Energy and Buildings 39 (2007) 166181

www.elsevier.com/locate/enbuild

Architectural design of an advanced naturally

ventilated building form
Kevin J. Lomas *
Institute of Energy and Sustainable Development, De Montfort University, Leicester LE1 9BH, UK
Received 3 April 2006; received in revised form 4 May 2006; accepted 24 May 2006

Abstract
Advanced stack-ventilated buildings have the potential to consume much less energy for space conditioning than typical mechanically
ventilated or air-conditioned buildings. This paper describes how environmental design considerations in general, and ventilation considerations in
particular, shape the architecture of advanced naturally ventilated (ANV) buildings. The attributes of simple and advanced naturally ventilated
buildings are described and a taxonomy of ANV buildings presented. Simple equations for use at the preliminary design stage are presented. These
produce target structural cross section areas for the key components of ANV systems. The equations have been developed through practice-based
research to design three large educational buildings: the Frederick Lanchester Library, Coventry, UK; the School of Slavonic and East European
Studies, London, UK; the Harm A. Weber Library, Elgin, near Chicago, USA. These buildings are briefly described and the sizes of the as-built
ANV features compared with the target values for use in preliminary design. The three buildings represent successive evolutionary stages: from
advanced natural ventilation, to ANV with passive downdraught cooling, and finally ANV with HVAC support. Hopefully the guidance, simple
calculation tools and case study examples will give architects and environmental design consultants confidence to embark on the design of ANV
buildings.
# 2006 Elsevier B.V. All rights reserved.

Keywords: Low energy buildings; Advanced natural ventilation; Ventilation areas; Case studies; Downdraught cooling

1. Background water and refrigerant) accounted for up to 50% of the emissions

The imperative of reducing the emission of greenhouse gases,
and in particular CO2, caused by the burning of fossil fuels has
stimulated interest in the design of low energy buildings. In the 20
buildings monitored by Bordass et al., in the well known UK
PROBE Studies [1] there was a factor of 6 difference in the CO2
emissions produced for space conditioning and lighting a given
floor area (Fig. 1). Nine of the 10 highest CO2 emitters were air-
conditioned (AC) or mixed mode (MM) (these used chilled
beams, with displacement ventilation, etc. rather than full AC),
and 9 of the 10 lowest emitters were naturally ventilated (NV) or
advanced naturally ventilated (ANV). The term advanced
natural ventilation was coined to encompass buildings which
utilised the stack effect to drive an air flow and so has been
adopted for the buildings which are the subject of this paper. In
the AC and mechanically ventilated buildings, the CO2 emissions
resulting from the fans and pumps required to move air (and
* Tel.: +44 116 257 7961; fax: +44 116 257 7977.
E-mail address: klomas@dmu.ac.uk.
associated with space heating and cooling. Because AC buildings
tend to be deep-plan, the CO2 emissions for artificial lights were
also substantial. Buildings which are particularly densely
occupied, with long periods of usage and with high internal
heat gains (e.g. from computers and other equipment) might
justify the use of AC, but as the PROBE results show, some
relatively lightly used buildings nevertheless had AC.
NV and ANV buildings utilise naturally occurring wind
pressures, and/or the buoyancy force generated by internal heat
sources, to drive an air flow, thereby avoiding the use of fans.
Admitting cool night air into a building, to purge daytime heat
accumulated in exposed thermal mass, can avoid the need for
mechanical cooling entirely or, in warmer locations, reduce
cooling loads, energy use and associated CO2 emissions.
Shallow-plans, which typify simple NV buildings, or the use of
atria and lightwells in deeper-plan buildings, can improve the
use of natural light reducing the CO2 emission associated with
artificial lighting.
Whilst global warming is seen as a treat to NV and ANV
buildings, the overheating risk can be overstated. Current

