O Departamento de Engenharia Mecânica do IST acumula a experiência de várias décadas de empenhamento em actividades de ensino, investigação e desenvolvimento. A diversidade de competências intrínsecas a um corpo docente próprio qualificado e constituído quase integralmente por titulares do grau de doutor (cerca de 90 doutores) associadas à qualidade e variedade dos equipamentos experimentais que se encontram instalados nos seus laboratórios conferem ao actual Departamento de Engenharia Mecânica do IST a excelência indispensável a um ensino de elevada qualidade.


Seminar in Lisbon

19 Março 2015, 16:02 - Jorge Manuel Fontes Coelho

Título: "Incompact3d: a high order flow solver to tackle turbulence problems using supercomputers"

Sylvain Laizet
Department of Aeronautics
Imperial College London, UK

Local: sala de reuniões do DEM
Data: 7 de abril (3ª feira)
Hora: 14h:30 às 15h:30.

Simulating and understanding turbulent flows remains one of the most challenging problems in mechanics.Significant progress has been made recently using high performance computing, and computational fluiddynamics (CFD) is now a critical complement to experiments and theories in order to understand turbulentflows and how to apply them in various engineering contexts.Only very few codes for Direct and Large Eddy Simulations (DNS/LES) are capable of undertaking massivesimulations with several billion mesh nodes on thousands of computational cores. Most of them aresimulating idealized homogeneous, isotropic turbulence, using spectral methods with periodic boundaryconditions in at least two spatial directions. Engineering problems have more complex geometries and fullspectral approaches are not practical. In conventional CFD, especially in an industrial context, complexgeometries are usually treated using nonstructuredelement meshes, requiring loworderschemes andsophisticated tools for the generation of highly distorted meshes. The resulting accuracy is most of the timeincompatible with a detailed analysis of engineering problems. Note that the spectral element method seemsto be a very promising strategy to undertake complex problems with the spectral accuracy. However, usingthis technique on thousands of computational cores is a challenging task that requires important numericaldevelopments to conciliate accuracy, efficiency and scalability.In this talk, I will present an innovative numerical method which can reconcile accuracy, efficiency, versatilityand scalability using a Cartesian grid. Various examples will be shown in this talk such as fractalgeneratedturbulence, gravity currents in an open basin, impinging jets on a heated plate and a microjetdevice tocontrol a turbulent jet.

Todos os Anúncios


Seminários da ACCAII

19 Março 2015, 14:56 - Jorge Manuel Fontes Coelho

Título: "Electricity demand profile prediction based on household characteristics"

Orador: Joaquim Viegas,
Hora: 13 h - 14 h
Local: Sala de reuniões do piso 4 Mecânica III

The energy market has been significantly changing in thelast decade due to liberalization, growing competition between utilities,technological advancements, and policy towards green energy exploitation andhigher efficiency in energy conversion. Also, technological advancements inthe fields of metering, communications and computation are enablingutilities to monitor and save huge amounts of data from daily operation.This work proposes a methodology for predicting the typical daily loadprofile of electricity usage based on static data easily obtained fromsurveys. The methodology intends to: (1) determine consumer segments basedon the metering data using the k-means clustering algorithm, (2) correlatesurvey data to the segments, and (3) develop statistical and machinelearning classification models to predict the demand profile of theconsumers. The developed classification models contribute to make the studyand planning of DSM programs easier, provide means for studying the impactof alternative tariff setting methods and generate useful knowledge forpolicy makers.

Todos os Anúncios