What impact does a resistance path-way have on zonal air flow in naturally ventilated buildings?
David Cunningham, Loughborough MRes EDS
Historically all buildings were naturally ventilated, yet with buildings reaching ever skywards and becoming more complex, alternatives to natural ventilation methods were sought. As a result there was a shift towards using air conditioning units in a bid to optimise the internal quality, particularly in terms of comfort and temperature (P. F. Linden 1999). Thankfully, from an energy demand reduction perspective, rising concerns about global warming in the 1990s reawakened the interest in naturally ventilated offices (Gratia & Deherde 2007). The primary purpose of natural ventilation is to provide acceptable internal air quality, as opposed to providing fresh air for respiration. Individuals require approximately 7.5 litres of fresh air per second for respiration; while typical air changes needed for thermal comfort require at least ten times this amount (P. F. Linden 1999).
Heating and Ventilation Air Conditioning (HVAC) units are commonplace in offices and multi-occupancy buildings; with nearly 68% of the total energy used in service and residential buildings attributable to these common systems (Stavrakakis et al. 2008), but with the drive to reduce the demand for energy, designers and engineers are now attempting to return to the utilisation of natural ventilation wherever possible.
The aim of this research is to investigate the effect of partitioning in naturally ventilated spaces. Natural ventilation is proven to work well within simple, open-plan geometry; with more complicated buildings however, there is a greater need to partition the space to provide different zones. It is important to develop an intuition for the way air moves around a partitioned naturally ventilated building, and how this is affected by changes in the design and in the external conditions (Linden, 1999). It is within the domain of change in design that the current project is situated.
Principal motivation for this research is the reduction of energy demand in buildings and the main objectives are to provide insight for building designers regarding the relationship between natural ventilation performance and partitioning; and provide quantifiable data on the effect of partitioning in naturally ventilated buildings.
A naturally ventilated building which is operating correctly, can typically consume between 40% (Stavrakakis et al. 2008) and 60% (Hunt & P. F. Linden 2001) less energy than that of comparable air-conditioned buildings. This energy demand reduction is achieved, in part, by using natural ventilation to purge some of the heat contained in a building during summer conditions, which has been found to be both practical (Coley 2002) and very energy efficient (Priyadarsini 2004; A. W. Woods et al. 2009).
It has been shown that high-rise buildings can be completely naturally ventilated (Pasquay 2004) providing it is acceptable for the internal and external temperatures to be the same occasionally. Typically, air entering a building is heated to a temperature of 17–19oC prior to being supplied to the occupants; with an interior temperature being maintained in the region 21–22oC (A. W. Woods et al. 2009). Occupants of naturally ventilated buildings have a propensity to tolerate a larger range of temperatures than in air-conditioned buildings (Emmerich et al. 2011), so it is likely that the proviso of similar external and internal air temperatures will be acceptable.
The most common models to predict the performance of naturally ventilated buildings, are computational fluid dynamics (CFD) models (Stavrakakis et al. 2008). However CFD models rely on simplifications and numerical equations to represent fluid movement in a building. It can sometimes be more appropriate and beneficial to use physical modelling techniques such as salt-bath modelling. This provides a physical simulation of fluid movement rather than numerical. Salt-bath modelling is also referred to as the ‘filling box’ model (Germeles 1975), or the `emptying water-filling box’ model (Li 2000).
In the early design stages of a naturally ventilated building, salt-bath modelling can be used to identify the stratification level in any given space. This stratification level is the warm upper layer of air that due to its lesser density sits on top of the cooler, denser, fresh air that is brought into the building. In a study looking at the effects of thermal radiation on airflow with displacement ventilation (Li et al. 1993), it was found that vertical temperature stratification is a combined effect of convection (both natural and forced), conduction and thermal radiation.
Ventilation flow rate governs the temperature gradient in a stack driven ventilation system (Mundt 1995), which in turn is controlled by the size and positioning of the openings in the building; if the openings are not sufficiently large, or erroneously situated, the stratification layer will descend into occupied regions (Hunt & P. F. Linden 2001). It has been found (Coley 2002) that ventilation rates will increase at a power of 0.5, as either vertical separation between the vents grows, or the difference in temperature between the inside and outside air increases. Internal stratification and flow regimes are also affected by the presence of an exhaust stack attached to one of the openings, namely the high level exhaust opening (S Fitzgerald & a Woods 2008).