0378-7788/$ see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.enbuild.2006.05.004
 K.J. Lomas / Energy and Buildings 39 (2007) 166181
Fig. 1. The CO2 emissions from 20 buildings and ECON19 [32] benchmarks.
evidence for the UK, although rather weak, suggests that ANV
can keep buildings comfortable though the next century in all
but the hottest (London) region [2,3].
Conventionally conceived NV buildings are shallow plan
with an extended perimeter, and facade openings which provide
the fresh air inlet and exhaust air outlet (Table 1). These
features can be incompatible with the planning constraints
imposed by tight urban sights and the noise and pollution in city
centres. The use of manually operated windows can
compromise security, increasing concerns about theft by
building occupants (a particularly important consideration
for library buildings of the type described in this paper).
Mechanically controlled perimeter windows enable night
ventilation but the building may then be vulnerable to break-
in or other malicious acts.
At the design stage an ability to reliably predict the likely
internal conditions in a building, for example by using dynamic
thermal models and computational fluid dynamics programs,
can be reassuring and it is important to have a clear idea of how
the internal conditions in the finished building will be
controlled. Relying, as they do, on variable and ill defined
pressure differences set up across the building by the wind, the
likely performance of simple NV buildings is hard to predict
and control.
ANV buildings that utilise the stack effect, in which air
warmed by internal sources of heat drives the air flow, do not
necessarily rely on wind pressures. If properly designed and
167
controlled, an air flow can be assured at all times when there is
an internal source of warmth, including at night. In fact, in an
unconstrained displacement flow regimen, where heat sources
generate isolated plumes of warm air, the flow rate is directly
proportional to the strength of the source, and the interface
between the cooler air the warm air above remains fixed [4].
With heat sources distributed over a surface, the air flow is also
dependent on the source strength in steady state conditions [5].
This happy coincidence, between heat input and air flow rate,
enables rather simple but robust control of air flow and makes
prediction of performance at the design stage comparatively
reliable. Further, the interface between the cool and warmer air
can be designed to lie above head height.
The benefits of control-ability and predictability, which stack
driven natural displacement ventilation offers, can be lost if wind
pressures begin to dominate the flow. An inability to harness
these pressures is not a disadvantage; after all it is during still
warm summer conditions when it is most difficult to keep ANV
buildings thermally comfortable. Therefore, designing the
buildings to be wind neutral is a useful guiding principle.
In a recent paper [2] a taxonomy was proposed, in which
stack ventilated buildings were divided into four main types
(Fig. 2). The edge-in, centre-out approach (E-C) is exemplified
by the Queens Building at De Montfort University, Leicester,
UK [69] and the edge-in, edge-out strategy (E-E) by the UK
Building Research Establishments (BRE) Energy Efficient
Office of the Future [10].
168 K.J. Lomas / Energy and Buildings 39 (2007) 166181

Table 1
Characteristics of different simple and advanced natural ventilation strategies (after [2])
Simple natural ventilation Advanced natural ventilation (ANV)
Single Cross Edge-in Edge-in Centre-in Centre-in
sided flow edge-out (E-E) centre-out (E-C) edge-out (C-E) centre-out (C-C)
Architectural implications
Air inlet objecta No No No No Yes Yes
Exhaust stacksb No No Yes Yes Yes Yes
Plan depthc 2.5 (5) 5 10d 10d 5
Deep plane No No No Yes Yes Yes
Indoor air quality provided
Occupant inlet control Yes Yes Yes Yes No No
Close control No No Possf Possf Yes Yes
Displacement vent possible No No Yes Yes Yes Yes
Draught control Poor Poor Poor Poor Good Good
Performance predictability Poor Poor Good Good Very good Very good
Protection from local environment
Urban noise attenuation Poor Poor Poor Poor Good Good
Perimeter security Poor Poor Poor Poor Good Good
Robustness to climate change
Night vent cooling Yes f Yesf Yesf Yes f Yes Yes
Possible mech vent assistg No No Yes Yes Yes Yes
Comfort cool Difficult Difficult Difficult Difficult Easy Easy
Heat recovery No No No No Noh Noh
a
Such as plenum and lightwell.
b
Might utilise other feature, such as an exhaust air lightwell.
c
Rules of thumb (e.g. CIBSE, 2001)based on multiples of the floor-to-ceiling height. For single sided vent this is the room depth, but for cross-flow vent it is the
floor plate width perimeter-to-perimeter.
d
With a row of centrally located stacks, exhausting both sides of the building (E-C), or a central air inlet shaft supplying both sides of the building (C-E), the
perimeter-to-perimeter depth may be 10.
e
Exceeding, about, 20 m perimeter-to-perimeter.
f
If mechanically controlled perimeter air inlets are used.
g
E.g. fan in a stack or fan pressurised supply.
h
However, since the air is exhausted through discrete vertical stacks, heat recovery is possible when a mixed mode variant of the building is operated in mechanical
mode (e.g. HAWL).

