Air Conditioning for Atrium Architecture With Method of Mechanical Ventilation

May 30, 2018 | Author: Malcolm Chan | Category: Ventilation (Architecture), Turbulence, Fluid Dynamics, Convection, Mechanical Fan
Report this link


Description

ASSIGNMENT 2 Air Conditioning for Atrium Architecture with Method of Mechanical VentilationMalcolm Chan Hao Xian UNIMKL-003434 Introduction Ventilation provides continuous fresh air within a building to maintain air circulation, remove odor, excessive moisture and heat. There are mainly two different types of ventilation introduced in buildings, natural and mechanical ventilation. Natural being one that takes advantage of natural air circulation and passive stack effect to transcend air to outdoor. Passive stack adopts the concept of convection which utilizes the buoyancy effect of hot air, where the air density within a building reduces and causes it to rise up. Mechanical represents methods that utilizes actuation such as fan or any electrical airconditioning system, hence the energy usage. The main interest regarding ventilation in buildings for the past decade is to design more efficient ventilation methods that save energy at the same time. This is the same reason why more and more new architectures are now experimenting with new natural ventilation methods to adopt a “green” approach by consuming less energy. However, there are certain disadvantages that are unavoidable in natural ventilations. The most significant is that natural ventilation relies solely on natural driving forces to drive the flow of fresh air through a building, which means that it is also affected by atmospheric temperature and humidity. Also, natural ventilation is not equipped with fans, therefore not a dynamic system. This means that if more occupants occupy a building, the system would not be able to respond to such a change, as fresh air inside a building will quickly diminish and heat spots will begin to develop causing stuffiness and high temperature. This article will be discussing a new method of building ventilation coupled with a specific architecture to provide a more dynamic ventilation system that is energy efficient at the same time. By using the same concept of convection, and passive stack, this method proposes a slight modification where the building architecture is of atrium type, equipped only with ventilation fans and utilizes “forced” convection, in which less dense, hot air is pushed out of the building by fresh air delivered into the building by the ventilation fans. The atrium architecture will provide the transport route for air discharge out of the building. Vents will be supplied in top floor of building to create draw to expel the air out. Figure 1: Schematic of Air Flow in Atrium Based Architecture Figure 1 presents the perfect example in which the fresh, low temperature air is being delivered into the building via mechanical devices; through turbulence and air circulations, will absorb heat from the building’s stagnant air. Increase in temperature reduces air density causing it to flow upwards and drawn out from exhausts or vents due to pressure difference. The advantages of such a system are that it retains the pros of natural convection which would reduce energy consumption, since the use of air-conditioning is now discarded. Instead, ventilation fans are used to provide a much better dynamic response towards change of temperature within the building. The main objective of the present study is to investigate numerically the unsteady performance of the proposed ventilation system of two different distinctive designs. The two architecture design for the ventilation systems are shown in Figure 2 and 3 which are almost identical to each other, with both having different amounts of inlet in the ground floor of the architecture. For comparison purposes, all dimensions of both designs are kept constant and are given in Figure 4. The model is a 3-dimensional cylindrical shaped building, 3-storey high, with atrium feature. Both Model A and B have 4 outlets in the third-storey but Model A has 4 inlets on the first storey while Model B only have one. Figure 2: Model of Two Different Architecture Design (A & B) Outlet Balcony 3rd Floor 2nd Floor Inlet 1st Floor Figure 3: Schematic Diagram for Two Different Architecture Design (A & B) d e c b a a (m) b (m) c (m) d (m) e (m) No of Outlets No of Inlets Model 1 9.6 0.1 0.3 4 2 4 1 Model 2 9.6 0.1 0.3 4 2 4 4 Figure 4: Dimensions of Domain Geometry Initial and Boundary Conditions At time = 0, the temperature of air within building and wall is 310 K everywhere. The initial inlet velocity is at temperature of 310 K. For time > 0, the temperature of air within building and wall is still 310 K. The temperature of the inlet air now reduces to 300K. Figure 5: Effect of Heat Flux and Inlet of Outdoor Air The wall for both models is constantly exposed to temperature of 310 K. This creates a temperature difference between the inlet air and room temperature. This closely resembles the real case scenario of ventilation problems where the heat spots generated within the building are mainly due to the excessive heat on building walls due to heat flux. On the other hand, the