In the present work the applicability and the improvement of the detection sensitivity of a particle mass spectrometer (PMS) were studied. The PMS was originally used by Hospital and Roth for the investigation of the growth of soot-particles produced in combustion processes. It consists of a molecular beam sampling system and a high vacuum chamber which contains the necessary measurement instrumentation for the in-situ particle sizing.<p><p>Three reactors were employed to investigate the applicability: a low-pressure flame reactor, a microwave reactor and a spark reactor. The reactors are different in respect of the energy transfer into the reactive gases and the bulk material respectively. The flame reactor was used to produce well-known soot particles as reference particles for the experiments related to the improvement of the detection sensitivity of the PMS. Soot-, Silver-, Carbon-, Siliconoxide-, Tinoxide and Nickel- Nanoparticles were produced. The results have shown that the applicability of the PMS could be extended and be applied to particles generated in the microwave reactor as well as in the spark reactor.<p><p>Neutral charged Soot- and Tinoxid-Nanoparticles which were produced in the flame reactor were charged by electron impact ionisation in the high vacuum chamber of the PMS. The charging effect increases with decreasing amount of charged particles in the aerosol. Thus, in the case of neutral or weakly charged aerosols the particles could be characterised with the PMS leading to a further extension of the application range.<p><p>A new detector was developed to improve the detection sensitivity of the PMS. The detector consists of a secondary electron multiplier (SEM) and an external conversion dynode. The dynode was set to a high electrical potential. Thus, the particles were accelerated to a velocity which is sufficient to release secondary electrons or ions from the dynode after the impaction of the particles with the surface of the dynode. Afterwards these secondary electrons and ions respectively were amplified by the SEM. In comparison with the conventional detector, a Faraday-Cup, an increase of the signals of about 10⁵ was achieved with this new detector.

Modern inland vessels are open-top double-hull ships with an unusually large aspect ratio, a shallow draught and an extremely long cargo hold. These ships have a very low bending and torsional rigidity. Due to minor collisions, grounding in shallow water, corrosion and fatigue the state of the ship structure changes appreciably in course of time. These imperfections reduce the stability and strength of the structure. The safety against total collapse decreases. This thesis is showing the results of a systematic investigation of typical ship-constructions with the Finite-Ele-ment-Method (FEM). The failures of a complex structure like a ship hull is not entails by one item. Usually more than one negative factors are responsible. For example these are an unfavourable cargo distribution, the reduction of the plate thickness and the large imperfections during ser-vice. Normally ships are construct for a relative long working season (20-30 years and more). In course of this long term the features of the structure are changing considerably. Already the dimensions of new components must take care of it. As an example, the traditional calculations methods recognize an extra charge for the corrosion. To investigate the structure stability it is very difficult to take into account the imperfections (e.g. pre-deformations) with analytical calculation methods. This is a typical application for the FEM. The introduction part of the thesis will showing different possibilities to calculate the structural stability and strength of an inland vessel. On account to the points mentioned above (imperfections, load distributions) the traditional calculation methods are not sufficient. Using the Finite-Element-Method (FEM) a new calculation concept will be introducing to achieve these elaborate analyses. In future the influence of local damages to the global structure should be a minimum. The heart of the procedure is the registration and numerical consideration of typical damages. It is to investigate which imperfections are important to consider and which are negligible. A classification of the imperfections will support this. To consider these imperfections the mentioned calculation concept will be used. First, the simple example of a stiffened plate and a box girder will be using to discuss the influence of imperfections on the structure behaviour. Later this knowledge of these investigations will be utilizing for the analysis of typical inland vessels. The FE-models of a pusher barge and a modern inland container vessel will be introducing and their calculation results will be showing. These are deformations and stresses for various cargo distributions. Additionally the non-linear results for extreme loads will be presenting to simulate the collapse behaviour of the vessel. At least constructive modifications (e.g. longitudinal girder, high-tensile steel) will be discussing to show their effects. With this, the new ship should be lightweight, easy to build and with a maximum security against global breakdown. The last point is a short discussion of a few "special features" for future investigations. For example these are the application of multi-processor software (to reduce the calculation time) and the using of optimizing software