Creating a scale model
The SSEES building opened in 2005, it is a naturally ventilated building situated on Taviton Street in London. It is one of the schools that make up University College London (UCL). The building was designed by Short and Associates, and resembles a ‘D’ shape in ‘plan’ view with a light well at its centre. SSEES operates with a centre-in, edge-out form of stack ventilation; meaning fresh air is brought into the centre of the building through a light well type structure (Lomas & Yingchun Ji 2009), and flows out towards the edge of the building. It is naturally ventilated year-round; a process made possible through the use of ‘downdraught cooling’ during the spring/summer which cools the air as it enters the building.
It was decided that a section of the 3rd floor would be most suitable for analysis, as it is a mid-level floor and therefore provides typical stack pressures, and it is heavily partitioned. The area chosen was split into 4 zones consisting of a segment of the open plan research area (zone 1 – identified by a number 2 in Figure 1), a portion of the corridor between the research room and academic offices (zone 2), the space inside the acoustic corridor (zone 3), and two cellular academic offices (zone 4 – identified by a number 3 in Figure 1). Figure 1 also identifies the exhaust stack of the building with a number 4.
For this project the partition at the research room is referred to as resistance pathway 1, and the acoustic corridor that partitions the academic offices referred to as resistance pathway 2 (identified in Figure 1 by a blue and green arrow respectively).
The scale model was produced to represent the SSEES building section with an approximate scale of 1:50, resulting in an exhaust stack of 60cm in length. It was deemed most appropriate to use this scale, in relation to the water tank being used (1.4mx1.6mx1.0m made of Perspex 0.03m thick). To make the study more generalised to resistance pathways in naturally ventilated spaces, it was decided that the opening sizes of the inlet and acoustic corridor would be equal to approximately 2% of the total floor area. The exhaust opening sizes were scaled from the actual exhaust windows in the SSEES building, and are larger than 2% of the floor area. Doing this further focuses the study on the partitions and occupied areas.
For this experiment an opening area of 1.05m2 is converted to 1.05cm2 and then multiplied by 5 to give a 1:50 scale. Meaning, the circular inlet opening in this model needs to be 5.25cm2 with a diameter of 2.585cm. With the acoustic corridor, to achieve an area of 2% of the total floor area, each opening would need an area of 1.31cm2 and a diameter of 1.293cm. Resistance pathway 1 has rectangular openings at the top and the bottom of the partition, with a scale area of 25.4cm2. The outlet opening leading to the exhaust stack, situated in each of the academic offices, has an area of 1.51m2 which is scaled to a diameter of 3.1cm.
In reality, it is not possible for the scale model to have the exact opening sizes as calculated. The openings will be made in the model using a drill and therefore the openings will be determined by the drill bits available. Table 1 lists the opening sizes used for the inlet, resistance pathways and the exhaust openings.
Table 1 – The table shows the size and area of each opening in the scale model
Salt-bath models consist of a scale model being immersed in a fresh water tank, with heat sources represented by saline solution, introduced through a nozzle (Baines & Turner 1969) fed by a header tank. The saline solution is created by inserting 500 grams of sodium chloride into a small mixing tank attached to the mechanical pump. 228ml of blue food colouring is added to the mixing tank, followed by approximately 20 litres of water. The pump is then turned on for 20 minutes, with the valves closed, to allow the water, food colouring and sodium chloride to thoroughly mix. Once mixed, the saline solution is forced by the pump against gravity, to the constant header tank which is situated at a higher level than the water tank. The reason for this is it allows the saline solution to enter the water tank using gravity as the lone driver.
The saline solution has a greater density than that of fresh water, similar to the density differences found between warm and cool air. It is these differences in temperature, that create the small pressure drops that are required to drive ventilation flows between two chambers (Tovar et al. 2007). Delivery of the saline solution through a nozzle means that a turbulent plume is achieved, the behaviour of which agrees well with the behaviour of a plume generated by a heat source as predicted in CFD modelling (Y. Ji & Cook 2007). This could be deemed essential, as it has been found that a turbulent plume in a confined region, can lead to stratification of the fluid surrounding the plume (Worster & Huppert 2006).
Using a saline solution allows the model to achieve dynamical similarity, as the effects of friction and diffusion can be appropriately scaled when brine is used to create density differences (Hunt & P. F. Linden 2001). Further advantages of using salt-bath modelling are that flow visualisation is straightforward and complex flow patterns can be easily determined (P. F. Linden et al. 1990). Furthermore, visual images obtained during the experiments can be used to provide validation data for simple mathematical models (Sandbach & Gregory F. Lane-Serff 2011) and clear physical explanation for the phenomenon of ‘transport against the gradient ‘ (Baines & Turner 1969).
