Dissertationen zum Thema „Akustické modely“

The aim of the diploma thesis was to create a program with which it will be possible to control a collaborative robot by voice. First chapters contain a search of the current state in the field of collaborative robotics in terms of safety, work efficiency, robot programming and communication with the robot. Furthermore, the issue of machine processing of the human voice is discussed. In practical part was proposed an experiment in which we work with off-line simulation of UR3 robot in PolyScope 3.15.0 software. This simulation was linked to a Python program which uses SpeechRecognition and urx libraries. Simple voice instructions have been designed to move robot to defined position.

This thesis deals with the properties of soundproof partition structures in the low-frequency sound at impact sound insulation and security of acoustic comfort inside buildings. The prediction of impact sound is a simulation laboratory measurements of impact sound. The work is based on the occurrence of frequent complaints of inhabitants of residential homes for subjectively poor impact sound insulation of horizontal dividing structures, whose top layer is formed laminate. Although these structures conform in terms of impact sound insulation in accordance with the requirements of CSN 73 0532: 2010, residents complain about the subjective perception of the sounds of lower frequencies. A noise with a distinctive character of sound at low frequencies has been proved by measuring the spectral analysis and evaluation of sound pressure levels caused by the movement of persons roof construction to the floor. On the measurement and evaluation carried out in accordance with the measurement and evaluation of noise in non-working environment can be related requirement under the Regulation No. 272/2011 Coll. "On the protection of health from the adverse effects of noise and vibration." Occupational noise limits for protected buildings interior space do not apply to noise from ordinary use of the apartment. Under current legislation, the problem is in the Czech Republic at present insoluble. Therefore, this work explores ways evaluation of impact sound and delivery is determined conclusions. With the low-frequency impact sound insulation is also related to the latter part of this dissertation, where the computing program ANSYS (version 14.0) is simulated laboratory measurements of impact sound insulation of the real structure. The paper presents the results of simulation, and the sound pressure level in the receiving room to třetinooktávového band 630 Hz. These results are compared with measured values modeled in the laboratory.

This diploma thesis solves creation of a computing model of sound propagation for determination of acoustic pressure level around multiple point sound source with consideration of sound directivity of each source (compresor and fan). Thesis contains neccesary theoretical knowledge for creation of computing model and basics of sound measuring, theoretical computing model, measurement data of real unit and model configuration for real measured values. Computing code of the model in Visual Basic for MS Excel is a part of the thesis.

An adaptable binaural model for surround sound has been developed in this master’s thesis. The adaptability is based on measurements of the listener’s head. This model is based on what was found to be the best suited material combination of successful models in earlier studies. This includes an ellipsoidal model for interaural time difference, an one-pole, one-zero head shadow filter and the use of Blauert’s directional bands for spectral manipulation. The model can play back six channel surround content using the standardized 5.1 surround sound loudspeaker setup. This standardized loudspeaker placement is used when creating virtual sound sources. Arbitrary sound directions are made in the horizontal plane by creating virtual sound sources using vector base amplitude panning between the standardized loudspeaker positions.To test the performance of this model, a listening test was conducted. The hypothesis tested was that the adaptable model would produce equal or lower localization error, compared to the commercial model. 20 test subjects participated. The test featured three different test types; standardized 5.1 loudspeaker setup, a commercial model for surround sound in headphones, and the adaptable model. Localization accuracy for ten selected directions in the right half plane was tested. The results from the adaptable model were compared to the result of the commercial model. The loudspeaker setup acted as a reference.Mean localization error was found to be thrice as high for the adaptable model, compared to the commercial model. Both models had the same standard deviation. 95% of the confidence intervals for these models did not overlap, i.e. there is a significant difference between the two methods. With this one can safely conclude that the commercial model provided a smaller localization error than the adaptable model. Hence the hypothesis has to be disproved.Both the commercial model and the thesis model performed significantly worse than the loudspeaker setup. One difference between commercial model, and the thesis model, was that that the commercial model had added room reflections and reverberation. This can create the sensation that the sound is coming from outside the head, and make it easier to localize. This contradicts with the knowledge that reverberation diffuses the sound field, making the direct sound that provides the directional information become less prominent.

The goal of this thesis is to design and test techniques for unsupervised adaptation of speech recognizers on some audio data without any textual transcripts. A training set is prepared at first, and a baseline speech recognition system is trained. This sistem is used to transcribe some unseen data. We will experiment with an adaptation data selection process based on some speech transcript quality measurement. The system is re-trained on this new set than, and the accuracy is evaluated. Then we experiment with the amount of adaptation data.

This diploma thesis is aimed to find an effective method for continuous speech recognition. To be more accurate, it uses speech-to-text recognition for a keyword spotting discipline. This solution is able to be applicable for phone calls analysis or for a similar application. Most of the diploma thesis describes and implements speech recognition framework Sphinx-4 which uses Hidden Markov models (HMM) to define a language acoustic models. It is explained how these models can be trained for a new language or for a new language dialect. Finally there is in detail described how to implement the keyword spotting in the Java language.

