As buildings become taller, seemingly straightforward physical phenomena can develop into significant engineering challenges. One of these is stack effect — the movement of air through a building driven by differences in temperature, and therefore air density, between the internal and external environments.
The underlying physics is simple: warm air rises and cold air falls. In a super-tall building, however, the resulting pressure differences can act over several hundred metres of height and interact with façades, entrance doors, lift shafts, stairwells and mechanical ventilation systems.
The result can be much more than an uncomfortable draught. Severe stack effect can interfere with the operation of doors and elevators, increase infiltration through the façade, affect heating performance and provide pathways for the movement of smoke and odours through a building.
The problem becomes particularly important for very tall buildings located in climates with cold winters — a combination increasingly encountered in cities across northern and inland China.
What causes stack effect?
Stack effect occurs because air density varies with temperature.
During cold weather, the conditioned air inside a building is generally warmer and less dense than the external air. The warmer internal air tends to rise through the building, particularly through relatively unrestricted vertical paths such as lift shafts and stairwells.
This creates a characteristic pressure distribution over the height of the building. Air tends to be drawn into the lower levels while internal air is pushed outward at higher levels.
Somewhere between these regions is a neutral pressure level, where the internal and external pressures are approximately equal.
The pressure difference produced by stack effect can be represented approximately by:
ΔP = ρᵢ g (h − hₙₚₗ)(Tᵢ − Tₒ) / Tₒ
where the pressure difference depends principally on the height relative to the neutral pressure level and the difference between internal and external temperatures.
This relationship immediately illustrates why stack effect has become particularly relevant to modern super-tall buildings: the greater the building height and temperature difference, the greater the potential pressure difference.
During hot weather the process can reverse, producing an inverted stack effect, although in many climates the severe winter condition represents the more significant design case.
Why does stack effect matter?
The most visible problems often occur at doors and elevators.
Large pressure differences across elevator doors can interfere with their operation, while manually operated doors may become difficult to open or close. The original study considered pressure differences of approximately 50–65 Pa across elevator doors and approximately 75–100 Pa across internal doors as indicative operating criteria, although appropriate criteria should ultimately be established for the particular door and elevator systems being designed.
There are also less obvious consequences.
Large volumes of air can move through entrance lobbies and other occupied spaces, creating uncomfortable conditions. Cold external air entering the lower levels can increase heating demand, while uncontrolled internal airflow can transport smoke and odours between different parts of the building.
The challenge is therefore not simply controlling pressure. It is understanding the network of airflow paths that allows the pressure difference to drive air through the building.
A simple phenomenon within a complicated building
This distinction is important. The basic physics of stack effect is straightforward.
A modern super-tall building, however, is not a sealed tube.
It contains hundreds or thousands of potential airflow paths: façade leakage, revolving and swing doors, lift doors, lift shafts, stairwells, service penetrations, internal doors, mechanical ventilation systems and connections between different zones of the building. Some of these paths offer relatively high resistance to airflow. Others — particularly continuous vertical shafts — can provide highly effective routes through the building.
Consequently, two buildings of similar height in the same climate can experience substantially different stack-effect behaviour depending upon their internal configuration and the airtightness of their envelopes.
Prevention is considerably easier than mitigation
Stack effect should ideally be considered during design rather than after construction.
Once a building has been completed, modifications can become architecturally difficult, operationally disruptive and expensive. Measures that could have been incorporated relatively easily into the original design may require significant intervention when retrofitted.
Three principles are particularly important:
Façade airtightness. Leakage through the building envelope provides one of the principal paths through which stack-effect pressures drive airflow. Curtain-wall façades are never perfectly airtight, but controlling their leakage can substantially influence building performance.
Vertical separation. Long, continuous vertical airflow paths should be avoided where practical. Dividing elevator systems into separate vertical zones, for example, can interrupt the airflow path that would otherwise extend over much of the height of the tower.
Internal airflow resistance. Doors and vestibules between vertical shafts and occupied areas provide additional resistance to airflow and can significantly reduce pressure communication throughout the building.
Entrance design is also important. Revolving doors are particularly useful because they allow people to enter and leave while maintaining separation between the internal and external environments.
The importance of façade airtightness
Façade performance deserves particular attention.
Even relatively small leakage rates become important when subjected to the pressure differences generated over the height of a tall building. The original study compared Chinese façade leakage classifications with ASTM/AAMA criteria and highlighted substantial differences in allowable leakage rates.
For a building of perhaps 200–300 metres or more, the stack-effect pressures can be substantially greater than the relatively low pressure at which façade leakage specifications may be quoted.
A façade specification should therefore not simply exist on paper. Appropriate testing and construction quality control are important if the intended airtightness is to be achieved in the completed building.
For stack-effect control, airtightness is one of the most powerful tools available to the designer.
Modelling the building as an airflow network
Because of the large number of interacting airflow paths, stack-effect behaviour can be investigated using numerical network airflow modelling.
The building is represented as a series of interconnected pressure zones. Air movement between adjacent zones is calculated according to pressure differences, temperatures and the flow resistance of the connecting elements.
The governing principles are relatively simple and are fundamentally based upon conservation of mass. The complexity comes from the building itself: a large tower can contain thousands of individual airflow paths.
A model may include:
internal temperatures under winter and summer conditions;
façade leakage characteristics;
lift shafts and stairwells;
external, elevator and internal doors;
mechanical ventilation openings; and
HVAC operation and internal pressurisation.
From this, the analysis can determine the location of the neutral pressure level, pressure differences across critical doors and elevators, airflow rates, façade infiltration and exfiltration, and internal pressures relevant to mechanical design.
Using modelling to improve the design
The greatest value of numerical modelling is not simply predicting whether a stack-effect problem exists.
It allows different solutions to be investigated before they are constructed.
If excessive pressure is predicted across a particular elevator bank, for example, the model can be modified to introduce additional separation, alter door arrangements or investigate changes to façade leakage. The effectiveness of each measure can then be assessed numerically.
This allows mitigation measures to be targeted where they provide the greatest benefit rather than making extensive changes throughout the building.
The same modelling can also be used to help design a full-scale measurement programme once the building is complete.
Measuring stack effect in completed buildings
Full-scale measurements provide another important tool, particularly when investigating an existing building.
Pressure differences can be measured across elevator doors, entrance doors and other critical partitions. These measurements can be combined with observations of airflow and building operation to identify the dominant flow paths.
Comparison between measured and predicted behaviour can also help refine numerical models and evaluate proposed remedial measures before physical modifications are made.
This is particularly valuable when addressing an existing building, where changes to entrances, lift lobbies or internal partitions may be costly.
Designing stack effect out of super-tall buildings
The increasing height of buildings has made stack effect an increasingly important aspect of environmental and building-performance design.
The fundamental phenomenon is simple, but its interaction with the complex airflow network within a super-tall building is not.
The most effective approach is therefore preventative.
A tight building envelope, appropriate vertical separation of elevator shafts and stairwells, well-designed vestibules and careful control of internal airflow paths can substantially reduce the magnitude of the problem.
Where problems already exist, mitigation remains possible through additional doors and vestibules, improved sealing, changes to entrance arrangements, specialised elevator-door systems and appropriate management of building services. However, these measures are generally more difficult and expensive once construction has been completed.
For mega- and super-tall buildings, stack effect is best treated as a design issue rather than an operational problem to be solved later.