There are 2 things that can affect the plume created in a salt-bath model; flow rate of the saline solution (maintained at 300 cc3/m or 0.005 l/s in this research) and reduced gravity – the buoyancy force which is defined as g’:
g’= g. ∆ρ/ρ = g. ∆T/T
Where g is acceleration due to gravity, ∆ρ/ρ is the fractional change in density, caused by a change in temperature ∆T/T (P. F. Linden 1999).
The salt-bath experiments were carried out on five different configurations of the scale model. Run one was simulated with no partitions inserted, creating an open plan building space, while run two was simulated with resistance pathway 1 inserted. The third run resistance pathway 1 for resistance pathway 2, and run four was simulated with both resistance pathways inserted in conjunction. For the fifth simulation, both resistance pathways remained in place but only one heat source was simulated. Stratification levels in each zone were recorded and are stated in proportion to the room height, ; where h is the stratification height from ground level, and H is the total height of the rooms.
Figure 3 shows the stratification levels that were recorded during the first salt bath experiment, on the open plan layout of the scale model. As such, the results shown in Figure 3 are the base case stratification levels. Comparisons between the base case stratification levels and those found in subsequent experiments show the effect of partitioning on the model.
During the filling stage of the experiment, the saline solution reached a stratification level at 0.47 of the room height after the first minute. By the second minute of filling the solution had filled 0.60 of the room, and levelled at 0.63 of the total room height in the third minute. The stratification level remained steady at this level, deepening in colour as the test went on. After the fourteenth minute of the experiment the saline solution reached a steady state at 0.67.
Once the saline solution had reached a steady state, and the stratification level was stable, the delivery system was switched off and the model was allowed to empty. The rate at which the model cooled was significantly slower than the rate at which it filled; with the stratification level rising to 0.47 of the total room height in the first minute.
The graph shows the stratification levels throughout the filling and emptying phases of the first model run.
Within five minutes of the emptying simulation the saline solution filled 0.33 of the room, lowering to 0.27 after the tenth minute; by the fifteenth minute the saline solution occupied only 0.17 of the model height. The simulation reached a steady state after nineteen minutes, resulting in a stratification level 0.13 of the total room height, a level at which it remained through the rest of the simulation.
Run 2 – Resistance pathway 1 inserted
Similar results were found during the second run of salt-bath modelling as were found in the first run. The rate of stratification increase in this run is almost identical to that of the first, with the saline solution filling 0.67 of zone two after only three minutes. During the first two minutes zone one fills at a slightly faster rate (5cm/min) than zone two (4.5cm/min). After this point, and up until the sixteenth minute, zone two has a lower stratification level than zone one. It is thought that the deeper stratification level in zone two is due to the saline solution in this zone being prevented from fully mixing with the fluid in zone one.
During the emptying phase of the modelling, it was found that zone one empties faster than zone two over the initial four minutes; but the emptying rates of each zone became equal after this point. This could be due to the saline solution that enters the exhaust stack from zone two being replaced by the solution in zone one. After five minutes of emptying the stratification levels had gone from filling 0.67 of each zone to filling only 0.33 of each zone. Furthermore, after ten minutes of the model emptying, the saline solution occupied only 0.23 of the room height in each zone.
Run 3 – Resistance pathway 2 inserted
A greater effect on the rate and level of stratification was observed when the salt-bath model was run with the acoustic corridor inserted, as opposed to the simulation with only resistance pathway 1 inserted. The overall rate of stratification was slower than the previous two runs; after the first minute saline solution occupied 0.33 of zone three. At the same point in the simulation, saline solution filled 0.20 of zone one and only 0.07 of zone two.
After seven minutes of running the filling phase of the simulation, the stratification in zones one and two had levelled out at 0.53 of the total room height. However zone three still had a deeper stratification level, filling 0.60 of the zone. By the eighth minute the stratification layer in zone two had increased beyond that of zone one and was level with zone three at 0.60; the stratification in zone one remaining 0.53 full of saline solution.
The stratification levels in zones one and three meet after ten minutes of simulation. With the exception of a slight deviation in the eleventh minute, the solution occupies 0.67 of each of these zones for the remainder of the model simulation. However the stratification layer in zone two continues to deepen, steadying at 0.73 of the partition height. Eventually the stratification layer in zone two settles after the seventeenth minute, filling approximately 0.77 of the space.