This thesis is a step towards the long-term and high-reaching objec-tive of building dialogue systems whose behaviour is similar to a human dialogue partner. The aim is not to build a machine with the same conversational skills as a human being, but rather to build a machine that is human enough to encourage users to interact with it accordingly. The behaviours in focus are cue phrases, hesitations and turn-taking cues. These behaviours serve several important communicative functions such as providing feedback and managing turn-taking. Thus, if dialogue systems could use interactional cues similar to those of humans, these systems could be more intuitive to talk to. A major part of this work has been to collect, identify and analyze the target behaviours in human-human interaction in order to gain a better understanding of these phenomena. Another part has been to reproduce these behaviours in a dialogue system context and explore listeners’ perceptions of these phenomena in empirical experiments. The thesis is divided into two parts. The first part serves as an overall background. The issues and motivations of humanlike dialogue systems are discussed. This part also includes an overview of research on human language production and spoken language generation in dialogue systems. The next part presents the data collections, data analyses and empirical experiments that this thesis is concerned with. The first study presented is a listening test that explores human behaviour as a model for dialogue systems. The results show that a version based on human behaviour is rated as more humanlike, polite and intelligent than a constrained version with less variability. Next, the DEAL dia-logue system is introduced. DEAL is used as a platform for the re-search presented in this thesis. The domain of the system is a trade domain and the target audience are second language learners of Swedish who want to practice conversation. Furthermore, a data collection of human-human dialogues in the DEAL domain is presented. Analyses of cue phrases in these data are provided as well as an experimental study of turn-taking cues. The results from the turn-taking experiment indicate that turn-taking cues realized with a di-phone synthesis affect the expectations of a turn change similar to the corresponding human version. Finally, an experimental study that explores the use of talkspurtinitial cue phrases in an incremental version of DEAL is presented. The results show that the incremental version had shorter response times and was rated as more efficient, more polite and better at indicating when to speak than a non-incremental implementation of the same system.
QC20100830

Rotary kilns are industrial furnaces that have been widely used in limestone calcination, cement industry and hazardous waste incineration for centuries. In this work a rotary kiln used for the iron ore pellet sintering process in the grate-kiln pelletizing system has been studied. In order to increase the energy efficiency, a large amount of air is supplied to the kiln through air channels connected to the cooler. This air is necessary for the coal combustion process and the heat transport to the kiln bed. However, the geometry of the kiln hood connecting the air channels and the cooler is complicated. As a consequence, the jet flame is unstable. In order to improve the performance of the jet flame it is therefore necessary to study the kiln aerodynamics to reveal the flow field. Even though it is a complicated problem containing fluid dynamics, combustion and heat and mass transfer, it can be simplified into a down-scaled cold model to make it feasible to understand the flow field both experimentally and numerically. In this work, the whole kiln is generalized as a high Reynolds number turbulent round jet interacting with asymmetric co-flows. With the aid from previous PIV measurement data of a down-scaled water model of the kiln, Computational Fluid Dynamics (CFD) simulations using the commercial code ANSYS CFX 16.0 have been pursued for two main purposes: 1) To find a turbulence model that is computationally inexpensive and able to capture the main features of the mean flow field; 2) With the turbulence model chosen in 1), to study the geometrical effect on the development of the primary jet. In Paper A, three turbulence models were employed, the standard k-epsilon model, a modified k-epsilon model with slightly higher turbulence production and the SSG Reynolds stress model. Wall functions were applied since resolving the viscous wall region was not a concern in this work. It is found that the standard k-epsilon model fit the experimental data best compared to the other two models and that all three turbulence models predict an asymmetric development of the primary jet, especially far downstream, In Paper B, again using the down-scaled kiln model, isothermal cases with four different nozzle diameters were simulated with the standard k-epsilon model. The aim is to investigate the effect of initial Reynolds number on the jet development in asymmetric co-flows from the air channels. It is found that, with increasing Reynolds number, the jet becomes shorter and the mixing between the primary jet and surrounding flow is better. A low-velocity region or external recirculation zone (ERZ) form near the kiln upper wall and shrink with increased nozzle diameter or decreased initial Reynolds number. The ERZ may stabilize the flame since it is a low-velocity region and consequently attract the jet to reside predominantly in it or in the shear layer. As a conclusion, by enlarging the jet exit diameter, the jet can be prolonged, while to a certain extent, the benefit from the ERZ would need to be sacrificed. Since the asymmetric development and the ERZ in the kiln were not studied during the previous PIV measurement campaign, more experimental studies are planned to provide more experimental evidences of the details of the flow and to lay grounds for validation of the CFD simulation results.

Wind turbines are an environmentally friendly and sustainable power source. Unfortunately, the noise impact can cause deteriorated living conditions for nearby residents. The audibility of wind turbine sound is influenced by ambient sound. This thesis deals with some aspects of noise from wind turbines. Ambient sounds influence the audibility of wind turbine noise. Models for assessing two commonly occurring natural ambient sounds namely vegetation sound and sound from breaking waves are presented in paper A and B. A sound propagation algorithm has been compared to long range measurementsof sound propagation in paper C. Psycho-acoustic tests evaluating the threshold and partial loudness of wind turbine noise when mixed with natural ambient sounds have been performed. These are accounted for in paper D. The main scientific contributions are the following.Paper A: A semi-empiric prediction model for vegetation sound is proposed. This model uses up-to-date simulations of wind profiles and turbulent wind fields to estimate sound from vegetation. The fluctuations due to turbulence are satisfactory estimated by the model. Predictions of vegetation sound also show good agreement to measured spectra. Paper B: A set of measurements of air-borne sound from breaking waves are reported. From these measurements a prediction method of sound from breaking waves is proposed. Third octave spectra from breaking waves are shown to depend on breaker type. Satisfactory agreement between predictions and measurements has been achieved. Paper C: Long range sound propagation over a sea surface was investigated. Measurements of sound transmission were coordinated with local meteorological measurements. A sound propagation algorithm has been compared to the measured sound transmission. Satisfactory agreement between measurements and predictions were achieved when turbulence were taken into consideration in the computations. Paper D: The paper investigates the interaction between wind turbine noise and natural ambient noise. Two loudness models overestimate the masking from two psychoacoustic tests. The wind turbine noise is completely concealed when the ambient sound level (A-weighed) is around 10 dB higher than the wind turbine noise level. Wind turbine noise and ambient noise were presented simultaneously at the same A-weighed sound level. The subjects then perceived the loudness of the wind turbine noise as 5 dB lower than if heard alone. Keywords: Wind turbine noise, masking, ambient noise, long range sound propagation
QC 20100705