From hazardous waste incineration (HWI) in rotary kilns ashes are generated as solid residues. In this thesis the effect of limestone addition to HWI-ashes was examined for the elution behaviour of heavy metals after a temperature treatment. Experimental results were compared to theoretic modelling. After treatment at 1200°C, the typical temperature for hazardous waste incineration, a minimum of heavy metal leaching from HWI-ashes was found at CaO-contents between 15 and 20%. On the one hand, lower melting temperatures of the ashes were measured after limestone addition. The melting process may lead to immobilisation of heavy metals in glassy phases of the ash. On the other hand, thermodynamic equilibrium calculations prove the decrease of the solubility of heavy metals as for example zinc, lead and chromium due to a changed chemical speciation. In future a HWI-ash optimised with regard to heavy metal leaching could be disposed off to lower landfilling costs, so that the incineration costs of hazardous waste in rotary kilns may also decrease.

Based on the problem of a total evaluation of skip steel cables of blast furnaces two computer aided procedures for the analysis of the measured data of electromagnetic steel cable testing are developed. The first procedure is useful to detect inner and outer broken wires and simulates the analysis of a skilled inspector by a pattern classification algorithm using fuzzy-logic. Like an inspector this system has to be trained. This is implemented with the help of a genetic algorithm. The second procedure is helpful serves as evaluation of accumulated inner faults of steel cables. Therefore two characteristic values describing the inner condition are determined by frequency analysis of the measured data. - Ausgehend von der Problemstellung einer umfassenden Beurteilung des Schädigungszustandes von Schrägaufzugseilen an Hochöfen werden zwei rechnergestützte Auswerteverfahren für die magnetinduktive Streuflußprüfung von Drahtseilen entwickelt. Das erste Verfahren dient der Erkennung innerer und äußerer Drahtbrüche und bildet die Auswertemethodik eines erfahrenen Inspekteurs in einem Musterklassifikator basierend auf Fuzzy-Logic nach. Den Erfahrungsprozeß des Inspekteurs vollzieht das Verfahren durch eine Optimierung mittels genetischen Algorithmus anhand realer Meßdaten nach. Das zweite Verfahren dient der Abschätzung des inneren Schädigungszustandes eines Drahtseiles hervorgerufen durch Schadenshäufungen. Dazu erfolgt die Berechnung zweier charakteristischer Kennwerte aus eine Frequenzanalyse der aufgezeichneten Meßdaten.

Fine solid particles, especially nano-particles, are of great interest for the synthesis of new materials. There is also a strong demand for newer and more precise contact technologies of electronic components due to the higher getting integration of electronic devices. These new technologies may be based on submicron conducting particles, e.g. for producing solder pastes. In the present thesis a process was developed for the production of micron and submicron conducting particles by the use of electrostatic atomisation of liquid metals. The atomization process is based on the instability of conducting liquids in a strong inhomogenous electric field. The free liquid surface is transformed into a cone shape, from where a thin liquid jet is emitted. This jet breaks up into a spray of fine unipolary charged droplets immediately after the emission. Under defined circumstances the droplets have a narrow size distribution and can be extracted for other applications without additional classification methods. The rapid solidification leads to defined spherical particles, depending on the used metal, and the unipolar charge to a low agglomeration. The high electrical field, which is necessary for the atomization, causes electrical discharges of the ambient gas and disturbes the atomization process. To reduce these discharges a special pressure chamber was constructed, which is filled with quenching gas as SF6 with over-pressures of up to 12 bar. The electrostatic atomisation was applicated for liquid metals only in an ultra high vacuum up to now. The production of aerosol particles may have advantages not only due to available cheaper online measurement techniques. By the use of aerosol measurement techniques a direct statement about the particle size, the particle number concentration and the mean charge on the particles is possible. The atomization was investigated for metals as well as for eutectic alloys with a low melting point and a low surface tension. The results show, that spherical particles with a diameter down to a few hundred nanometer could be produced. In addition the production of thin and homogenous granular films was possible. To compare and verify the experimental results a simplified numerical model of the spraying process with respect to the space charge was developed. By calculating the particle trajectories from the emission source to the collector plate an estimation of the spray cloud geometry and the emitted electrical current was possible. An interesting observation was the self induced size segregation of the particles, which occurred in the experiments and in the numerical simulations. This effect can be used for a defined expansion or focussing of the deposition area on the target. Here there may be found some interesting applications for the point spraying.