In both building types, cold winter air can be drawn in over
perimeter heating elements to pre-warm it and in summer
operable windows can be used to enhance airflows, and create air
movement, without disrupting the basic airflow strategy.
Mechanically operated air inlets permit night ventilation (and
in the Queens Buildings lecture theatres, also daytime
Fig. 2. Schematic diagrams of the different forms of stack ventilation.
ventilation). With centrally located stacks (E-C), deep-plan
buildings are possible, as in the Queens Building. Whilst
centrally located atria can, in principle, assist buoyancy driven
flow, stacks require less space, have more reliable ventilation
performance, can have terminations which are less susceptible
to wind effects and can, if necessary, incorporate low-powered
axial fans to encourage airflow under particularly adverse
conditions (as in the BRE office). The disadvantage of the
edge-in strategy is that the perimeter inlets are susceptible to the
noise, pollution and security concerns associated with design on
urban sites (Table 1).
The three case study buildings described in this paper, for
which the author provided strategic design advice and
performance evaluations, on behalf of the client and the
architect, Short and Associates, all utilise a centre-in ANV
strategy: the centre-in, edge-out (C-E) strategy is exemplified
by the School of Slavonic and East European Studies building
(SSEES) building, London, UK [2,1113]; the larger, very
deep-plan, Frederick Lanchester Library (FLL), in Coventry,
UK employs both the C-E and C-C strategy [2,3,11,1319];
whilst the Harm A Weber Library (HAWL), in Elgin, near
Chicago, Illinois, USA [20,21] uses the C-E approach with
localised E-E ventilation of perimeter offices.
 K.J. Lomas / Energy and Buildings 39 (2007) 166181 169

The centre-in strategy has a number of strategic design

advantages: it enables the external facade to be sealed, which
overcomes security, noise and pollution concerns; the air supply
route can become a lightwell if necessary, thereby introducing
daylight into a deep-plan; the air inlet can be protected from
wind effects, giving even greater confidence about the likely
airflows (than in buildings with perimeter openings); and the
fresh air can be pre-heated. Further, by locating the exhaust
stacks at the perimeter of the building, as in the three case study
buildings, the basic floor plate is left clear permitting more
flexible internal space planning.
Most interestingly, the supply air can be comfort cooled or
fully conditioned enabling the same basic, but versatile, plan
form to incorporate either an ANV or a mixed-mode (MM)
approach to ventilation without the overhead of having two
different air distribution systems (one for mechanical mode and
the other for natural mode), Table 1. This offers the prospect of
introducing measures to combat future climate change and for
applying the basic design strategy to a range of climate types
these advantages are illustrated by the case study buildings.

2. Common features of the three case study buildings

There are numerous geometrical differences between the

case study buildings, as dictated by the client, site, budget,
summer cooling strategy, etc., however, design considerations
imposed by the natural ventilation mode of operation have a
major impact on the overall built form and so there are strong
generic similarities between them: it is these on which this
paper concentrates. The geometry of the three buildings and the
intended ventilation strategies, are illustrated in Figs. 38 and
their key features, and dimensions, particularly those related to
the ventilation strategy, are tabulated in Appendix A.
All three case study buildings were for educational
establishments with clients who would own and operate the
buildings and so were concerned about whole of life operating
costs and particularly energy consumption. The buildings
contain cellular offices for staff and teaching spaces and
extensive areas for library books, which, for security reasons,
and because of the noisy sites, required the building facade to be
sealed.
The climate to which the UK buildings were exposed is, of
course, much less severe than that in the Chicago region (see
Appendix A1). For example the UK sites have around 230
cooling degree days (CDD) to base 15.5 8C, compared to 766
for Chicago; the mean daily maximum temperature (MDMa) in
the warmest month (July) is around 20 8C at the UK sites and
28.7 8C in Chicago; and there were under 3% of working hours
when the ambient temperature exceeded 25 8C (WH25) at the
UK sites and over 15% in Chicago. Comparing the two UK
climates it is evident that London, even without considering the
urban heat island influence, is warmer than Manchester (CDD
229 cf. 77; MDMa 22.4 8C cf. 19.4 8C; and WH25 2.9% cf.
Fig. 3. Floor plan of the Frederick Lanchester Library (after [13]).
0.6%). Interestingly, the Chicago climate has a greater mean
diurnal swing in both spring and autumn than the UK climates
(over 9.5 K cf. under 8 K), which suggests that night ventilation
cooling could be a useful energy saving resource. The diurnal
swing for London is, in fact, likely to be less than the climate
file indicates due to the urban heat island effect [13,22].
To contend with these climatic differences, the three case
study buildings illustrate a progressively more complex
environmental control strategy: from pure ANV for the FLL
(which is located in the UK Midlands); through ANV with
comfort cooling using passive downdraught cooling (PDC) in
the SSEES building (because of the reduced summer night
cooling potential caused by the London urban heat island and
because the UK design guidelines relevant at the time [23]
require the use of a near-extreme weather year for the design of
naturally ventilated buildings2); to ANV with full HVAC
support in the HAWL (because of the severe Chicago climate).
As noted above, the summer-time mechanical cooling