This thesis studies how the mechanical design of a surveillance camera affects the acoustic performance, mainly in terms of the frequency response within the human hearing range. During the project, the mechanical characteristics that affect frequency response were investigated by measuring the camera’s audio behavior in an anechoic chamber. A theoretical and adaptable acoustic model was built in COMSOL to simulate the frequency response of the sound path. Measurement and simulation results were compared to identify critical aspects of the mechanical design and adjust accordingly for better acoustic performance.

The recent goals from some countries to become renewable energy based and reduce carbon dioxide emissions have caused the wind industry to grow. Together, the size of the wind farms and the noise emission have grown, while the noise emission regulations have to be fulfilled. Numerical simulations based on engineering approaches are in many cases a fast alternative that may supplement actual sound measurements at the site on question. However, the sound propagation models have many assumptions and estimations, as different variants can affect the resulting sound propagation. The accuracy of the sound propagation models Nord2000, CONCAWE, and ISO 9613-2 are investigated in this research by comparing the predicted to the measured sound pressure levels from a wind farm in northern Sweden. Different parameters were investigated in each model, as wind speed and direction, roughness length, ground class, temperature gradient, and receiver height. The computational calculations were run on SoundPLAN software for a single point, the nearby dwelling. For the different parameters investigated, the settings were defined and inputted in the software, and the calculations were run. The equivalent sound pressure level results from the computational models were compared to the equivalent sound pressure level of the sound measurements filtered from background noise. The results indicate that the model ISO 9613-2 did not perform well for the specific site conditions at the wind farm. On the other hand, the CONCAWE and Nord2000 showed high accuracy, for downwind conditions at 8 m/s. For upwind conditions at 8 m/s, Nord2000 is more accurate, as the refraction of the sound rays are better calculated on this model. For the variants investigated on the Nord2000 model, the results that better approximate to the sound levels of the sound measurements are the roughness length 0.3, ground class D, and temperature gradient  0.05 K/m. Thus, these settings would be recommended for calculations with Nord2000 for noise assessment in a permit process.

The result of this work is a fully working and significantly optimized implementation of a dynamic decoder. This decoder is based on dynamic recognition network generation and decoding by a modified version of the Token Passing algorithm. The implemented solution provides very similar results to the original static decoder from BSCORE (API of Phonexia company). Compared to BSCORE this implementation offers significant reduction of memory usage. This makes use of more complex language models possible. It also facilitates integration the speech recognition to some mobile devices or dynamic adding of new words to the system.

This diploma thesis deals with the analysis and design optimization of a speaker cabinet (enclosure) produced by the additive technology of 3D sand printing. The introductory part is devoted to the theory, both in the field of noise, vibration and the method of their calculations, as well as the theory of speakers and their systems. Numerical and experimental modal analysis is performed, which were compared to determine the mechanical properties of the new material, including material damping. This is followed by experimental and numerical harmonic analysis, with output in the form of a numerical model describing the behaviour of the structure during its excitation. The data are compared with a modal analysis of the internal acoustic space and the critical shapes and their frequencies are determined. At the end of the work, construction modifications are proposed to increase the rigidity of enclosures, the influence of which is evaluated on the created numerical models.

Much of the planned installation of wind turbines in Sweden will be located in the northern region, characterized by a lower population density so that problems related to sound pollution and visual acceptance are of lower concern. This area is generally distinguished by complex topography and the presence of forest, that significantly affects the wind characteristics, complicating their modelling and simulation. There are concerns about how good an industrial code can simulate a forest, a question of paramount importance in the planning of new onshore farms. As a first step, a sensitivity analysis was initially carried out to investigate the impact on the ow of different boundary conditions and cell discretization inside the forest for a 2D domain with a homogeneous forest. Subsequently, a comparative analysis between the industrial code WindSim and Large Eddy Simulation (LES) data from Segalini. et al. (2016) was performed with the same domain. Lastly, simulations for a real Swedish forest, Ryningsnäs, was conducted to compare a roughness map approach versus modelling the forest as a momentum sink and a turbulence source. All simulations were conducted for neutral stability conditions with the same domain size and refinement. The main conclusions from each part can be summarized as follows. (i) The results from the sensitivity analysis showed that discretization of cells in the vertical direction inside the forest displayed a correlation between an increasing number of cells and a decreased streamwise wind speed above the canopy. (ii) The validation with the LES data displayed good agreement in terms of both horizontal mean wind speed and turbulence intensity. (iii) In terms of horizontal wind speed for Ryningsnäs, forest modelling was prevailing for all wind directions, where the most accurate simulation was found by employing a constant forest force resistive constant (C2) equal to 0.05. All forest models overestimated the turbulence intensity, whereas the roughness map approaches underestimated it. Based solely on the simulations for Ryningsnäas, a correlation between lower streamwise wind speed and higher turbulence intensity can be deduced.

