Eight roughness configurations in Clemson boundary layer wind tunnel are presented. For these configurations, flow parameters such as turbulent intensities, integral length scales, large- and small- scale turbulence, and spectra of velocity components of the wind are obtained and studied to the simulated turbulence. At the same time, new analyzing tools, orthogonal wavelet techniques, are applied to provide additional information in time domain. This makes it possible to study the intermittency event, one important characteristic associated with pressure peak activities in turbulence. Three parameters, scale energy, intermittency factor and intermittency energy are defined. Variation of these quantities as a result of different configuration is discussed. Finally, the corresponding variations in measured pressure peaks in relation with the variations of configuration as well as with the intermittency parameters are investigated. The work here is of important significance for future wind tunnel and field data comparison, and this could help to find the best simulation among all configurations. / Master of Science

Doctor of Philosophy / Department of Mechanical and Nuclear Engineering / Zhongquan Zheng / Windscreens are widely used in outdoor microphone measurement for acoustic sensing systems. In many cases of outdoor microphone applications, wind noise interferes with the signals. The performance of measurement microphones thus heavily depends on correct designs of windscreens that are used to maximize the signal to noise ratio of the sensing system. The purpose of the study is to investigate the wind noise reduction between the unscreened microphone and the screened microphone under different frequencies of incoming wind turbulence. In this study, a modified immersed boundary method using the distributed forcing term has been applied to simulate the flow/acoustic interaction between air and the porous medium. Because of the high accuracy requirement in the vicinity of the interface between air and the porous medium, spatial derivatives of flux need to be discretized using high order schemes. In this study, several different schemes have been tested in the vicinity of the interface including a second-order upwind scheme, a third-order upwind scheme, and a fifth-order Weighted Essentially Non-Oscillatory (WENO) scheme. Based on the test results, the fifth-order WENO scheme is selected for most of the simulation cases. The model equations for flow outside the windscreen are the Navier-Stokes equations; flow inside the windscreen (porous medium) uses the modified Zwikker-Kosten equation. The wind turbulence in this study is generated by two different ways. The first is to place different sizes of solid cylinders and spheres in the upstream of the microphone under two-dimensional and three-dimensional conditions. The second is to use a Quasi-Wavelet method to generate the background atmospheric turbulence to simulate the real physical phenomena. Both two-dimensional and three-dimensional simulations for the flow over the unscreened and the screened microphone are presented and discussed under both low Reynolds number and high Reynolds number flow conditions. The results show that the windscreen effect is significant and the wind noise reduction level between the unscreened and the screened microphone can reach around 20dB either for low Reynolds number cases or for high Reynolds number cases. For low Reynolds number cases, Low flow resistivity windscreens are more effective for low frequency turbulence; high flow resistivity windscreens are more effective for high frequency turbulence. For high Reynolds number cases, the medium flow resistivity windscreens perform better compared to low flow resistivity windscreens and high flow resistivity windscreens.

Approved for public release, distribution is unlimited / Vertically propagating waves (VPWs) generated by prominent mountain ridges are a severe hazard to military aircraft operations. Properly forecasting the initiation and duration of such a phenomenon is critical, yet quite often missed by turbulence forecasters. A primary reason for poor forecast skill is vague VPW forecasting guidelines at the Air Force operational centers, focusing a majority of attention on the less severe, more common trapped lee wave response. The United States Air Forces in Europe Operational Weather Squadron (USAFE OWS) has requested a tool to aid in improving forecast ability of VPW events. Satellite analysis from October 2003 through March 2004 indicated an occurrence of six major VPW events to the lee of the Alps. Actual verification of turbulence in each VPW was unavailable due to the minimal pilot report (PIREP) database kept for military flights over Europe, therefore, a subjective assessment of turbulent conditions was determined depending on the resulting cloud signature. Using NCEP GFS model analysis and upstream upper air soundings during these events, an average synoptic condition and critical weather parameters were created. These developed tools were then tested from October 2004 through March 2005 to prove their reliability. In a limited data set these tools identified all VPW events, with only a 25% false alarm rate. This is compared to a 6% forecast ability with 0% false alarm rate determined during the 2003-2004 winter season by USAFE OWS forecasters. These new rules should be valuable in that they will provide a much needed capability for synoptic scale turbulence forecasters to better determine hazardous aviation conditions associated with VPWs. / Captain, United States Air Force