inlet flow is only of the outdoor air temperature which is lower than the wall temperature. This difference is utilized in the ventilation system to help absorb and expel hot air within the building. This is the reason why the inlet air flow is set to 300 K (27oC) while the wall temperature is slightly higher to resemble the effect of heat flux; 310 K (37oC). The inlet of the building is calibrated to a 1.5 m/s speed of inwards airflow for single fan system. The inlet is reduced to 0.375 m/s for 4 fan system. The outlet of the building is enabled as outlet points. This enables cool air below to replace hot air above and pushes it out. Operating Conditions Gravity is enabled in simulation. The working fluid selected is air with constant physical properties. The density is altered to Boussinesq model; Density = 1.225 kg/m3 Specific Heat (cp) = 1006.43 J/kgK Viscosity ( = kg/ms Thermal Conductivity (k) = 0.0242 W/mK Thermal Expansion Coefficient = 0.00333 K-1 This allows the buoyancy effect as well as heat transfer between fluids to take place. When air temperature rises, density will drop, hence mass of air will drop. When gravity is enabled in simulation, fluid stratification happens and hot air will rise upwards while cooler air will replace the predeceasing hot air. This feature allows convection; forced or natural to happen in simulation. Mathematically, Boussinesq Approximation presents the density of fluid as a function of temperature. Density is treated as constant value for all solved equations except for buoyancy terms, in which the overall new density is expressed in terms of thermal expansion coefficient and original density value. Where is the density of the flow is the temperature difference is the thermal expansion coefficient Solutions Control PRESTO! is selected as the pressure discretiszation scheme as it uses the concept of staggered grid instead of co-located arrangement. Hence, pressure values are calculated on grid faces and interpolations are done naturally. This avoids interpolation errors and pressure gradient assumptions on boundaries. Because of this, PRESTO! is also extremely suitable for simulation of high Rayleigh number flows or natural convection. Since the nature of the case studied is of convection nature, PRESTO! discretization is preferred. PISO is used for pressure velocity coupling to ensure stability for long duration of unsteady simulation. Due to initial difficulties for simulation to converge, Second Order Upwind is selected as momentum discretization scheme. This would help reduce false diffusion as well as increase transportiveness of the domain since upwind schemes take into account the direction of flows. The remaining discretization remains with the default selection of first order upwind. This is to conserve unnecessary computation cost. The solution domain for both of the models as in Figure 4, were created and meshed by non-uniform spacing tetragonal cells. Both Model 1 and 2 have an approximate of 285582 cells with a maximum skewness of 0.75. The time step selected is 0.01 second. There are viable motives behind the use of small time step size, especially with 3-dimensional geometry as this will increase computational cost. The reason is that one of the turbulence model utilized is LES. LES is a technique that explicitly solves for large eddies and uses sub-grid scaling for solving smaller eddies implicitly. Explicit methods are well-known to only be conditionally stable, and in this case, the step size has to be small enough to ensure that the flow does not reach beyond the first cell in the domain within the first time step. Small step size will ensure stable time marching and improved accuracy. The simulation will be done in two sections in which both models, A and B will be simulated using turbulence model, K-Epsilon RNG and Large Eddy Simulation. After which, the findings will be compared to conclude on the performance of the system. Limitations The main limitation in this case study is the Boussinesq Approximation of fluid density. With the application of gravity and fluid stratification, the fluid can now move vertically according to changes in density, however, this rendered the main model solution impossible. This is because the initial solution was to feature a structure in which the outlet is formed on the ceiling of the structure. This enables the structure to be open air concept and improve ventilation. However, existing air within the structure refuse to exit the structure and drops down again once it reaches the outlet, and computation diverged. Therefore, the outlet was then changed to the side of the structure instead. Results and Findings Below are the temperature variations for all four cases in the centre of the structure on the second storey. 