During the emptying stage of the model run the cooling of the zones one and three occurs at the same rate for the first minute. In the fifth minute, saline solution occupied 0.47 of zone one, but only 0.33 of zone three. Zone three continues to empty at a faster rate than zone one, until the tenth minute when the stratification level is maintained momentarily. In the fourteenth minute the layers of saline solution in zone three and zone one converge at 0.23 of the room height. The stratification levels settle after the nineteenth minute, leaving 0.13 of the model containing saline solution.
Run 4 – Both resistance pathways in conjunction
Within the first minute of filling, zones two and four have a stratification layer that is 0.33 the depth of the room height. In comparison zone one level is 0.20, and zone three has a level of 0.13. The difference between the stratification levels across the zones becomes even more apparent in the third minute of filling; when levels get to 0.50 in zone one, 0.73 in zone two, 0.27 in zone three and 0.63 in zone four.
After running the model for five minutes the stratification levels in zone two levelled at 0.87 and remained at this level throughout the experiment. By this point the stratification layer in zone four had also reached a steady state and settled at 0.67. However zone one was still filling in a relatively uniform manner and had a stratification depth of 0.57 of the total room height, with zone three having a level 0.47 of the room depth. The stratification levels throughout the filling stage of this simulation are shown below.
Graph shows the varying stratification levels in each of the 4 zones during the filling stage of the scale model.
Between the fifth and eighth minute the stratification level inside zone three increases at a rate of 2cm every minute where it meets the level recorded in zone two. After ten minutes of simulation the levels observed in zone one become equal to the steady state level of zone four at 0.67 of the room height, with zones two and three having saline solution layers up to 0.87 of the room depth. No further changes in the stratification layers occurred across the model, with the exception of zone three thinning out slightly to its final steady state of 0.80.
The rate of emptying (shown in Figure 5) during the initial two minutes of the model run was almost identical across all 4 zones. Following this there was a deviation between zone one and zone four, with the latter reducing to 0.33 at five minutes and the former at 0.43. In spite of the faster rate of decrease seen initially in zone four, by the eleventh minute the stratification level was equal to that of zone one at 0.23. Zone one stratification layer decreases in a remarkably uniform manner, reducing at a rate of 0.5cm per minute until the saline solution occupies only 0.20 of the model height after twelve minutes.
Looking at the graph it is possible to see that zone two and zone three have a very similar emptying rate. A steady stratification depth of 0.20 is observed in zone four until the twenty-first minute, after which it becomes slightly thinner still and reaches a completely steady state of 0.17 throughout the simulation. Zone one does not maintain such a steady level, with small decreases in saline solution depths as the test approaches the twenty minute stage. At twenty-one minutes the stratification level in zone one settles at its final depth of 0.13. While the simulation approaches the twenty-five minute mark, zone two settles at the same level as zone one.
Run 5 – Single heat source
It was not initially planned to run a simulation using a single heat source, as the model covers three rooms each with their own heat sources. However it was thought that carrying out a simulation using a single heat source may provide a clearer visual representation of how fluid moves through the resistance pathways. In the SSEES building, the level of occupancy can vary from day to day. At most times it is likely that the academic offices and the research room are both occupied, but this will not always be the case. This model run shows what happens to the stratification levels when only one of the rooms is occupied.
Logically, with the heat source located in zone one it is this zone that fills the fastest. The saline solution spreads across the ceiling of the zone one, flows through the opening at the top of the first resistance pathway and begins to stratify in zone two. Once the layer of saline solution in this zone has reached the level at which the inlet openings of RP2 are situated, the fluid starts to build up inside zone three. In turn the solution makes its way through the acoustic corridor into zone four and starts to stratify.
There were two main points of interest about the stratification levels once the simulation had reached a steady state; the first of which is how remarkably flat the stratification layer becomes in zone four. During the previous simulations with a resistance pathway inserted, the stratification level was continually more distinct in the academic offices than in the research room; but none more so than in this simulation.
The other interesting observation found during this simulation is the heights at which the stratification levels settled. In each of the previous model runs, the steady state maintained the stratification layer depths at 0.67 in zone one and the academic offices, regardless of the model configuration. With only one heat source present, the steady state created a stratification layer as deep as 0.73 of the model height in zone one, and closer to 0.33 in the academic offices.