Approved for public release; distribution in unlimited. / Optical turbulence plays an important role in the propagation of electromagnetic waves through the atmosphere because it broadens and distorts the optical beam. A variety of optical, thermal, and acoustic instruments are used to detect the atmospheric turbulence and an acoustic echosounder has proven to be a valuable tool to probe the fine dynamic structure of atmospheric turbulence within first hundred meters above the surface. The first planar acoustic echosounder constructed at the Naval Postgraduate School was by Weingartner and Wroblewski, under Walters' supervision. Moxcey later modified this design by reducing the number of drivers from 25 to 19 and placing the drivers closer together into a hexagonal, close-packed array. This thesis explored the potential sources of the transducer ringing and implemented solutions to the problem. Additionally, we also improved the receiving sensitivity of the echosounder and lowered the electronics noise when receiving. Finally, we applied these techniques to another array assembled with new drivers to improve its performance compared to the previous echosounder array, while measuring and quantifying the level of improvement achieved. / Lieutenant Commander, Republic of China Navy

Para vegetação esparsa e de porte alto a determinação dos parâmetros de rugosidade é comprometida pela dificuldade em se observar condições que satisfaçam a lei logarítmica da velocidade do vento. Estimou-se o comprimento de rugosidade (z0) e o deslocamento do plano zero (d) por alguns métodos com medidas micrometeorológicas e da estrutura física de arbustos esparsos em região semi-árida, durante o experimento HAPEX-Sahel. A velocidade do vento foi medida em quatro alturas acima da superfície (3,0; 4,1; 5,3 e 8,5 m), e os fluxos determinados por correlações dos turbilhões a 9m de altura. Métodos baseados no perfil logarítmico foram aplicados em condições de atmosfera neutra. A altura média da vegetação era h = 2,06 ± 0,47 m. O método convencional (ajuste estatístico) resultou em estimativas satisfatórias de d e z0 em condições nas quais a validade do perfil logarítmico foi satisfeita. Com uma única altura de medida localizada acima da subcamada inercial as estimativas resultaram em valores ou fisicamente inconsistentes ou que não caracterizam a rugosidade da superfície. Quando se utilizou a velocidade de fricção dada pela correlação dos turbilhões na solução do perfil logarítmico, as estimativas melhoraram. A combinação do perfil logarítmico com a relação z0 = λ (h - d) proporcionou estimativas satisfatórias para os valores de λ = 0,188 e 0,190 determinados em função da estrutura física da vegetação, o que não foi observado para o valor médio da literatura (0,166). Relações entre a estrutura física da vegetação e o transporte de momentum estimaram apropriadamente d e z0. A rugosidade da área foi melhor descrita por d = 0,95 m = 0,46 h e z0 = 0,204 m = 0,1 h, sendo λ = 0,185. As velocidades horizontal do vento e de fricção foram mais sensíveis a variações em z0 do que em d. / For sparse and tall vegetation the estimate of roughness parameters is compromised by the difficulty in observing conditions that satisfy the windspeed logarithmic law. The roughness length (z0) and the zero-plane displacement (d) were estimated by some methods with micrometeorological measurements and the physical structure of sparse shrubs in semi-arid region, during the HAPEX-Sahel experiment. The wind speed was measured at four heights above of surface (3.0, 4.1, 5.3 and 8.5 m), and the turbulent flows determined by eddy correlations at the height of 9m. Methods based on the logarithmic profile have been applied in neutral atmosphere conditions. The average height of the vegetation was h = 2.06 ± 0.47 m. The conventional method (statistical fit) resulted in good estimates of d and z0 only under conditions of validity of the logarithmic law. Only one height of measurement located above of the inertial sublayer is enough to result in physically inconsistent values. When the friction velocity, given by eddy correlation, was used in the logarithmic law, the estimates improved. The combination of the logarithmic law with z0 = λ (h - d) provided satisfactory estimates of the surface roughness for λ = 0.188 and 0.190 determined in function of the physical structure of the vegetation; but for λ = 0.166, the average value of literature, the estimates where not good. Relationships between the physical structure of the vegetation and the momentum transfer estimated appropriately d and z0. The area roughness was better described by d = 0.95 m = 0.46 h and z0 = 0.204 m = 0.1 h, being λ = 0.185. Wind speed and friction velocity were more sensible to variations in z0 than in d.