1 2
Appendix A presents Manchester data for the FLL as this is the nearest TRY The third hottest year recorded in London (Heathrow) between 1976 and
site to Coventry. 1995: the London Design Summer Year is 1989.
170 K.J. Lomas / Energy and Buildings 39 (2007) 166181

Fig. 4. Frederick Lanchester Library showing air supply strategy (left) and air exhaust strategy (right) (after [13]).

equipment could be introduced into the SSEES and HAWL

Fig. 5. Floor plan of the School of Slavonic and East European Studies Building
(after [13]).
without compromising the basic centre-in ANV strategy.
The buildings have a square (or in the case of the SSEES,
approximately square) footprint, which yields a low surface
area to volume ratio. This, together with the high insulation
standards used in the roofs and walls (Appendix A), produces
low specific fabric heat gains and losses. The windows are of
clear low-emissivity double glazing to admit natural light to
perimeter offices and work spaces, and the SSEES and FLL
have artificial lighting which responds to daylight levels. The
windows are well shaded to reduce perimeter heat gains: either
by deep window reveals (HAWL); by adjacent buildings
(SSEES); or by the stacks, stair towers and metal shading fins
(FLL). Concrete (SSEES) or steel (FLL and HAWL) columns
and beams support the exposed flat concrete ceilings, which are
essential for effective night ventilation cooling. Castellated
beams (FLL) or open trusses (HAWL) enable stratified warm
air to move across the ceiling soffit. The plan forms, insulation
standards and window designs represent good, energy efficient
practice, irrespective of how buildings are conditionedbut the
deep-plans are unusual for NV buildings.
The lightwells are, of course, a critical and distinctive feature
of the three buildings. These supply fresh air to each above-
ground floor via low level openings to encourage a displacement
ventilation flow. Higher floor-to-ceiling dimensions are advanta-
geous with such a flow regimen. The flow of air from the lightwell
to occupied spaces is control by either dampers (FLL) or
windows (SSEES, HAWL) set below desktop level. Secondary
 K.J. Lomas / Energy and Buildings 39 (2007) 166181
Fig. 6. School of Slavonic and East European studies building showing the
natural ventilation cooling strategy (after [13]).
Fig. 7. Floor plan of the Harm A Webber Library (after [20]).
171
Fig. 8. Harm A Webber Library showing the natural ventilation strategy (after
[20]).
heating is provided by column radiators (SSEES, HAWL) or
trench heaters (FLL) at the point where the air leaves the
lightwell and enters the occupied spaces. Perimeter heating is
used in all buildings to offset any fabric heat loss.
The lightwells in the SSEES and HAWL are centrally located
to so that the flow path from the inlets to the perimeter air outlets
is approximately the same in all directions. In the larger
(50 m  50 m) FLL, four lightwells are used, one in the centre of
each quadrant, and a central lightwell acts like a large, glazed air
exhaust shaft. The use of a triangular lightwell in the SSEES
building was primarily an architectural choice, precipitated by
the shape of the building and constructional considerations,
rather than ventilation or environmental control considerations.
The lightwells have clear glazing in the walls and at the top
to admit natural light to the centre of the buildings and to
provide visual connectivity between the interior and the
outside. It is critical that the lightwell tops are sealed shut in
winter to prevent warmed buoyant air within them leaking
away. They must also incorporate summertime solar radiation
172 K.J. Lomas / Energy and Buildings 39 (2007) 166181
control to stop the ambient air, and in the case of the SSEES and
HAWL buildings the cooled air, being heated. The FLL and the
HAWL have an enclosed greenhouse formed by a horizontal
glazed screen just above the air inlet to the top-most floor. The
greenhouses have moving blinds and can be copiously
ventilated to remove solar-originated heat in summer. The
SSEES building uses the same principle, but the greenhouse
takes the form of an upper and lower ETFE cushion. There is no
blind system, but the space between the cushions can be
ventilated and the lightwell top is tilted towards the north. The
lower ETFE layer has dampers around its perimeter with
cooling batteries below. In spring and summer the dampers
admit ambient air for ventilation cooling and if necessary the air
can be chilledpassive downdraught cooling.
In all three buildings air is supplied to the lightwell(s) via a
horizontal plenum located between the ground floor and the
basement. The plenum feeds all sides of the lightwell, in the
SSEES and HAWL, but just two sides of each lightwell in the
FLL. Vertical drops from the plenum supply fresh air to
basement areas in the SSEES and HAWL, which are exhausted
by stacks. Because the basements are partially earth-bound and
are poorly day lit, they tended to house unique (support) spaces
(e.g. book archives and computer rooms), some of which may
require air conditioning. (For example, the whole of the FLL