This thesis deals with different parallelizations of training procedure for artificial neural networks. The networks are trained as phoneme-state acoustic descriptors for speech recognition. Two effective parallelization strategies were implemented and compared. The first strategy is data parallelization, where the training is split into several POSIX threads. The second strategy is node parallelization, which uses CUDA framework for general purpose computing on modern graphic cards. The first strategy showed a 4x speed-up, while using the second strategy we observed nearly 10x speed-up. The Stochastic Gradient Descent algorithm with error backpropagation was used for the training. After a short introduction, the second chapter of this thesis shows the motivation and introduces the neural networks into the context of speech recognition. The third chapter is theoretical, the anatomy of a neural network and the used training method are discussed. The following chapters are focused on the design and implementation of the project, while the phases of the iterative development are described. The last extensive chapter describes the setup of the testing system and reports the experimental results. Finally, the obtained results are concluded and the possible extensions of the project are proposed.

In this thesis we apply linear feedback control to spatially evolving flows in order to minimize disturbance growth. The dynamics is assumed to be described by the linearized Navier--Stokes equations. Actuators and sensor are designed and a Kalman filtering technique is used to reconstruct the unknown flow state from noisy measurements. This reconstructed flow state is used to determine the control feedback which is applied to the Navier--Stokes equations through properly designed actuators. Since the control and estimation gains are obtained through an optimization process, and the Navier--Stokes equations typically forms a very high-dimensional system when discretized there is an interest in reducing the complexity of the equations. One possible approach is to perform Fourier decomposition along (almost) homogeneous spatial directions and another is by constructing a reduced order model by Galerkin projection on a suitable set of vectors. The first strategy is used to control the evolution of a range of instabilities in the classical family of Falkner--Skan--Cooke flows whereas the second is applied to a more complex cavity type of geometry.

This master thesis deals with the implementation of various types of recurrent neural networks via programming language lua using torch library. It focuses on finding optimal strategy for training recurrent neural networks and also tries to minimize the duration of the training. Furthermore various types of regularization techniques are investigated and implemented into the recurrent neural network architecture. Implemented recurrent neural networks are compared on the speech recognition task using AMI dataset, where they model the acustic information. Their performance is also compared to standard feedforward neural network. Best results are achieved using BLSTM architecture. The recurrent neural network are also trained via CTC objective function on the TIMIT dataset. Best result is again achieved using BLSTM architecture.

Over the past decades, attention has been brought to the importance of indoor air quality and the serious threat of bio-aerosol contamination towards human health. A novel idea to transport hazardous particles away from sensitive areas is to automatically control bio-aerosol concentrations, by utilising airflows from ventilation systems. Regarding this, computational fluid dynamics (CFD) may be employed to investigate the dynamical behaviour of airborne particles, and data-driven methods may be used to estimate and control the complex flow simulations. This thesis presents a methodology for machine-learning based control of particle concentrations in turbulent gas-solid flow. The aim is to reduce concentration levels at a 90 degree corner, through systematic manipulation of underlying two-phase flow dynamics, where an energy constrained inlet airflow rate is used as control variable. A CFD experiment of turbulent gas-solid flow in a two-dimensional corner geometry is simulated using the SST k-omega turbulence model for the gas phase, and drag force based discrete random walk for the solid phase. Validation of the two-phase methodology is performed against a backwards facing step experiment, with a 12.2% error correspondence in maximum negative particle velocity downstream the step. Based on simulation data from the CFD experiment, a linear auto-regressive with exogenous inputs (ARX) model and a non-linear ARX based neural network (NN) is used to identify the temporal relationship between inlet flow rate and corner particle concentration. The results suggest that NN is the preferred approach for output predictions of the two-phase system, with roughly four times higher simulation accuracy compared to ARX. The identified NN model is used in a model predictive control (MPC) framework with linearisation in each time step. It is found that the output concentration can be minimised together with the input energy consumption, by means of tracking specified target trajectories. Control signals from NN-MPC also show good performance in controlling the full CFD model, with improved particle removal capabilities, compared to randomly generated signals. In terms of maximal reduction of particle concentration, the NN-MPC scheme is however outperformed by a manually constructed sine signal. In conclusion, CFD based NN-MPC is a feasible methodology for efficient reduction of particle concentrations in a corner area; particularly, a novel application for removal of indoor bio-aerosols is presented. More generally, the results show that NN-MPC may be a promising approach to turbulent multi-phase flow control.