O incêndio florestal é uma complexa combinação da energia liberada na forma de calor devido à combustão dos produtos oriundos da pirólise da vegetação e o transporte dessa energia para o ar e para a vegetação à sua volta. O primeiro é o domínio da química e ocorre na escala de moléculas e o segundo é o domínio da física e ocorre em escalas de até quilômetros. É a interação desses processos sobre uma ampla gama de escalas temporais e espaciais envolvidas no incêndio florestal que faz a modelagem do seu comportamento uma tarefa tão difícil. A propagação do incêndio através de vegetação rasteira e folhas mortas foi simulada numericamente usando a formulação física do WFDS. A abordagem utilizada foi tridimensional e transiente, e baseada em uma descrição dos fenômenos físicos que contribuem para a propagação de um incêndio de superfície através de uma camada de combustível. Neste cenário de incêndio, existem duas regiões: vegetação e ar, cada uma com suas propriedades físicas e químicas e, embora elas precisem ser integradas no mecanismo de solução, há diferentes fenômenos que ocorrem em cada uma. Na região de vegetação, a abordagem é representá-la como partículas submalha cercadas de ar. O caráter heterogêneo da vegetação, como sua natureza, folhagens, pequenos galhos, etc. foi levado em conta usando propriedades físicas médias características da floresta amazônica. Os fenômenos na região de vegetação são a evaporação da sua umidade, a pirólise e a transferência de calor por radiação e por convecção. Na região do ar, a combustão com chama ocorre em um ambiente turbulento, onde as transferências de calor por radiação e por convecção desempenham um papel significativo. Para incorporar a radiação dos gases de combustão, o modelo físico emprega o método de volumes finitos, que resolve a equação de transferência de calor por radiação como uma equação de transporte para um número finito de discretos ângulos sólidos, e que pode ser usado em uma ampla faixa de espessuras óticas e meios participantes. A combustão turbulenta para a fase gasosa é modelada com base no modelo Eddy Dissipation Concept (EDC). O modelo de combustão turbulenta adota a hipótese de reação química infinitamente rápida entre o combustível e o ar e é controlado apenas pela velocidade de mistura desses reagentes. Esse modelo representa bem a física de incêndios em ambientes ventilados, como é o caso dos incêndios florestais. Para incluir os efeitos do transporte turbulento é utilizado o método Large Eddy Simulation (LES), que calcula explicitamente as grandes estruturas turbulentas, mas trata a dissipação e a cascata inercial em escalas menores usando aproximações na escala submalha. As regiões de vegetação e ar trocam massa e energia. O comportamento da mistura gasosa resultante da degradação térmica da vegetação e das reações de combustão é regido pelas equações de Navier-Stokes. As equações que regem os modelos físicos são formuladas como equações diferenciais parciais que são resolvidas por métodos numéricos. O método utilizado para discretização das equações é o método de diferenças finitas em malha deslocada. O modelo numérico utilizado resolve as equações de Navier-Stokes para fluidos compressíveis usando o filtro de Favre. A dissipação de energia cinética é obtida através de um fechamento simples para a tensão turbulenta: o modelo de coeficiente constante de Deardorff. O transporte turbulento de energia e massa é contabilizado pelo uso, respectivamente, de números de Prandtl e de Schmidt turbulentos constantes. Os resultados das simulações do modelo físico descrito foram comparados aos dados experimentais obtidos em campo para a propagação do incêndio na floresta amazônica. Apesar da idealização das condições de combustível, vento e as incertezas dos dados experimentais, as previsões do modelo estão na mesma ordem de grandeza dos experimentos. As taxas de propagação do incêndio experimentais variam de 0,12 +/-0,06 a 0,35+/-0,07 m/min. Mesmo considerando-se o desvio padrão da taxa de propagação do incêndio experimental, os valores das taxas simuladas ficaram dentro do erro experimental somente em dois de sete casos. As simulações mostraram que os parâmetros importantes para o modelo são a área superficial por volume da vegetação, sua massa específica aparente e sua umidade. Como o coeficiente de absorção por radiação é função direta da massa específica aparente e da área superficial por volume da vegetação, esses parâmetros afetam o comportamento numérico do incêndio de superfície. De acordo com os resultados das simulações numéricas, a umidade da vegetação também tem importância no incêndio de superfície. A temperatura inicial da vegetação e a umidade do ar na faixa de variação analisada não influenciam a taxa de propagação do incêndio. As simulações também mostraram que o processo de radiação é muito importante, e afeta diretamente todos os demais processos e a taxa de propagação do incêndio. A convecção tem importância muito menor que a radiação na condição de ausência de vento externo. A coerência das taxas de propagação do incêndio experimental e numérica em função da massa específica aparente de material combustível e da umidade da vegetação foi investigada. O modelo numérico é coerente em todas as nove combinações de casos. Já o experimento é coerente em quatro combinações. Com