basement is a 24 h access computing suite.)
The plena are supplied with ambient air via dampered slots
at the buildings perimeter (FLL and HAWL) or by air supply
corridors and discreet inlets (SSEES). Each plenum has inlets
located on more than one side of the building so that the
dampers at one or more inlets can be closed in adverse wind
conditions whilst retaining an open air inlet elsewhere. The
inlets are heavily louvered and incorporate either bird and
rodent mesh (SSEES, HAWL) or insect mesh (HAWL). Heater
batteries pre-heat the air and are located either across the base
of the lightwell (FLL) or behind the air inlets (SSEES, HAWL).
The latter strategy avoids insulating the plenum and enables the
bottom of the lightwell to be clear glazing, thereby providing
natural light to the basement.
The perimeter stacks are an architecturally striking feature of
these ANV buildings and are crucial the ventilation strategy. In
all three buildings they are reasonably uniformly distributed
around the perimeter, which: assists with aesthetics; enables the
stacks to contribute to solar shading of windows; creates thermal
buffers between the inside and outside; helps to ensure zones of
warm stale air do develop in the building; and offers planning
flexibility by enabling perimeter cellular spaces to be easily
locked into an exhaust stack. This latter benefit is fully
exploited in the SSEES building, which has many offices: the rear
stacks, of triangular shape, cover the entire back wall and a
double facade runs right across the front face.3 There are draught
lobbies to entrance doors which prevent the stacks drawing in
ambient air, which is particularly important in winter when the
stack forces are greatest and the draught risk higher.
3
The outer front facade was designed to harmonise with the Georgian
architecture of the surrounding area.
Exhaust air enters the stacks through dampered openings set
below the ceiling soffit. The stacks are vertical and well
insulated to keep the air in them warm and buoyant. They
discharge above roof level to provide the necessary stack height
and to position the terminations (which are to be neutral to wind
effects) out of the turbulent airflow zone at roof level. The
terminations are louvered to prevent the ingress of precipitation
and they contain bird or insect mesh. Above the roofline the
stacks have a rectangular cross section, which simplifies the
design of the dampers which seal each stack at roof level.
In the HAWL the stacks discharge into a sloped roof plenum
which exhausts via five ridge-mounted terminations: a position
which the client preferred to the more dramatic perimeter
location for stacks.
Experience from CFD analyses has indicated that, unless
perimeter stacks extend above the level of the top floor inlet a
long way, which can be costly and impractical, stale air from
lower floors can re-enter the top most floor [15]. Thus dedicated
top floor exhaust paths have become a feature of these ANV
buildings: separate short stacks in the HAWL; short stacks plus
partitioned perimeter stacks in the FFL; and partitioned stacks
at the rear, but dedicated partitioned chimneys at the front, in
the SSEES.
3. Preliminary sizing of advanced natural ventilation
system components
3.1. Preliminary design
In the preliminary design phase the geometry of a building
can be extremely fluid as the architect grapples with a multitude
of design constraints and design driversstructure, space
planning, fire, safety, cost, etc. This might include considering
the number of storeys, the disposition of open plan and cellular
spaces and the external visual appearancein particular the
position, size and number of stacks and the style of the roof top
exhaust air terminations.
Under such circumstances it is helpful if the design team can
work with simple equations and guidelines. In the context of
centre-in ANV buildings, these need to assist with sizing and
locating the main components: the plena; the lightwell(s); the
stacks; and the air inlets and outlets to and from these. The
critical measure is the free area available for air flow and so the
equations that follow generate target structural areas for
preliminary design, i.e. the size of the opening to be created by
the architect (and into which any air flow control, heating and
other equipment will be inserted). The target areas are
expressed as a percentage of the floor area to be ventilated,
which enables them to be used with buildings of arbitrary size
and shape.
The air flow rates required to maintain thermal comfort
during warm, still summer conditions invariably dictate the
maximum free area of opening required (some spaces have
higher ventilation needs, see e.g. [24], but in these specialist
buildings, natural ventilation is probably inappropriate).
Initially, the free areas are calculated on the basis of the
overall expected internal heat gain in the particular building
 K.J. Lomas / Energy and Buildings 39 (2007) 166181
Table 2
Estimated structural inlet and outlet areas for use at the preliminary design stage of advanced natural ventilation systems