37.5 37 36.5 36 35.5 35 34.5 34 33.5 33 32.5 0 10 20 30 Time (min) 40 50 60 Temperature (Degree C) Temperature (Degree C) Graph 1: Temperature Trend for K-Epsilon RNG Single Fan System 37 36.5 36 35.5 35 34.5 34 33.5 33 32.5 32 31.5 0 5 10 Time (s) 15 20 25 Graph 2: Temperature Trend for K-Epsilon RNG 4 Fan System 38 Temperature (Degree C) 37 36 35 34 33 32 31 0 10 20 Time (s) 30 40 50 Graph 1: Temperature Trend for LES Single Fan System 38 Temperature (Degree C) 37 36 35 34 33 32 31 0 5 10 15 20 25 30 35 Time (s) Graph 4: Temperature Trend for LES 4 Fan System From the figures above, it can be found that the temperature reaches a saturation point after the temperature has been reduced to the point around 32 OC. After which, the temperature begins to fluctuate around the temperature range of 32 – 33 OC. This specific scenario may be due to the geometry of the model structure. It may cause fluid obstruction and the fluid propagation will absorb less heat in the structure. Since most of the time, structures are built and ventilation system is only installed afterwards to be adapted to the structure, nothing significant can be done to repair this effect. The only solution would be to adapt the ventilation system design in parallel to the architecture development stage. It may also be due to extreme intensity of wall temperature or that the temperature difference between the inlet and surrounding air is not large enough. These two will cause the inlet air to be incapable to absorb the required heat released through the wall of the structure. In other words, the specific heat capacity of the fluid used, air is not high enough. Hence, either a fluid of higher heat capacity has to be used, or to simply further decrease the inlet air temperature. The former is impossible since the limitation is air; therefore, the only solution to reduce the temperature further is to reduce the inlet temperature. It should be noted that although the inlet air is only capable to reduce the surrounding temperature for approximately 5 OC, the temperature that is released by the wall of the structure is at 310 K, which is extreme for most cases. This is because most structure has multiple layers of walls constructed from concrete, bricks, or cement. These will allow losses in terms of heat before the heat flux from external wall. If the surrounding indoor air temperature is lower than 310 K, the ventilation system may improve its performance. There is a possibility that the computation time is not enough, where there may be a further steep drop in temperature, however that possibility is very low, since the temperature has reached saturation for more than approximately 20 minutes for all four cases. It is unlikely for the temperature to experience a drop again. It can also be seen that for both LES and K-Epsilon RNG, there is not much drastic difference in terms of temperature change to time. Both turbulence model presents the same pattern and values in overall. Both computation results in rapid cooling of air within the same period of time. For LES and RNS single fan system, the temperature of air is drastically cooled until saturation within the first 20 minutes, where as for the 4 fan system instead, the temperature is cooled until saturation in 10 minutes time. This shows that the increase in amount of fans can help improve cooling time but not the performance of temperature reduction, which is consistent with the findings above regarding the working fluid’s specific capacity. Also note that for 4 fan system, the inlet air velocity is calibrated down to 0.375 m/s. This is to equate the air velocity of 1.5 m/s for a single fan system. This signifies that just the increase of amount of fans, without increasing the overall velocity of air, can improve air circulation and heat exchange because the interaction between the inlet and surrounding air is higher and hence improve the performance. Temperature Contours for K-Epsilon RNG Single Fan (Figure 6 to Figure 16) Refer respective figures for time step. Temperature Contours for LES Single Fan (Figure 17 to 27) Refer figures for respective time step. Temperature Contours for K-Epsilon RNG 4 Fan System (Figure 28 to 32) Refer figures for respective time step. Temperature Contours for LES 4 Fan System (Figure 32 to 39) Refer figures for respective time step. LES and K-Epsilon RNG For all the figures above, there is one important finding which is that there is a layer hot air trapped on the ceiling of the structure. This is because the outlet provided is located at the side walls of building, hot air that failed to escape through the outlets will proceed to rise above it and form a heat canopy at the top of the building. There is no significant effect to the occupants of the structure as heat exchange is still taking place at lower storey. However, the constant exposure of the building ceiling to heat can lead to thermal damage that may cause creep. This is very critical to the health of a structure, because creep can lead to the collapse and failure of the whole structure [1]. However, the usage of an outlet at the top of the structure would be able to solve this issue as the trapped air would be released. It can also be noted that after a period of cooling, all cases reaches a steady state. This is evident from the repetitive heat pattern for all cases. The contours of temperature for the cases are all similar for time steps after the steady state, where the first storey would be in the lowest temperature of 27-29 OC. The second storey on the other hand, remains at 31-33OC. On the third storey, the temperature is around 33-35 OC, while the ceiling