by formalizing composable, subsumable, modifiable reasoning modules, called geometric Perspectors, which engineers can use to specify the nature of the dependency of a Perspective on other Perspectives. Because every Perspective has one associated Perspector, together they can be composed into Perspector Graphs to specify complex transformations of source Perspectives into a dependent Perspective. Each Perspector can formalize an automated transformation, or the Perspector algorithm can simply provide CAD tools to an engineer to perform the needed transformation. The focus of this research is on automated Perspectors, but manual Perspectors can also be easily incorporated. Because a Perspector’s input and output are both Perspectives, a Perspector Graph (and its associated Perspectives) can be subsumed into one Perspector to represent higher-level transformations. To enable greater reuse of Perspectors, a Perspective can also specify Perspector Parameters that its Perspector uses when constructing its Features. Perspectors become a formal specification that produces formal and therefore repeatable results. Using PerspectorApp, the chapter gives empirical evidence that engineers can select from a potentially small number of predefined, reusable Perspectors and easily compose them into Perspector Graphs to construct useful dependent geometric Perspectives more quickly and accurately than current practice and theory allows, and may enable engineers to engage in novel automated design and analysis.
Chapter 3 formalizes the representation of a project model as a directed acyclic graph of Perspectives and dependencies, to satisfy the first research question, and the management processes to control the iterative construction of this project model, to satisfy the third research question. Chapter 4 formalizes the composable, subsumable, modifiable reasoning mechanisms, called Perspectors, which engineers use to specify the nature of the dependency between Perspectives, to satisfy the second research question. Figure 2: The Perspective Approach. The contributions of this thesis correspond to the research questions. They are the representation, reasoning and management that make up the Perspective Approach. Specifically the research claims as contributions: 1. Perspective – a generic geometric view that formalizes the existence, status and nature of its dependency on other Perspectives. 2. Perspector – a generic reasoning mechanism that constructs information in a dependent Perspective based on information in the source Perspective 3. Management Processes – that specify how an engineer can easily construct and control the integration of Perspectives. The Perspective Approach provides a simple, flexible way to conceptualize and construct integrated project models by formalizing simple, general tools that enable engineers to construct task specific views from other engineer’s views, and control the integration of these views as the project progresses. An integrated Project Model emerges as the engineers iteratively construct and integrate their task-specific Perspectives. This research focuses on the construction and integration of geometric Perspectives. However, the general ideas of Perspective, Perspector, and Management Processes could be applied more generally to provide a general framework to integrate many task-specific views of AEC projects, or any endeavor where multiple disciplines require task-specific views that are constructed and integrated from information in other views. Empowering engineers to specify the automated construction of dependent views, and enabling them to control their integration is intended to facilitate more efficient, formal design exploration, iteration, and execution. |
|