base nas comparações entre cada dois casos experimentais e as respectivas simulações numéricas, nota-se que as taxas de propagação a partir das simulações numéricas foram mais coerentes que as experimentais. / Forest fire is a complex combination of energy released as heat due to the combustion of the products from the vegetation pyrolysis and the transport of this energy to the surrounding air and vegetation. The first is the domain of chemistry and occurs on the molecular scale, and the second is the domain of physics and occurs at scales up to kilometers. It is the interaction of these processes on a wide range of temporal and spatial scales involved in forest fires that makes modeling its behavior such a challenging task. The spread of fire through small plants and dead leaves was simulated numerically using WFDS physical formulation. The approach used was three-dimensional and transient, based on a description of the physical phenomena that contribute to the spread of a surface fire through a layer of fuel. In this fire scenario, there are two regions: vegetation and air, each one with its physical and chemical properties and, although they need to be integrated into the solution mechanism, there are different phenomena that occur in each one. In the vegetation region, the approach is to represent it as subgrid particles surrounded by air. The heterogeneity of the vegetation, such as its nature, leaves, twigs, etc. was taken into account by using average physical properties that are representative of the Amazon forest. The phenomena in the vegetation region are the evaporation of its moisture, pyrolysis, heat transfer by radiation and convection. In the air region, the flaming combustion occurs in a turbulent environment, and heat transfer by radiation and convection play a significant role. To incorporate the radiation from the combustion gases, the physical model employs the finite volumes method, solving the radiation transfer equation as a transport equation for a finite number of discrete solid angles, which can be used in a wide range of optical thicknesses and participating media. Turbulent combustion for the gaseous phase is modeled using the Eddy Dissipation Concept (EDC) model. The mixing controlled turbulent combustion model adopts the assumption of infinitely fast chemical reaction between the fuel and air. This model represents well the fire physics in ventilated areas, as is the case of forest fires. To include the turbulent flow effects, it is used the Large Eddy Simulation (LES) method, which explicitly calculates the large turbulent structures, but models the dissipation and inertial cascade using approximations in the sub-grid scale. The vegetation and air regions exchange mass and energy. The behavior of the gas mixture resulting from the vegetation thermal degradation and combustion reactions is governed by the Navier-Stokes equations. The equations governing the physical model are formulated as partial differential equations, which are solved by numerical methods. The method used for discretization of the equations is the finite difference method on a staggered grid. The numerical model solves the Navier-Stokes equations for compressible fluids using the Favre filter. Dissipation of kinetic energy is achieved through a simple closure for the turbulent stress: the constant coefficient Deardorff model. The turbulent transport of heat and mass is accounted for by use of constant turbulent Prandtl and Schmidt numbers, respectively. The physical model simulation results were compared to experimental data obtained in the field for the spread of fire in the Amazon forest. Despite of the idealized conditions of fuel, wind and the uncertainties of the experimental data, the model predictions and the experiments are in the same order of magnitude. Experimental rate of spread range from 0.12 +/- 0.06 to 0.35 +/- 0.07 m/min. Even considering rate of spread experimental standard deviation, simulated rate values were within experimental error only in two of seven cases. The simulations showed that the important parameters for the model are the vegetation surface area to volume ratio, its bulk density and moisture. As the radiation absorption coefficient is a direct function of vegetation bulk density and surface area to volume ratio, these parameters affect the numeric behavior of the surface fire. According to the numerical simulations results, vegetation moisture is also important in the surface fire scenario. Vegetation initial temperature and air humidity in the range analyzed does not influence the rate of spread. The simulations also showed that the radiation process is very important and directly affects all other processes and rate of spread. Convection heat transfer has much less significance than radiation heat transfer in the absence of external wind. The consistency of the experimental and numerical rate of spread, as a function of combustible material bulk density and vegetation moisture was investigated. The numerical model is consistent in all nine case combinations. The experiment is consistent in four cases. Based on comparisons between each two experiments and their numerical simulations, it is noted that the rate of spread variation from the numerical simulation is more consistent than the experimental one.

