Refine
Has Fulltext
- yes (71) (remove)
Document Type
- Conference Proceeding (67)
- Article (4)
Institute
- In Zusammenarbeit mit der Bauhaus-Universität Weimar (50)
- Professur Informatik im Bauwesen (10)
- Institut für Strukturmechanik (ISM) (4)
- Professur Theorie und Geschichte der modernen Architektur (2)
- Institut für Bauinformatik, Mathematik und Bauphysik (IBMB) (1)
- Institut für Konstruktiven Ingenieurbau (IKI) (1)
- Junior-Professur Computational Architecture (1)
- Professur Bodenmechanik (1)
- Professur Tragwerkslehre (1)
Keywords
- CAD (71) (remove)
It's not uncommon that analysis and simulation methods are used mainly to evaluate finished designs and to proof their quality. Whereas the potential of such methods is to lead or control a design process from the beginning on. Therefore, we introduce a design method that move away from a “what-if” forecasting philosophy and increase the focus on backcasting approaches. We use the power of computation by combining sophisticated methods to generate design with analysis methods to close the gap between analysis and synthesis of designs. For the development of a future-oriented computational design support we need to be aware of the human designer’s role. A productive combination of the excellence of human cognition with the power of modern computing technology is needed. We call this approach “cognitive design computing”. The computational part aim to mimic the way a designer’s brain works by combining state-of-the-art optimization and machine learning approaches with available simulation methods. The cognition part respects the complex nature of design problems by the provision of models for human-computation interaction. This means that a design problem is distributed between computer and designer. In the context of the conference slogan “back to command”, we ask how we may imagine the command over a cognitive design computing system. We expect that designers will need to let go control of some parts of the design process to machines, but in exchange they will get a new powerful command on complex computing processes. This means that designers have to explore the potentials of their role as commanders of partially automated design processes. In this contribution we describe an approach for the development of a future cognitive design computing system with the focus on urban design issues. The aim of this system is to enable an urban planner to treat a planning problem as a backcasting problem by defining what performance a design solution should achieve and to automatically query or generate a set of best possible solutions. This kind of computational planning process offers proof that the designer meets the original explicitly defined design requirements. A key way in which digital tools can support designers is by generating design proposals. Evolutionary multi-criteria optimization methods allow us to explore a multi-dimensional design space and provide a basis for the designer to evaluate contradicting requirements: a task urban planners are faced with frequently. We also reflect why designers will give more and more control to machines. Therefore, we investigate first approaches learn how designers use computational design support systems in combination with manual design strategies to deal with urban design problems by employing machine learning methods. By observing how designers work, it is possible to derive more complex artificial solution strategies that can help computers make better suggestions in the future.
Interactive visualization based on 3D computer graphics nowadays is an indispensable part of any simulation software used in engineering. Nevertheless, the implementation of such visualization software components is often avoided in research projects because it is a challenging and potentially time consuming task. In this contribution, a novel Java framework for the interactive visualization of engineering models is introduced. It supports the task of implementing engineering visualization software by providing adequate program logic as well as high level classes for the visual representation of entities typical for engineering models. The presented framework is built on top of the open source visualization toolkit VTK. In VTK, a visualization model is established by connecting several filter objects in a so called visualization pipeline. Although designing and implementing a good pipeline layout is demanding, VTK does not support the reuse of pipeline layouts directly. Our framework tailors VTK to engineering applications on two levels. On the first level it adds new – engineering model specific – filter classes to VTK. On the second level, ready made pipeline layouts for certain aspects of engineering models are provided. For instance there is a pipeline class for one-dimensional elements like trusses and beams that is capable of showing the elements along with deformations and member forces. In order to facilitate the implementation of a graphical user interface (GUI) for each pipeline class, there exists a reusable Java Swing GUI component that allows the user to configure the appearance of the visualization model. Because of the flexible structure, the framework can be easily adapted and extended to new problem domains. Currently it is used in (i) an object-oriented p-version finite element code for design optimization, (ii) an agent based monitoring system for dam structures and (iii) the simulation of destruction processes by controlled explosives based on multibody dynamics. Application examples from all three domains illustrates that the approach presented is powerful as well as versatile.
In distributed project organisations and collaboration there is a need for integrating unstructured self-contained text information with structured project data. We consider this a process of text integration in which various text technologies can be used to externalise text content and consolidate it into structured information or flexibly interlink it with corresponding information bases. However, the effectiveness of text technologies and the potentials of text integration greatly vary with the type of documents, the project setup and the available background knowledge. The goal of our research is to establish text technologies within collaboration environments to allow for (a) flexibly combining appropriate text and data management technologies, (b) utilising available context information and (c) the sharing of text information in accordance to the most critical integration tasks. A particular focus is on Semantic Service Environments that leverage on Web service and Semantic Web technologies and adequately support the required systems integration and parallel processing of semi-structured and structured information. The paper presents an architecture for text integration that extends Semantic Service Environments with two types of integration services. Backbone to the Information Resource Sharing and Integration Service is a shared environment ontology that consolidates information on the project context and the available model, text and general linguistic resources. It also allows for the configuration of Semantic Text Analysis and Annotation Services to analyse the text documents as well as for capturing the discovered text information and sharing it through semantic notification and retrieval engines. A particular focus of the paper is the definition of the overall integration process configuring a complementary set of analyses and information sharing components.