Roterugnar är cylindriska kärl som används för att höja materials temperaturer i en kontinuerlig process som kallas för kalcinering. Roterugnar kan tillämpas i olika processer såsom reduktion av oxidmalm samt återvinning av farligt avfall. Fördelen med roterugnar ligger i dess förmåga att hantera råmaterial som sträcker sig från slam till granulära material med en mängd olika partikelstorlekar, och därigenom upprätthålla distinkta miljöer såsom en bädd av fasta partiklar som samexisterar med ett oxiderande fribord. Sex olika bäddbeteende har dokumenterats med avseende på fyllningsgrad samt Froude nummer. Syftet med denna studie var att utveckla en tvådimensionell suspensions modell med CFD genom att använda den kommersiella mjukvaran COMSOL 5.5 för att simulera de två faser, gas och fast, som en blandad fas efter verk av Philips et. al., Physics of Fluids A: Fluid Dynamics 4.1 (1992) 30-40 och Acrivos & Zhang., International Journal Multiphase Flow 20.3 (1994) 579-591. Denna modell undersöktes genom att jämföra den med de dokumenterade flödesregimerna samt genom parameter som partikelstorlek, partikeldensitet och viskositeten hos gas i flödesregimen känd som rullande läge. Dessutom undersöktes temperaturprofilen för den roterande ugnen genom att utforska hur blandningsvariationer av den fasta bädden i den roterande ugnen påverkas av värmeöverföringen när värme tillförs från väggen under rullande läge. Resultaten av den tvådimensionella suspension modellen visade att det var bara möjligt att simulera glidläge korrekt; andra lägen kunde inte beskrivas som dokumenterat i litteraturen. Det indikeras att vilovinkeln och viskösa krafter i den roterande ugnen var låga vilket resulterade i att suspensions modellen inte kunde avbilda exakt de återstående flödesregimerna som dokumenterat. Till exempel avbildades rullningsläget mer likt forsandeläge då partiklarna fall fritt efter höjning av bädden. Partikelstorlek och partikeldensitet har visat sig ha en betydande påverkan på suspensions modellen eftersom de viskösa krafterna blir låga för en partikelstorlek och partikeldensitet under 0,4 mm respektive 1500 kg/m3. Angående gasens viskositet visades det sig att ju närmare värdet 2.055e-3 (Pa*s) den blev desto större blev sedimentationsflödet vilket resulterade i att bäddpartiklarna dras ner och förblir där. Suspensions modellen kunde således simulera en fast och flytande fas och inte en gasfas som avsett. Slutligen visade temperaturanalysen att påverkan av den termiska konduktiviteten var mer signifikant än den specifika värmekapaciteten i intervallet 1 - 50 (W/(m*K)) respektive 300 - 800 (J/(kg*K)) på grund av den tid det tog att nå en homogen temperaturprofil.
Rotary kilns are cylindrical vessels used to raise materials temperature in a continuous process known as calcination. Rotary kilns find application in various processes such as reduction of oxide ore and hazardous waste reclamation. The advantage of the rotary kiln lies in its ability to handle feedstock ranging from slurries to granular materials with a variety of particle size, thereby maintaining distinct environments such as a bed of solid particles coexisting with an oxidising freeboard. Six different bed behaviours within the kiln have been documented with respect to the filling degree and Froude number. The aim of this study was to develop a two-dimensional suspension model with CFD by using the commercial software COMSOL 5.5 to simulate the two phases, gas and solid, as a mixed phase, following the works of Philips et. al., Physics of Fluids A:  Fluid Dynamics 4.1 (1992) 30-40 and Acrivos & Zhang., International Journal Multiphase Flow 20.3 (1994) 579-591. This model was investigated by comparing it against the documented flow regimes as well as through parameters such as particle size, particle density and viscosity of gas in the flow regime known as rolling mode. In addition, the temperature profile of the rotary kiln was investigated by exploring how the mixture variation of the solid bed within the rotary kiln affects the heat transfer when heat is supplied from the wall during a rolling mode. The results of the two-dimensional suspension model showed that it was only possible to simulate the slipping mode accurately; others mode could not be described as documented in literature. It is indicated that the angle of repose and viscous forces within the rotary kiln were low resulting in the suspension model not being able to accurately depict the remaining flow regimes as documented. For instance, the rolling mode was depicted more as a cataracting mode due to the free fall of particles after elevation of the bed. The particle size and the particle density were found to have a significant impact on the suspension model as the viscous forces became low for a particle size and particle density below 0.4 mm and 1500 kg/m3 respectively. As for the viscosity of gas it was found that the closer it got to the value 2.055e-3 (Pa*s) the sedimentation flux became too large resulting in the bed particles being pulled down and remaining there. Thus, the suspension model could simulate a solid and liquid phase and not a gas phase as intended. Lastly, the temperature analysis revealed that the impact of the thermal conductivity was more significant than the specific heat capacity in the range of 1 - 50 (W/(m*K)) and 300 - 800 (J/(kg*K)) respectively, due to the time it took to reach a homogeneous temperature profile.

At the Torpshammar dam two rectangular beams are situated upstream of the spillway gates to stabilize the sidewalls holding the embankment of the dam. A computational fluid dynamics (CFD) simulation of the dam with the bottom outlets open was made to investigate how the flow and discharge capacity is affected by the beams. The results can be used to avoid unexpected consequences due to turbulence caused by the beams, make the beams strong enough to hold the pressure from the flow and get an estimation of the discharge capacity with the beams. Turbulence is one of the hardest things to simulate so the results were compared with previous simulation work made without the beams and physical model tests to validate the results. Also, a sensitivity analysis was made to investigate the method used. The beams lowered the velocity (to 17 m/s) and the discharge capacity (to 255 m3/s) compared to the previous work. The force on the beams was directed upward and downstream. The beams increased the turbulence and the vortex shedding frequency was higher for the beam closest to the outlet. The velocity and discharge capacity differed with 6 % compared to model test results. The results can therefore only be used as an estimation, a more detailed computational model and more computational cells are needed to get a better result. The sensitivity analysis showed that the velocity and turbulence depend on the method and further studies need to be made to decide which method gives the closest similarity with reality.