is covered with a layer of hot air, as explained above. This pattern of temperature distribution remains the same even as the time steps continue marching. This is due to the exposure of wall to a constant and consistent heat flux, and the inlet airflow at a constant velocity. Referring back to Figure 10, 20, 30, and 34, it can be seen that the contours is consistent with the findings. This shows that the potential of the system is fulfilled within the first 10 minutes for 4 fan system, 20 minutes for a single fan system. This signifies that in the nature of such ventilation system where there is a constant heat source and inlet air, both with consistent temperature, the ventilation will reach a point where the pattern of temperature reduction will saturate, at which is also the maximum potential of the system. It can also be noted here that there is a distinctive difference between the two turbulence models utilized although both presented the same pattern of temperature change. It can be seen for LES model, the fluid motion from the inlet are more focused , in which the air remains as a core stream of low temperature air, even after entering the domain geometry for a certain amount of time steps. On the other hand, for K-Epsilon RNG, the fluid motion after the inlet is dispersive in nature. The temperature of the core stream increases its temperature in a short period of time. In overall, both turbulence model are similar, however, to conclude the better turbulence model for this specific case study is challenging due to the lack of any similar benchmark to compare with. Velocity Vector for K-Epsilon Single Fan System (Figure 40 to 50) Refer figures for respective time step. Velocity Vectors for LES Single Fan System (Figure 51 to 60) Refer figures for respective time step. LES It can be seen that the fluid from the inlet will propagate across the structure initially. This forms two significant eddies on the first storey of the building, bottom left, and the right hand side , as seen in the figures presented above. These two large eddies continue to grow as the time marches forward. The eddy on the right side of the structure has a more random propagation, in which smaller eddies start to grow within the main vortex. The fluid motion on the second and third storey of the structure appears to of random order and chaotic across the whole domain. The main pattern shows the fluid however, does move across the model vertically, through the sides of the entrance. The fluid then proceed into the third storey of the building by propagating on the ground of the third storey and then into the outlet. The fluid motion above the outlet appears calm. This is consistent with previous finding, as this layer of fluid represents the hot air trapped in the same location. Since there is not heat exchange taking place, there is no major movement in that location. The fluid also propagates upwards in the third storey up until the height in which the trapped hot air resides, and then proceeds to leave the structure through the outlet. In overall, the main pattern of fluid motion remain similar across time marching. The only difference being minor growth of vortices as the fluid propagates upwards. The velocity of the inlet air is also more significant, where the main core stream retains its velocity even after well into the time marching. Only the outer region of the inlet flow is affected by shearing, and losses its velocity. This phenomenon is similar to previous findings in which the core temperature of the inlet air does not vary much, where only the outer region experiences heat exchange. K-Epsilon RNG The fluid movement and pattern does not differ far from those in LES turbulence model. The fluid from the inlet, similarly, propagates across the structure, however not as much as the case of LES. It can also be seen that there are two major formation of eddies at the same location as described in LES case. However, the formation of eddies are more defined and significant, where the main circulating loop can be seen clearly, and there are no formation of any smaller eddies within, in contrary to those in LES. The fluid motion on the second storey is also similar to the case of LES, in which the fluid appears to be chaotic and random. The difference is that, permanent eddies begin to form in the corner of the model as seen in Figure 43 onwards. Eddies begin to form 20 minutes into time marching, and will proceed to resides in those locations. The propagation of the fluid across the structure on the second storey and third storey remain similar to the case of LES in terms of the main pattern, where the fluid moves upwards and on the ground of the third storey, before exiting through the outlet. The calm and slow layer of trapped hot air on the top of the building remains at the same location as with the LES case. The velocity of the inlet air is in the contrast to the case of LES. The main core stream is dispersive, and quickly losses its velocity magnitude after time marching begun. It can be commented that the core flow is heavily influenced by shearing and fluid friction, which is why the velocity magnitude decreases rapidly. This is consistent to previous