The interest in nano particles from the view of basic and applied research has constantly risen in recent years. Since their physical properties are different from those of bulk materials, new industrial applications are emerging, e.g. in microelectronics. For quality control, particle production technologies call for on-line techniques that respond to the chemical nature of the particles. A particle generation process usually leads to particles with different surface properties. These surface properties, i.e. impurities ( like an oxygen layer ) or the crystal habit of single particles, determines much of their behaviour, especially in the diameter range of a few nanometers. While this fact has hardly been recognized, we anticipate that new technologies will establish a demand not only for analysis but also for separation of chemically pure particles with respect to their surface properties. In these PHD thesis we introduce a method and a device to separate aerosol particles according to their photoelectric properties. These strongly depend on particle surface composition as well as the particle morphology.

In order to meet the cost targets of PEM fuel cells for commercialization significant cost reductions of cell stack components like membrane/electrode assemblies and bipolar plates have become key aspects of research and development. Central topics of his work are the bipolar plates and humidification for portable applications. Best results concerning conductivity of an extensive screening of a variety of carbon polymer compounds with polypropylene as matrix could be achieved with the carbon black/graphite/polypropylene-base system. Successful tests of this material in a fuel cell stack could be performed as well as the proof of suitability concerning material- and manufacturing costs. Dependent on application a decrease of material cost to 2 ¤/kg to 1,8 ¤/kW seems to be possible. Finally bipolar plates consisting of a selected carbon polymer compound were successfully integrated and tested in a 20-cell stack which was implemented in a portable PEFC-demonstrator unit with a power output between 50 and 150 W.

Der Beitrag stellt die Modellierung eines virtuellen Kraftsensors mittels neuronaler Netze vor. Ausgangspunkt ist ein hydraulisch angetriebener zweiachsiger Roboter mit elastischem Arm, dessen Strukturschwingungen durch Regelung des Aktuators gedämpft werden. Dieses Regelungskonzept erfordert jedoch den Einsatz eines aufwendigen Sensors zur Messung der auf den Aktuator wirkenden Kraft. Auf Basis des mehrschichtigen Perzeptrons wird ein Black-Box-Modell identifiziert, dass im Eingang nur Prozessgrößen verwendet, die von ohnehin vorhandener Sensorik gemessen werden, und die gesuchte Kraft im Sinne eines virtuellen Sensors schätzt. Die Validierung des Modells erfolgt am realen System und schließt einen Vergleich mit bereits bestehenden Ansätzen ein.

This paper presents the nonlinear black box modeling of a hydraulic translatory drive using neural networks. The type of neural network employed here is the multilayer perceptron. Feeding previous inputs and outputs into the network leads to two different black box model structures, namely the series-parallel and the parallel model. Their suitability for modeling the hydraulic drive on the basis of measurements on a test bed is compared.

Dieser Beitrag stellt die Modellierung eines pneumatischen Antriebs mit Black-Box-Methoden vor, wobei eine qualitative Analyse des Systems zu einer detaillierteren Modellstruktur und damit höheren Güte beiträgt. Als Black-Box-Strukturen werden zum einen Fuzzy-Netzwerke vom Takagi-Sugeno-Typ mit mehrdimensionalen Zugehörigkeitsfunktionen und zum anderen neuronale Netzwerke in Form des mehrschichtigen Perzeptrons verwendet. Beide Ansätze gestatten die Identifikation nichtlinearer dynamischer Ein-Ausgangs-Modelle und werden hinsichtlich ihrer Approximationseigenschaften in diesem Beitrag miteinander verglichen.