The effect of resistance pathways
During the filling phase of the salt-bath modelling, steady state stratification levels in the research room and the academic offices remained constant across all model runs. However, the stratification levels from run three show an increase of 15% in zone 2, which is the space inside resistance pathway 2. This would indicate that heat is contained within the acoustic corridor, somehow prevented from flowing into the academic offices.
Stratification levels found during run four highlight even further that partitioning naturally ventilated building affects air flow. Table 2 shows a 30% increase in stratification layer depth is found in zone 2, which in this run is the corridor between the research room and academic offices. The fifth run shows a change from the base case on the stratification levels of all four zones, with the zone one increasing by 9% and zone four reducing by 45%.
Table 2- The table shows the steady state stratification levels from the filling stage of the salt-bath modelling experiments. The table diplays the base case stratification level from run one, levels of each zone from subsequent runs and the percentage of change from the base case.
The results from the emptying phase (shown in Table 3) of the model runs confirm that resistance pathways have an effect on stratification layer depth in naturally ventilated buildings. Corresponding with the pattern seen during the filling stage, the stratification levels in run two are identical to the base case, but changes are found in runs three and four.
Table 3 – The table shows the steady state stratification levels from the emptying stage of the salt-bath modelling experiments. The table diplays the base case stratification level from run one, levels of each zone from subsequent runs and the percentage of change from the base case.
Run three shows that after reaching a steady state the stratification levels in the research room and academic offices are equal to those found in the base case. However the depth of saline solution layer inside the acoustic corridor shows a 54% increase from the depth of the stratification level; again this shows that the acoustic corridor is retaining heat.
Interestingly, the emptying phase in run four provides even stronger indications that resistance pathways affect the performance of naturally ventilated buildings. As opposed to the results shown previously, the stratification level in the research room and academic offices are not equal. The results show that for the first time the corridor between the partitions settled at the same height as the base case, as did the research room. Both the acoustic corridor and the academic offices record a 31% increase in the volume of saline solution retained during the emptying stage.
The salt-bath experiments were used to provide data on how stratification levels can change in a building with the introduction of partitioning. Prior to starting this research project it was expected that increasing the number of resistance pathways in a naturally ventilated building would have a greater impact on the buildings air flow. Through carrying out salt-bath modelling tests on the scale model with both of the resistance pathways inserted this was shown to be true. Partitioning can lead to increases in the depth of stratification layers by 20-30% in unoccupied areas during the day; further leading to a 31% increase in the depth of stratification layers, in occupied areas, during cooling.
It has been shown that partitions with openings situated along the length of the top and bottom have less of an effect on stratification levels in naturally ventilated buildings when compared to acoustic corridors. Acoustic corridors retain heat inside their internal void during night-cooling; although this does not affect the stratification levels during occupied hours, it does have an effect on how the building purges. The greatest effect that resistance pathways have on air flow is found when they are situated close to one another. Additional work could be carried out looking at how the distance between two partitions affects the air flow through, and stratification levels within naturally ventilated buildings.
This research would be complimented by investigating how changes in the number of heat sources present affects the air flow and stratification levels of a naturally ventilated sub-zoned building. Advantages of this study would provide greater insight as to how best design a naturally ventilated building that has to accommodate fluctuating heat sources. Further studies could incorporate running simulations consecutively on the same configuration, to investigate if stratification levels are affected by the presence of warm air retained in the building from the previous day.
This research was supported with funding from the Engineering and Physical Sciences Research Council (EPSRC) and was undertaken within the London-Loughborough Centre for Doctoral Training (Lo-Lo CDT) in Energy Demand Reduction.
Baines, W.D. & Turner, J.S., 1969. Turbulent buoyant convection from a source in a confined region. Journal of Fluid Mechanics, 37(01), p.51. Available at: http://www.journals.cambridge.org/abstract_S0022112069000413.
Coley, D., 2002. Sensitivities in the relationship between ventilation rate, vent separation and temperature. Energy and Buildings, 34(5), pp.467-468. Available at: http://linkinghub.elsevier.com/retrieve/pii/S037877880100127X.
Emmerich, S.J., Polidoro, B. & Axley, J.W., 2011. Impact of adaptive thermal comfort on climatic suitability of natural ventilation in office buildings. Energy and Buildings. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0378778811001691 [Accessed July 22, 2011].
Fitzgerald, S & Woods, a, 2008. The influence of stacks on flow patterns and stratification associated with natural ventilation. Building and Environment, 43(10), pp.1719-1733. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0360132307002041 [Accessed August 18, 2011].