comments on the temperature change of the core flow. Velocity Vector for K-Epsilon 4 Fan System (Figure 61 to 67) Refer figures for respective time step. Velocity Vector for LES 4 Fan System (Figure 68 to 74) Refer figures for respective time step. LES and K-Epsilon RNG For both LES and K-Epsilon RNG, it can be seen that the fluid from the inlet propagates across into the centre of the structure. This forms two significant spots where eddies form vigorously on the first storey of the building, bottom left, and right, as seen in the figures above. These two large eddies continue to grow as the time marches forward. After which, the motion will propagate upwards from the centre through the opening to the second storey. The difference is that for LES, smaller random eddies begin to form in these spots of large eddies. In the first storey, the fluid motion for K-Epsilon RNG appears to be more orderly and organized, and the fluid decreases rapidly in velocity magnitude, whereas for the case of LES, it can be seen that the fluid propagation is more chaotic and random. Again, the core stream retains its high velocity magnitude as well. The propagation of fluid upwards to second storey is of random pattern and not vertically upwards as in the case of K-Epsilon RNG. On the second storey, the main pattern of fluid motion shows the fluid moving across the second storey of the model vertically, through the sides of the entrance. The fluid then proceeds into the third storey of the building by propagating on the ground and then upwards by the wall of the second storey. In the third storey, the fluid continues to move on the ground of the third storey before exiting though the outlet. The fluid motion after the outlet also appears to be calm. This is again similar with previous finding, where this layer of fluid represents the hot air trapped in the same location. Initially, within the first 5 minutes of time marching, the fluid within the third storey of the structure experiences turbulence and eddies were formed before the fluid exits through the outlet. However, after the initial period of 5 minutes, the turbulence and chaotic condition disappears from the third storey, and the top height resembles the condition of trapped air as explained before. This could be best explained that the fluid propagation attained steady state condition only after the initial 5 minutes, where the fluid is still experiencing drastic changes before saturation due to heat transfer. In overall, the main pattern of fluid motion remain similar across time marching. The only difference being minor growth of vortices as the fluid propagates upwards. The velocity of the inlet air is also more significant, where the main core stream retains its velocity even after well into the time marching. Only the outer region of the inlet flow is affected by shearing, and losses its velocity. This phenomenon is similar to previous findings in which the core temperature of the inlet air does not vary much, where only the outer region experiences heat exchange. For all cases, it can be seen that the fluid motion, at most of the time propagates across the model structure and continuously experiences turbulence. Considering that the case study is regarding a ventilation system for an occupants’ building, the erratic behavior of fluid motion may serve to be disturbance for occupants within the structure. However, it should be noted that turbulence does not necessarily takes place within high velocity condition, which is exactly the case of the turbulence in discussion. Most of the turbulence that takes place within the domain has very low velocity, especially for the case of 4 fan system. Therefore, in real life scenario, the air motion should be well within comfort level for occupants within the structure. Figure 75: Velocity Vector for K-Epsilon RNG 4 Fan System at 1.2m height from ground Figure 76: Velocity Vector for LES 4 Fan System at 1.2m height from ground Figure 77: Velocity Vector for K-Epsilon RNG Single Fan System at 2.4m height from ground Figure 78: Velocity Vector for K-Epsilon RNG Single Fan System at 4.8m height from ground Figure 79: Velocity Vector LES Single Fan System at 2.4m height from ground Figure 80: Velocity Vector for LES Single Fan System at 1.2m height from ground LES and K-Epsilon From Figure 75, it can be seen again that the nature of K-Epsilon RNG computation produces a more orderly turbulence, in which, at every inlet from all four sides, vortices form on each two sides of the core stream flow. The patterns of these vortices are closely similar between each inlet. At the same location for the LES case, the vortices formed as viewed from Figure 76, are all random in order. These vortices are chaotic and present no apparent or recognizable pattern which is presented in K-Epsilon RNG case. From Figure 77, the vortices beneath the core flow are formed on both sides of the flow, where there is fluid circulation. At the centre of the whole domain of structure, it can also be seen that two major vortices form circulation around both sides of the structure as seen from Figure 78. As for the case of LES, comparing Figure 79 to Figure 77, it can be seen that there are