A number of experimental and theoretical studies, performed in the field of technical flow duct acoustics are presented in this thesis. The acoustical methods treated are implemented on turbocharged IC-engines and engine gas exchange system components. A new method based on the well-known two-load technique has been developed. The method was applied to characterise the source data of various piston-engines with non-linear behaviour including a 6 cylinder turbo-charged truck diesel engine. The source characterisation results were compared to the results obtained using the linear two-load technique. It was demonstrated that the new non-linear multi-load technique gives improved results when the source is slightly non-linear. The use of active one-port models has been tested to characterize an air terminal device (ATD) as a source of flow generated noise. In order to predict the noise generation at different operating points of the device a scaling law was derived and verified. In the experimentally derived scaling law a flow speed dependence of 3 was found for the narrow band spectra, corresponding to a dipole-like behavior of the source in the plane wave range. The proposed technique was validated successfully and the results indicated a good prediction of in-duct sound generation by the air terminal device. Sound reflection from hot flow duct openings has been investigated experimentally. The reflection coefficient was measured for flow temperatures up to 500 ºC and jet velocities up to 108m/s. The results have been compared with famous Munt’s theory. It was concluded that at low Mach number and Helmholz number cases the results agree well with the Munt’s model. This was the first experimental validation of the theory for hot flow conditions. Experimental procedures to determine the sound transmission through automotive turbo-charger compressors were developed and described in detail. An overview of a unique turbocharger testing facility established at KTH CICERO in Stockholm is given. The facility can be used to measure acoustic two-port data for turbo-compressors. Results from measurements on a passenger car turbo-compressor are presented and the influence of operating conditions on the sound transmission is discussed. Current wave action models developed in CMT for computation of the gas exchange processes in I.C. engines have been implemented to determine the acoustic wave transmission through the turbo- compressor. The models are validated with the experimental data and the results are presented for different operating conditions of a Volvo passenger car turbo-compressor.
QC 20100809

The high turbine inlet temperature of modern gas turbines poses a challenge to the material used in the turbine blading of the primary stages. Mechanical failure mechanisms are more pronounced at these high temperatures, setting the lifetime of the blade. It is therefore crucial to obtain accurate local metal temperature predictions of the turbine blade. Accurately predicting the external heat transfer coefficient (HTC) distribution of the blade is therefore of uttermost importance. At present time, Siemens Energy uses the boundary layer code TEXSTAN for this purpose. The limitations coupled to such codes however make them less applicable for the complex flow physics involved in the hot gas path of turbine blading. The thesis therefore aims at introducing CFD for calculating the external HTC. This includes conducting an extensive literature study to find and validate a suitable methodology. The literature study was centered around RANS modeling, reviewing how the calculation of the HTC has evolved and the performance of some common turbulence and transition models. From the literature study, the SST k − ω model in conjunction with the γ − Reθ transition model, the v2 − f model and the Lag EB k − ε model were chosen for the investigation of a suitable methodology. The validation of the methodology was based on the extensively studied LS89 vane linear cascade of the von Karman Institute. In total 13 test cases of the cascade were chosen to represent a wide range of flow conditions. Both a periodic model and a model of the entire LS89 cascade were tested but there were great uncertainties whether or not the correct flow conditions were achieved with the model of the entire cascade. It was therefore abandoned and a periodic model was used instead. The decay of turbulence intensity is not known in the LS89 cascade. This made the case difficult to model since the turbulence boundary conditions then were incomplete. Two approaches were attempted to handle this deficiency, where one was ultimately found invalid. It was recognized that the Steelant-Dick postulation could be used in order to find a turbulent length scale which when specified at the inlet, lead to fairly good agreement with data of the HTC. The validation showed that the SST γ − Reθ model performs relatively well on the suction side and in transition onset predictions but worse on the pressure side for certain flow conditions. The v2 − f model performed better on the pressure side and on a small portion of the suction side. Literature emphasized the importance of obtaining proper turbulence characteristics around the vane for accurate HTC-predictions. It was found that the results of the validation step could be closely coupled to this statement and that further work is needed regarding this. Further research must also be done on the Steelant-Dick postulation to validate it as a reliable method in prescribing the inlet length scale.

Für die Luftqualitätsbewertung in Städten sind Informationen zur raumzeitlichen Variabilität luftgetragener Feinstäube auf kleiner Skala von wichtiger Bedeutung. Standardisierte Messverfahren, zur Bestimmung von Partikelkonzentrationen, sind mit hohem Aufwand verbunden, weshalb dichte Messnetze fehlen. Partikelausbreitungsmodelle sind kompliziert in der Anwendung und/oder benötigen hohe Computerrechenleistung. Infolgedessen gibt es bezüglich örtlicher Partikelkonzentrationen große Informationslücken. Diese Arbeit untersucht die mikroskalige Variabilität von Aerosolen in Raum und Zeit mit unterschiedlichen Methoden. Es wurden Erhebungen mit mobilen Sensoren und eine Passantenbefragung durchgeführt. Weiterhin wurden in dieser Arbeit die physikalischen Partikeltransportmodelle ENVI-met und Austal2000 in ihrer Leistung bewertet und in angewandten Studien eingesetzt. Weiterhin wurde ein neuronales Netzwerk zur Vorhersage von Partikelkonzentrationen entwickelt. Die Untersuchungen erfolgten in den Städten Aachen und Münster. Es konnten unerwartete Verteilungsmuster hinsichtlich der Massekonzentration von Partikeln beobachtet werden. In einem innerstädtischen Park wurden diffuse Partikelquellen identifiziert, mit einem deutlichen Hinweis darauf, dass feuchtgelagerte Wegedecken einen maßgeblichen Anteil an lokalen Partikelimmissionen hatten. Weiterhin wurde Straßenverkehr als wichtiger Beitrag zum städtischen Aerosol identifiziert. Passanten, die verschiedenen Partikelkonzentrationen ausgesetzt waren, konnten diese perzeptiv nicht unterscheiden. Simulationsergebnisse von Austal2000 und ENVI-met wiesen Unterschätzungen im Vergleich zu Messwerten auf. Das entwickelte neuronale Netzwerk prognostizierte Partikelkonzentrationen teilweise mit hoher Genauigkeit. Das große Potenzial von neuronalen Netzen für die Vorhersage von Partikelkonzentrationen in räumlicher und zeitlicher Ausdehnung, auch für den Bereich der Luftqualitätsüberwachung, wurde aufgezeigt.
Knowledge about the micro-scale variability of airborne particles is a crucial criterion for air quality assessment within complex terrains such as urban areas. Due to the significant costs and time consumption related to the work required for standardized measurements of particle concentrations, dense monitoring networks are regularly missing. Models that simulate the transmission of particles are often difficult to use and/or computationally expensive. As a result, information regarding on-site particle concentrations at small scales is still limited. This thesis explores the micro-scale variability of aerosol concentrations in space and time using different methods. Experimental fieldwork, including measurements with mobile sensor equipment alongside a survey, and modeling approaches were conducted. Applied simulation studies, a performance assessment of two popular particle dispersion models, namely Austal2000 and ENVI-met, as well as the development of an ANN model are presented. The cities of Aachen and Münster were chosen as case studies for this research. Unexpected patterns of particle mass concentrations could be observed, including the identification of diffuse particle sources inside a park area with strong evidence that unpaved surfaces contributed to local aerosol concentration. In addition, vehicle traffic was proved to be a major contributor of particles, particularly close to traffic lanes. Results of the survey reveal that people were not able to distinguish between different aerosol concentration levels. Austal2000 and ENVI-met turned out to have room for improvement in terms of the reproduction of observed particle concentration levels, with both models having a tendency toward underestimation. The newly developed ANN model was confirmed to be a fairly accurate tool for predicting aerosol concentrations in both space and time, and demonstrates the principal ability of the approach also in the domain of air quality monitoring.