Total heat gain Airflow rates
(W/m2) 1 a 1 b
(ls ) (ach )
20 2.4 2.5
30 3.6 3.7
40 4.8 4.8
50 6.0 6.0
60 7.1 7.4
Values in bold are target area for preliminary design purposes.
a
Air flow rate required for ventilation cooling per m2 of floor area.
b
Assumes 3.5 m floor to ceiling height.
c
Eq. (8) assumes gross structural area is twice the free area and the air speed is limited to 0.3 m/s.
d
Eqs. (3), (4), (9) and (10) presume no obstruction by grills, dampers, meshes, louvers, etc. and an air speed of 0.5 m/s.
e
Eq. (7) assumes gross structural area is twice the free area and the air speed is limited to 0.5 m/s.
type, tabulated values for which can be found in design guides
(e.g. [24]).
Assumptions about the likely internal temperature differ-
ential and air speeds have to be made and, for preliminary
design, values are chosen such that the calculated ventilation
opening areas are conservative, i.e. over, rather than under,
sized. Experience indicates that areas set aside for ventilation
(stacks, lightwells, etc.) at preliminary design, can be readily
surrendered for other purposes as the design evolves, but trying
to reclaim space, to compensate for under sizing early in the
design process, can be very difficult.
The target areas calculated are, of course, merely the starting
point. As the design evolves, the free areas can be refined, by re-
using the equations but with the improved knowledge about the
use which will be made of spaces and thus the likely heat gains.
Detailed design can involve more accurate manual calcula-
tions, such as the application of stack effect equations (see, for
example [24]), and, later in the design process, the use of
sophisticated computer-based methods such as thermal
simulation and computational fluid dynamics analysis (see
Appendix A for the analyses used in designing the case study
buildings). Indeed, it is experience gained through the use of
these methods that has informed the development of the simple
equations and guidelines presented here.
3.2. Displacement and stack ventilation
The sizing equations are based on considerations of a simple
stack-driven displacement ventilation regimen; that is, a low
level inlet supplying the space to be cooled and a high level
outlet into a stack. The volume flow of air, m (m3/s), required to
provide ventilation cooling for different internal heat gains is
given by:
QA
m m3 =s
C c DT
where Q is the heat gain (W/m2), A (m2) the floor area, DT (K)
the allowable temperature rise, and Cc is the volumetric heat
capacity of air (1200 J/m3 K).
Typically, the supply air temperature would be 23 K below
the target temperature for the occupied zone. The temperature
173
Target structural areas as percentage of area of floors served (%)
Lightwell Lightwell, plenum Plenum
outletsc outlet and stacks d inlete
1.6 0.5 1.0
2.4 0.7 1.4
3.2 1.0 2.0
4.0 1.2 2.4
4.8 1.4 2.8
close to the ceiling could be about 3 K, or in the case of higher
ceilings, 4 K above the mid-height temperature [24,25]. Thus
an assumption that the overall temperature difference, DT, is
7 K is reasonable. Given this, the volume flows necessary for
different heat gains can be found (Table 2, cols 2 and 3).
The allowable temperatures rise, DT, could be varied from
the value used here, for example 5 K might be more appropriate
in buildings with a lower ceiling height (and vice-versa). The
effect would be to proportionally increase (or decrease) the
target volume flows of air, and hence the target opening areas,
required.
The mass flow rate can be converted into a free area of
ventilation opening, A, via:
m
A m2 (2)
v
where v (m/s) is the assumed air speed.
The achievable air speed will decrease as the stack height
decreases, all other factors being equal. Assuming that
dedicated ventilation provision is made for the top floor of
the building (as in the case studies) then the top-but-one floor
will have the shortest stack height, say 6 mthe height of the
floor above plus the height from the roof level to the stack
termination. Using this value, and a DT of 7 K, it can be shown
(e.g. equations in [24], pp. 411) that a value for v of 0.5 m/s is
reasonable. Experience from CFD analysis corroborates this
rough assumption (e.g. [21]).
3.3. Location and size of lightwell
It is generally most appropriate to position the lightwell in
the middle of the floor plates which are to be ventilated:
although circumstances can arise which dictate otherwise, for
example when a building abuts its neighbours so that exhaust
(1) stacks cannot be located on all sides.
The volume flow of air required up the lightwell, ml, can be
calculated on the basis of total area of the building to be
ventilated from the lightwell, Ab, and the expected daily
average heat gain density in these areas, Qb. A building-average
heat gain is appropriate for lightwell sizing even if peak gains in
individual spaces are known, because, whilst some zones might
174 K.J. Lomas / Energy and Buildings 39 (2007) 166181
be at full occupancy, it is unlikely that all spaces will be so
simultaneouslypeople move around redistributing the heat
