ESIMPLE, a new pressure-velocity coupling algorithm for built-environment CFD simulations

• Autor(a) principal: Serra, Nuno
• Data de Publicação: 2021
• Outros Autores: Semião, Viriato
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.21/13747
• topic: Pressure-velocity coupling; Built environment simulation; CFD; Modelling technique

Built environments are major energy consumers and, therefore, tools supporting their efficient design and guaranteeing thermal comfort and indoor air quality are a key factor for energy, environmental sustainability and healthiness. This has been particularly stressed recently by the need to understand the phenomenon of transport of pollutants and/or pathogens leading to exhalation of droplets and aerosols in built environments from people potentially contaminated with the new coronaviruses. In the pursuance of such objectives, CFD procedures have been widely used as prediction tools due to its ability and flexibility in capturing the main features of built environment flows. On the other hand, CFD methods are supported by complex and timeconsuming calculation procedures, especially when used to predict heat and mass transport phenomena in built environment. A possible strategy to reduce computational time is the optimization of the pressure-velocity coupling. An extension of the SIMPLE algorithm (ESIMPLE) is proposed, and its performance compared with the well stablished algorithms, SIMPLE, PISO, SIMPLEC and LIMPO. For that, three test case scenarios are simulated: i) a cubical room (1 m3) with heated floor; ii) a small-scale room, with an occupant, mimicking an office room; and iii) a real-scale office room with an occupant. For the worst scenario, ESIMPLE yielded similar CPU-time required for convergence, and for the best scenario, a three times faster convergence rate was attained. Simultaneously, this newly proposed coupling scheme algorithm, yielded a lower number of iteration steps required for convergence, in 6 of the 9 simulated cases.