The Master's thesis deals with the creation of the numerical model that serves to analysis of the acoustic properties of separating construction of a wooden structure. The work includes theoretical basis for solving the subject, it discusses the airborne sound insulation of separating structures, focuses on wooden buildings, describes a computing environment ANSYS and model creation in it. More variants of model were compiled, CLT panel and frame construction. Model itself is using scripting language APDL and each step of creation is described. The results of numerical simulations describe the distribution of pressure in front and behind the separation construction. Results are compiled into graphs and single value of weighted airborne sound insulation is evaluated. In conclusion individual results of model variants are compared, model accuracy improvements are discussed.

(EN) Thesis deals with an approximation of static moduli in wells from dynamic moduli determined by acoustic well logging using T-matrix model. Proposed approach makes possible to determine moduli values, which are close to values of static moduli, which would be determined by loading tests. This approach is based on an idea, that an intact rock with sufficiently high compressional strength sc and sufficiently high value of static Young's modulus Es, manifests more or less linear elastic behaviour. In such case, the values of static and dynamic moduli are identical. This fact has been experimentally verified for rocks with values of sc and Es in order of higher tens of MPa and GPa respectively. In case of a rock damage presence in such rock, it's behaviour becomes nonlinearly elastic. The amount of nonlinearity is proportional to increasing amount of rock damage. This results in the difference between values of static and dynamic moduli. T-matrix model is used to quantify this difference. This model is based on an anisotropic rock matrix with ellipsoidal inclusions. These inclusions can affect each other. The result of this model calculation is a group of values of elastic constants, which we call effective moduli. These effective moduli include the effect of porosity in the rock as well and they...