sources (and the stack system will automatically draw more
air to the more densely occupied, and thus warmer, zones).
Periods of particularly dense occupancy also tend to be short
lived (especially at the whole building level) and a thermally
massive building, in which occupants can radiate heat to a
night-cooled ceiling slab, can simply ride out periods of dense
occupation (the time history of thermally massive buildings can
be in the order of several days).
The lightwell cross sectional area, Al , can be expressed as a
percentage of the whole building floor area by combining
Eqs. (1) and (2):
Al Qb
 100 % (3)
Ab vC c DT
Using the values of 0.5 m/s, 1200 J/m3 K, and 7 K for v, Cc
and DT, respectively, yields the ratio of the lightwell area to that
of the total floor area ventilated (Table 2), e.g. 0.7% and 1.2%,
for heat densities of 30 and 50 W/m2, respectively.
For a lightwell fed by a plenum (as in the FLL and HAWL),
the cross sectional area calculated is, in fact, for the bottom of
the lightwell. From an air supply standpoint, a bottom-fed
lightwell could taper because the air volume to be carried
diminishes floor-by-floor. However, from an interior day-
lighting standpoint, it is better to have a lightwell with a large
aperture at the top (and a larger aperture is almost certainly
needed if the lightwell is (also) to be fed by from above in
ventilation cooling mode, e.g. the SSEES building). Also, as
will be seen later, it tends to be the available area of lightwell
perimeter, rather than the cross sectional area of the lightwell,
which begins to dictate its size. Therefore, in practice, air
supply lightwells (even those fed with air from the bottom only)
tend to have vertical sides: this can also be less costly than
sloping sides.
3.4. Sizing the air inlet plenum
If the lightwell is fed only from the bottom, the plenum
needs to be able to deliver all the air the building needs. Thus,
Apo Qb
 100 % (4)
Ab vC c DTbpo
where bpo is the proportion of the plenum outlet that is blocked
(0, fully blocked; 1, no blockage), Apo the free cross-sectional
area of the plenum outlet into the lightwell and v and DT are
again 0.5 m/s and 7 K, respectively. If there is no obstruction at
the interface between the lightwell and plenum and, in fact, no
airflow control device is needed at this point, the free cross-
sectional area Apo is equal to the gross structural area, giving:
Apo Al (5)
Because the building perimeter, which is the location for the
plenum inlet, is longer than the lightwell perimeter, it tends to
be the outlet from the plenum into the lightwell which
determines the plenum depth, Dp:
Apo
Dp m (6)
Pl
where P1 (m) is the length of the lightwells perimeter. Because
a shallow plenum is desirable, as this reduces the overall
building height (and increases the head height in a basement
below), it is advantageous to make maximum use of the
perimeter available by connecting the plenum to all sides of
the lightwell (as in the SSEES and the HAWL).
The plenum can be supplied with ambient air either by a slot
around the buildings perimeter (FLL, HAWL) or by air
corridors (e.g. SSEES). The structural opening to these should
be sufficiently large that the necessary obstructions, dampers,
heater batteries and bird or insect meshes, do not inadvertently
reduce the free area. Therefore, the structural area of inlet to the
plenum is given by:
Api Qb
 100 % (7)
Ab vC c DTbpi
where bpi is the proportion of the plenum inlet that is
obstructed.
For preliminary design purposes, it is reasonable to assume
that bpi is 0.5. Methods of reducing blockage at the air inlet
include: raking the heater batteries (SSEES and HAWL) or
enlarging the mouth of the plenum (HAWL). Whilst the plenum
can carry services these should not unduly restrict the free area
or hinder maintenance.
If the air entering the plenum serves only the lightwell (and
not, for example, the basement below) then Qb and Ab are the
same as in Eqs. (3) and (4). However, if some of the air entering
the plenum is used to ventilate a basement (as in the HAWL and
the SSEES) or other areas (such as the perimeter offices, in the
HAWL) then the value of Ab in Eq. (7) will be larger than the
value used in Eqs. (3) and (4). To produce the target areas in
Table 2 it has been assumed that all air entering the plenum
supplies only the lightwell.
3.5. Sizing air outlets from lightwell
The gross structural area of the outlets from the lightwell to
each floor, Alo, can be given by:
Alo Qf
 100 % (8)
Af vo C c DTblo
where Af is the area of the floor to be ventilated; Qf the heat load
density on the floor (W/m2); blo the proportion of the inlet that
is blocked by obstructions; and vo is the speed of the air leaving
the outlets (m/s). At preliminary design stage the value of Qf
might simply be taken as Qb and refined as the occupancy for
each floor becomes better defined.
The air outlets from the lightwell are located adjacent to
occupied areas. In fact, the available daylight and fresh air,
together with a structure (the lightwell) on which to mount
services, makes the lightwell perimeter an ideal place to locate
work surfaces. Care must be taken, therefore, to avoid cold