Germeles, a E., 1975. Forced plumes and mixing of liquids in tanks. Journal of Fluid Mechanics, 71(03), p.601. Available at: http://www.journals.cambridge.org/abstract_S0022112075002765.
Gratia, E. & Deherde, a, 2007. Guidelines for improving natural daytime ventilation in an office building with a double-skin facade. Solar Energy, 81(4), pp.435-448. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0038092X06002015 [Accessed August 18, 2011].
Hunt, G.R. & Linden, P.F., 2001. Steady-state flows in an enclosure ventilated by buoyancy forces assisted by wind. Journal of Fluid Mechanics, 426(October 1991), pp.355-386. Available at: http://www.journals.cambridge.org/abstract_S0022112000002470.
Ji, Y. & Cook, M.J., 2007. Numerical studies of displacement natural ventilation in multi-storey buildings connected to an atrium. Building Service Engineering Research and Technology, 28(3), pp.207-222. Available at: http://bse.sagepub.com/cgi/doi/10.1177/0143624407077190 [Accessed July 27, 2011].
Li, Y., 2000. Buoyancy-driven natural ventilation in a thermally stratified one-zone building. Building and Environment, 35(3), pp.207-214. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0360132399000128.
Li, Y., Sandberg, M. & Fuchs, L., 1993. Effects of thermal radiation on airflow with displacement ventilation: an experimental investigation. Energy and Buildings, 19(4), pp.263-274. Available at: http://linkinghub.elsevier.com/retrieve/pii/037877889390011I.
Linden, P.F., 1999. the Fluid Mechanics of Natural Ventilation. Annual Review of Fluid Mechanics, 31(1), pp.201-238. Available at: http://www.annualreviews.org/doi/abs/10.1146/annurev.fluid.31.1.201.
Linden, P.F., Lane-Serff, G. F. & Smeed, D. a, 1990. Emptying filling boxes: the fluid mechanics of natural ventilation. Journal of Fluid Mechanics, 212(-1), p.309. Available at: http://www.journals.cambridge.org/abstract_S0022112090001987.
Lomas, K.J. & Ji, Yingchun, 2009. Resilience of naturally ventilated buildings to climate change: Advanced natural ventilation and hospital wards. Energy and Buildings, 41(6), pp.629-653. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0378778809000036 [Accessed July 27, 2011].
Mundt, E., 1995. Displacement ventilation systems—Convection flows and temperature gradients. Building and Environment, 30(1), pp.129-133. Available at: http://linkinghub.elsevier.com/retrieve/pii/0360132394E00029.
Pasquay, T., 2004. Natural ventilation in high-rise buildings with double facades, saving or waste of energy. Energy and Buildings, 36(4), pp.381-389. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0378778804000465 [Accessed July 27, 2011].
Priyadarsini, R., 2004. Enhancement of natural ventilation in high-rise residential buildings using stack system. Energy and Buildings, 36(1), pp.61-71. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0378778803000768 [Accessed August 18, 2011].
Sandbach, S.D. & Lane-Serff, Gregory F., 2011. Transient buoyancy-driven ventilation: Part 1. Modelling advection. Building and Environment, 46(8), pp.1578-1588. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0360132311000308 [Accessed July 27, 2011].
Stavrakakis, G. et al., 2008. Natural cross-ventilation in buildings: Building-scale experiments, numerical simulation and thermal comfort evaluation. Energy and Buildings, 40(9), pp.1666-1681. Available at: http://linkinghub.elsevier.com/retrieve/pii/S037877880800056X [Accessed July 27, 2011].
Tovar, R., Linden, P. & Thomas, L., 2007. Hybrid ventilation in two interconnected rooms with a buoyancy source. Solar Energy, 81(5), pp.683-691. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0038092X0600226X [Accessed July 27, 2011].
Woods, A.W., Fitzgerald, Shaun & Livermore, S., 2009. A comparison of winter pre-heating requirements for natural displacement and natural mixing ventilation. Energy and Buildings, 41(12), pp.1306-1312. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0378778809001613 [Accessed July 27, 2011].
Worster, M.G. & Huppert, H.E., 2006. Time-dependent density profiles in a filling box. Journal of Fluid Mechanics, 132(-1), p.457. Available at: http://www.journals.cambridge.org/abstract_S002211208300172X.
INVESTIGATING THE EFFECTS OF RESISTANCE PATHWAYS ON AIR FLOW THROUGH NATURALLY VENTILATED BUILDINGS
Conference paper and presentation in the Building Simulation Optimisation (BSO) 2012 conference.