Getriebeheulphänomenen wird in der industriellen Praxis zum Teil noch immer ausschließlich mit Hilfe einer optimierten Verzahnungsauslegung zur Minimierung des Drehfehlers begegnet. Bei auffälligen Resonanzerscheinungen werden zudem Strukturoptimierungen am Getriebegehäuse und den Karosserieanbindungspunkten des Antriebsstrangs vorgenommen, ohne auf die internen Systemkomponenten verstärkt einzugehen. Zudem kann bisher die verlässliche Erkenntnis, dass zur Einhaltung akustischer Grenzwerte die Konstruktion nochmals überarbeitet werden muss, erst spät im Entwicklungsprozess während der akustischen Versuchsdurchführung an Prüfständen oder im Fahrzeug getroffen werden. Durch sogenannte Sekundärmaßnahmen, die typischerweise nicht unmittelbar das Anregungsverhalten oder die Dynamik des Antriebsstrangs, sondern den karosserieseitigen Transferpfad betreffen, kann eine Reduktion des Schalldruckpegels im Fahrzeuginnenraum in gewissen Grenzen erzielt werden. Dies ist zumeist weder aus ingenieurstechnischer Sicht noch aus Sicht steigender Entwicklungskosten als optimal zu bezeichnen. Basierend auf einem detaillierten Abgleich zwischen Experiment und Simulation, angefangen auf Einzelteilebene über die Baugruppenebene bis hin zur Methodenentwicklung der Validierung von nichtrotierenden Gesamtsystemen mittels künstlicher Anregung, werden in dieser Arbeit möglichst akkurate Simulationsergebnisse angestrebt, um die Auswirkung von akustischen Optimierungen innerhalb der Simulationsumgebung realitätsnah vorherzusagen. Schließlich wird ein Prozess vorgeschlagen, der eine getriebeinterne Transferpfadanalyse zur Identifikation sensitiver Körperschallpfade vorstellt. Mit dieser Methode wird aufgezeigt, dass es möglich ist, akustische Schwachstellen auf der Antriebsstrangseite vorherzusagen. Dabei kann zum einen akustisches Optimierungspotenzial des Welle-Lager-Systems abgeleitet werden, zum anderen werden auch die aktuellen Möglichkeiten und Grenzen der Verfahren beleuchtet.:Inhaltsverzeichnis Abbildungsverzeichnis Formelzeichen und Abkürzungen 1 Einleitung 2 Stand der Technik 2.1 Getriebeakustik – Einflussgrößen und Begrifflichkeiten 2.1.1 Verzahnungsanregung von unter Last stehenden Zahnrädern 2.1.2 Verzahnungsinduzierte Körper- und Luftschallweiterleitung in Fahrzeugen 2.1.3 Maßnahmen zur Reduktion von Körperschallpfaden 2.2 Strukturdynamische Analysemethoden in der Getriebeakustik 2.2.1 Numerische Modalanalyse 2.2.2 Experimentelle Modalanalyse und Betriebsschwingformanalyse 2.2.3 Unkonventionelle Methoden zur Erregung von Bauteilstrukturen 2.2.4 Computerunterstützte Modellanpassung 2.2.5 Transferpfadanalyse 2.3 Getriebesimulation 2.3.1 Getriebetypische Kontaktmodellierung 2.3.2 Finite-Element-Modellierung und Reduktionsverfahren 2.4 Fazit 3 Zielsetzung und Vorgehensweise 4 Methoden zur Analyse von Getriebekomponenten 4.1 Verzahnungsanalyse 4.2 Strukturdynamische Untersuchungen an Einzelteilen 4.3 Strukturdynamische Untersuchungen an Baugruppen 4.4 Verhalten von Wälzlagern 4.5 Fazit 5 Methode zur Gesamtsystemvalidierung 5.1 Modellierung des Getriebesystems mit Prüfstandsanbindung 5.2 Getriebeinterne Anregung mittels Zahnaktor 5.2.1 Voruntersuchungen 5.2.2 Implementierung einer torsionalen Anregung in Frontgetrieben 5.2.3 Simulative und experimentelle Untersuchungen 5.3 Fazit 6 Getriebeinterne Transferpfadanalyse 6.1 Identifikation sensibler Körperschalltransferpfade 6.2 Optimierungsansätze 6.3 Fazit 7 Zusammenfassung und Ausblick 8 Literaturverzeichnis
Gear whine phenomena are typically mitigated by optimising gear design to minimise Transmission Error. Additionally, structural optimisations on the gearbox housing and on the gearbox-to-chassis mounts may be conducted most likely without a detailed consideration of internal components‘ dynamics. Moreover, it is not unusual that in the final stages of drivetrain developments when the gearbox is tested on test benches or in vehicles NVH targets are not met. The countermeasures applied often do not alter the source of excitation nor the dynamics of the drivetrain at this stage, but the vehicle transfer paths through the chassis. Thus, sound power level reduction may be exclusively limited to vehicle transfer path improvements. In most cases this is not regarded as an optimal solution neither from an engineering perspective nor from the commercial point of view due to increased development costs. Based on detailed correlation activities where simulation has been run against measurement the work strives to achieve accurate gearbox NVH predictions to forecast more realistically the effect of design optimisations. The correlation approach starts on single component level moving on to sub-assembly level and finally up to the development of a method which is able to artificially excite the entire but non-rotating drivetrain. After various successful correlation studies, a process has been elaborated which proposes a gearbox internal transfer path analysis to identify potentially critical structure-borne noise paths. The method illustrates the capability of successfully predicting weak spots at the active side of a drivetrain early in the development process. Additional room for improvement can be derived when implementing this method by considering the rotating components of a drivetrain such as gear blanks, shafts and bearings.:Inhaltsverzeichnis Abbildungsverzeichnis Formelzeichen und Abkürzungen 1 Einleitung 2 Stand der Technik 2.1 Getriebeakustik – Einflussgrößen und Begrifflichkeiten 2.1.1 Verzahnungsanregung von unter Last stehenden Zahnrädern 2.1.2 Verzahnungsinduzierte Körper- und Luftschallweiterleitung in Fahrzeugen 2.1.3 Maßnahmen zur Reduktion von Körperschallpfaden 2.2 Strukturdynamische Analysemethoden in der Getriebeakustik 2.2.1 Numerische Modalanalyse 2.2.2 Experimentelle Modalanalyse und Betriebsschwingformanalyse 2.2.3 Unkonventionelle Methoden zur Erregung von Bauteilstrukturen 2.2.4 Computerunterstützte Modellanpassung 2.2.5 Transferpfadanalyse 2.3 Getriebesimulation 2.3.1 Getriebetypische Kontaktmodellierung 2.3.2 Finite-Element-Modellierung und Reduktionsverfahren 2.4 Fazit 3 Zielsetzung und Vorgehensweise 4 Methoden zur Analyse von Getriebekomponenten 4.1 Verzahnungsanalyse 4.2 Strukturdynamische Untersuchungen an Einzelteilen 4.3 Strukturdynamische Untersuchungen an Baugruppen 4.4 Verhalten von Wälzlagern 4.5 Fazit 5 Methode zur Gesamtsystemvalidierung 5.1 Modellierung des Getriebesystems mit Prüfstandsanbindung 5.2 Getriebeinterne Anregung mittels Zahnaktor 5.2.1 Voruntersuchungen 5.2.2 Implementierung einer torsionalen Anregung in Frontgetrieben 5.2.3 Simulative und experimentelle Untersuchungen 5.3 Fazit 6 Getriebeinterne Transferpfadanalyse 6.1 Identifikation sensibler Körperschalltransferpfade 6.2 Optimierungsansätze 6.3 Fazit 7 Zusammenfassung und Ausblick 8 Literaturverzeichnis