The paper proposes a new method for general 3D measurement and 3D point reconstruction. Looking at its features, the method explicitly aims at practical applications. These features especially cover low technical expenses and minimal user interaction, a clear problem separation into steps that are solved by simple mathematical methods (direct, stable and optimal with respect to least error squares), and scalability. The method expects the internal and radial distortion parameters of the used camera(s) as inputs, and a plane quadrangle with known geometry within the scene. At first, for each single picture the 3D position of the reference quadrangle (with respect to each camera coordinate frame) is calculated. These 3D reconstructions of the reference quadrangle are then used to yield the relative external parameters of each camera regarding the first one. With known external parameters, triangulation is finally possible. The differences from other known procedures are outlined, paying attention to the stable mathematical methods (no usage of nonlinear optimization) and the low user interaction with good results at the same time.
We establish the basis of a discrete function theory starting with a Fischer decomposition for difference Dirac operators. Discrete versions of homogeneous polynomials, Euler and Gamma operators are obtained. As a consequence we obtain a Fischer decomposition for the discrete Laplacian. For the sake of simplicity we consider in the first part only Dirac operators which contain only forward or backward finite differences. Of course, these Dirac operators do not factorize the classic discrete Laplacian. Therefore, we will consider a different definition of a difference Dirac operator in the quaternionic case which do factorizes the discrete Laplacian.
Car following models are used to describe the behavior of a number of cars on the road dependent on the distance to the car in front. We introduce a system of ordinary differential equations and perform a theoretical and numerical analysis in order to find solutions that reflect various traffic situations. We present three different variations of the model motivated by reality.
HYPERMONOGENIC POLYNOMIALS
(2006)
It is well know that the power function is not monogenic. There are basically two ways to include the power function into the set of solutions: The hypermonogenic functions or holomorphic Cliffordian functions. L. Pernas has found out the dimension of the space of homogenous holomorphic Cliffordian polynomials of degree m, but his approach did not include a basis. It is known that the hypermonogenic functions are included in the space of holomorphic Cliffordian functions. As our main result we show that we can construct a basis for the right module of homogeneous holomorphic Cliffordian polynomials of degree m using hypermonogenic polynomials and their derivatives. To that end we first recall the function spaces of monogenic, hypermonogenic and holomorphic Cliffordian functions and give the results needed in the proof of our main theorem. We list some basic polynomials and their properties for the various function spaces. In particular, we consider recursive formulas, rules of differentiation and properties of linear independency for the polynomials.
Analysis of the reinforced concrete chimney geometry changes and their influence on the stresses in the chimney mantle was made. All the changes were introduced to a model chimney and compared. Relations between the stresses in the mantle of the chimney and the deformations determined by the change of the chimney's vertical axis geometry were investigated. The vertical axis of chimney was described by linear function (corresponding to the real rotation of the chimney together with the foundation), and by parabolic function (corresponding to the real dislocation of the chimney under the influence of the horizontal forces - wind). The positive stress pattern in the concrete as well as the negative stress pattern in the reinforcing steel have been presented. The two cases were compared. Analysis of the stress changes in the chimney mantle depending on the modification in the thickness of the mantle (the thickness of the chimney mantle was altered in the linear or the abrupt way) was carried out. The relation between the stresses and the chimney's diameter change from the bottom to the top of the chimney was investigated. All the analyses were conducted by means of a specially developed computer program created in Mathematica environment. The program makes it also possible to control calculations and to visualize the results of the calculations at every stage of the calculation process.
VARIATION OF ROTATIONAL RESTRAINT IN GRID DECK CONNECTION DUE TO CORROSION DAMAGE AND STRENGTHENING
(2006)
The approach to assessment of rotational restraint of stringer-to-crossbeam connection in a deck of 100-year old steel truss bridge is presented. Sensitivity of rotational restraint coefficient of the connection to corrosion damage and strengthening is analyzed. Two criteria of the assessment of the rotational restraint coefficient are applied: static and kinematic one. The former is based on bending moment distribution in the considered member, the latter one – on the member rotation at the given joint. 2D-element model of finite element method is described: webs and flanges are modeled with shell elements, while rivets in the connection – with system of beam and spring elements. The method of rivet modeling is verified by T-stub connection test results published in literature. FEM analyses proved that recorded extent of corrosion damage does not alter the initial rotational restraint of stringer-to-crossbeam connection. Strengthening of stringer midspan influences midspan bending moment and stringer end rotation in a different way. Usually restoring member load bearing capacity means strengthening its critical regions (where the highest stress levels occur). This alters flexural stiffness distribution over member length and influences rotational restraint at its connection to other members. The impact depends on criterion chosen for rotational restraint coefficient assessment.
Traffic simulation is a valuable tool for the design and evaluation of road networks. Over the years, the level of detail to which urban and freeway traffic can be simulated has increased steadily, shifting from a merely qualitative macroscopic perspective to a very detailed microscopic view, where the behavior of individual vehicles is emulated realistically. With the improvement of behavioral models, however, the computational complexity has also steadily increased, as more and more aspects of real-life traffic have to be considered by the simulation environment. Despite the constant increase in computing power of modern personal computers, microscopic simulation stays computationally expensive, limiting the maximum network size than can be simulated on a single-processor computer in reasonable time. Parallelization can distribute the computing load from a single computer system to a cluster of several computing nodes. To this end, the exisiting simulation framework had to be adapted to allow for a distributed approach. As the simulation is ultimately targeted to be executed in real-time, incorporating real traffic data, only a spatial partition of the simulation was considered, meaning the road network has to be partitioned into subnets of comparable complexity, to ensure a homogenous load balancing. The partition process must also ensure, that the division between subnets does only occur in regions, where no strong interaction between the separated road segments occurs (i.e. not in the direct vicinity of junctions). In this paper, we describe a new microscopic reasoning voting strategy, and discuss in how far the increasing computational costs of these more complex behaviors lend themselves to a parallelized approach. We show the parallel architecture employed, the communication between computing units using MPIJava, and the benefits and pitfalls of adapting a single computer application to be used on a multi-node computing cluster.