traces of vortices forming on both sides of the core stream flow similar to the case of K-Epsilon RNG. However, again, the nature of the turbulence is random and unpredictable. Comparison of Turbulence Model In overall, both the turbulence model of LES and K-Epsilon RNG would be able to provide a proper numerical solution of the case study in discussion with an acceptable accuracy. However, the turbulence model LES presents sets of results that are more chaotic and random in nature. The fluid motion and heat exchange that takes place in the domain geometry are unpredictable and presents no apparent pattern. For K-Epsilon RNG, the results presented, although are also of turbulence and chaos, however are more orderly as compared to LES. There are also numerous occasions where there are certain recognizable patterns of fluid motion that are repetitive in nature. These findings are consistent to the concept of both the turbulence model used. Large Eddy Simulation operates by solving the large eddies explicitly while using sub-grid scaling for solving smaller eddies implicitly. The filtered flow variables such as pressure and velocities are of random nature. In contrast, for cases of Reynolds Decomposition, K-Epsilon RNG for an example, the pressure and velocities are non random and based on fluctuating part of the flow. From a more technical perspective, LES turbulence model would be a more suitable choice for the numerical solution of the case study as it provides fluid motion that are of random order. In addition, the model discards any representation of closure by using eddy viscosity, instead, the Reynolds Stress are directly computed, improving its accuracy. However again, due to lack of any benchmark publications, the true potential or comparison between the two turbulence models is unknown. The more strategic approach would be to select a more accurate turbulence model. It should be noted that due to the complexity of solving Reynolds Stress directly, as well as extremely small time steps, computation cost is higher. Therefore, there is a certain tradeoff between computation cost and accuracy. Comparison of Ventilation System Design Comparing between the two designs of ventilation system, in which there is an increase in the amount of inlets on the first storey of the model structure, the 4 fan system is deemed to be a more efficient design and has a higher level of performance as compared to the single fan system. This is proven from the findings that the increase of fan inlets can help accelerate the process of heat exchange even though the velocity of the inlets were lower than the case of a single fan system. Moreover, due to the low velocity of the inlet, the fluid propagation across the structure has a lower velocity magnitude making it less noticeable and more comfortable for the occupants residing within the structure. However, there is one aspect of the performance of the ventilation system that both fails to address, and that is total cooling of the whole structure, as the fluid attains a steady state condition once temperature reduction reaches saturation. As discussed before, using fluid of higher heat capacity is impossible. Using air at a lower temperature is also not valid, since the outdoor air temperature is beyond manual control. This means the only way is to calibrate the design of architecture. The most direct solution is to elevate the height of the structure, where the air of higher temperature would proceed to float upwards to the higher ceiling which is far from any occupants. This would increase the span of area with a lower and acceptable temperature. This also signifies that this concept of ventilation system should be integrated simultaneously into the architecture process in order to design an efficient cooling process. Conclusion In conclusion, the application of a forced convection concept as a form of ventilation system is a viable one. It provides an alternative to save energy consumption while able to condition surrounding air to a comfortable level. The performance of the studied systems above presents encouraging results where rapid cooling is achievable in large scale buildings up to a height of 9.6 meter. Findings also direct to possibilities of improvements where temperature reduction can be amplified by either having cooler air or changes to the basic architecture layout of the structure. The findings also presents that LES turbulence model presents results that are more random and chaotic, while K-Epsilon RNG’s results present predictable patterns and less random. In this case, without any benchmark references, the more strategic choice would be to prefer LES turbulence model in this nature of study. References [1] Bazˇant, Z. P., & Yong, Z. (2002). Why Did the World Trade Center Collapse?—Simple Analysis. JOURNAL OF ENGINEERING MECHANICS , 2-6. [2]CFD Online http://www.cfd-online.com/Wiki [3] Fluent 6.3 User’s Guide, Fluent Inc. http://www.fluentusers.com [4] H. K. Versteeg and W. Malalasekera, “An introduction to Computational Fluid Dynamics,” Longman, New York, 1995. [5] S. V. Patankar, “Numerical Heat Transfer and Fluid Flow,” McGraw-Hill, New York, 1980.


Comments

Copyright © 2024 UPDOCS Inc.