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1982. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Includes bibliographical references. / by Sun Tong Lau. / M.S.

Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1982. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Vita. / Includes bibliographical references. / by David Michael Papurt. / Ph.D.

Para vegetação esparsa e de porte alto a determinação dos parâmetros de rugosidade é comprometida pela dificuldade em se observar condições que satisfaçam a lei logarítmica da velocidade do vento. Estimou-se o comprimento de rugosidade (z0) e o deslocamento do plano zero (d) por alguns métodos com medidas micrometeorológicas e da estrutura física de arbustos esparsos em região semi-árida, durante o experimento HAPEX-Sahel. A velocidade do vento foi medida em quatro alturas acima da superfície (3,0; 4,1; 5,3 e 8,5 m), e os fluxos determinados por correlações dos turbilhões a 9m de altura. Métodos baseados no perfil logarítmico foram aplicados em condições de atmosfera neutra. A altura média da vegetação era h = 2,06 ± 0,47 m. O método convencional (ajuste estatístico) resultou em estimativas satisfatórias de d e z0 em condições nas quais a validade do perfil logarítmico foi satisfeita. Com uma única altura de medida localizada acima da subcamada inercial as estimativas resultaram em valores ou fisicamente inconsistentes ou que não caracterizam a rugosidade da superfície. Quando se utilizou a velocidade de fricção dada pela correlação dos turbilhões na solução do perfil logarítmico, as estimativas melhoraram. A combinação do perfil logarítmico com a relação z0 = λ (h - d) proporcionou estimativas satisfatórias para os valores de λ = 0,188 e 0,190 determinados em função da estrutura física da vegetação, o que não foi observado para o valor médio da literatura (0,166). Relações entre a estrutura física da vegetação e o transporte de momentum estimaram apropriadamente d e z0. A rugosidade da área foi melhor descrita por d = 0,95 m = 0,46 h e z0 = 0,204 m = 0,1 h, sendo λ = 0,185. As velocidades horizontal do vento e de fricção foram mais sensíveis a variações em z0 do que em d. / For sparse and tall vegetation the estimate of roughness parameters is compromised by the difficulty in observing conditions that satisfy the windspeed logarithmic law. The roughness length (z0) and the zero-plane displacement (d) were estimated by some methods with micrometeorological measurements and the physical structure of sparse shrubs in semi-arid region, during the HAPEX-Sahel experiment. The wind speed was measured at four heights above of surface (3.0, 4.1, 5.3 and 8.5 m), and the turbulent flows determined by eddy correlations at the height of 9m. Methods based on the logarithmic profile have been applied in neutral atmosphere conditions. The average height of the vegetation was h = 2.06 ± 0.47 m. The conventional method (statistical fit) resulted in good estimates of d and z0 only under conditions of validity of the logarithmic law. Only one height of measurement located above of the inertial sublayer is enough to result in physically inconsistent values. When the friction velocity, given by eddy correlation, was used in the logarithmic law, the estimates improved. The combination of the logarithmic law with z0 = λ (h - d) provided satisfactory estimates of the surface roughness for λ = 0.188 and 0.190 determined in function of the physical structure of the vegetation; but for λ = 0.166, the average value of literature, the estimates where not good. Relationships between the physical structure of the vegetation and the momentum transfer estimated appropriately d and z0. The area roughness was better described by d = 0.95 m = 0.46 h and z0 = 0.204 m = 0.1 h, being λ = 0.185. Wind speed and friction velocity were more sensible to variations in z0 than in d.

Based on a Ph.D. thesis in the Dept. of Electrical Engineering, 1970. / Bibliography: p. 122-123.