 K.J. Lomas / Energy and Buildings 39 (2007) 166181
draughts, especially in winter. Thus the air speeds must be
limited (and, in winter, the air temperatures not too low) but a
displacement flow regimen, supplying all the occupied floor
area, must be achieved. In general, guides (e.g. [24]) suggest an
upper value of air speed of about 0.15 m/s. However, in summer
cooling mode, which is the design condition being considered
here, the speed can be higher (because the supply air
temperature will be elevated) therefore, for these sizing
purposes a value for the air speed, vo , of 0.3 m/s has been
assumed (Table 2).
The structure associated with the lightwell, the reheating
devices, and the dampers with their framing and louvers (or
other flow control objects), will introduce blockage. This might
mean that only 50% of the structural opening is actually free
area so a reasonable value for blo at preliminary design stage is
0.5.
The calculated free area for outlets, e.g. 2.4% of floor
area for an internal heat gain of 30 W/m2 (Table 2) does not
seem large. However, the inlets serving an entire floor will
cluster around the perimeter of the lightwells and the top of
the inlet may need to be no more than about 0.7 m above the
floorto ensure a displacement flow and to fit below work
surfaces. Thus, the length of a lightwells perimeter may
limit the free inlet area achievable, which can lead to the
lightwells being enlarged to accommodate the outlet areas
required.4 In very deep-plan buildings, the lightwell
perimeter may be insufficient and so multiple smaller
lightwells may be used, rather than a single large lightwell
(e.g. the FLL).
3.6. Sizing the stacks and air outlets
The stacks themselves are likely to be free of blockage and,
as noted above, it is reasonable to assume an air speed in them,
during ventilation cooling operation, of 0.5 m/s. Therefore the
total area of the stacks As exhausting a floor can be given by:
As Qf
 100 % (9)
Af vC c DT
As noted above Qf might simply be taken as Qb for
preliminary design. The required area can be distributed
around a number of equally sized stacks or in some other way
(the FLL has stacks and a central lightwell, the HAWL stacks
of differing cross-section and the SSEES stacks and a double
facade).
As successive floors exhaust into each stack, the volume
flows of air to be carried increases, thus the cross-sectional
areas could increase up the building. The central lightwell of
the FLL visibly illustrates this; it enlarges from 36 m2 on the
ground floor to 82 m2 at the roof topa form which is
consistent with daylighting considerations. At the level of the
stack terminations, the total area of all the stack outlets from the
4
The authors and architect have considered articulating lightwell perimeters
to artificially increase their perimeter lengthbut this can be costly, construc-
tionally difficult and in conflict with interior planning ideas.
175
whole building (Ast) will be given by
Ast Qb
 100 % (10)
Ab vC c DT
values for different Qb are given in Table 2.
Ideally, the stacks should be vertical and straight and
terminate above the roof line; as noted above, the top floor may
need to be ventilated separately and/or the stacks might be
internally partitioned.
The stack-top terminations can become rather large because
they must provide the same free area as the stacks which they
surmount but also: prevent rain penetration; and include insect
or bird mesh, dampers and devices to overcome unwanted wind
pressures. They must also have a suitable aesthetic appearance
as they may be the most striking visual feature of stack-
ventilated buildings.
The inlets to each stack will be located at high level, just
below the ceiling soffit in order to drain warm, stale stratified
air, and they will contain dampers to control the airflow. Thus
the structural area of the openings into each stack will exceed
the area of stacks calculated from Eq. (9) in proportion to the
degree of blockage caused by the dampers, e.g. by 50%. At
preliminary design stage it is unnecessary to size these
openings; however, if the number of stacks provided on a given
floor is small and if the stacks present a narrow face towards the
space, it may be difficult to get the necessary outlet area into the
stack. The stacks in the HAWL and especially the SSEES
building overcome such difficulties by presenting a wide stack
face towards the occupied spaces.
4. Comparison of as-built and target opening areas
It is instructive to compare the as-built areas of the openings
in the three case-study buildings with the target areas suggested
by the above sizing method for a number of reasons: to
understand which of the target opening areas are, in practice,
the most difficult to achieve; to illustrate the extent to which
opening areas deviate from the target figures; to illustrate how
design ingenuity can enable the desired free inlets to be
achieved in difficult circumstances; and, finally, to act as a
springboard for discussion of the finer points of these buildings
designs.
From the as-built dimensions (as given in Appendix A) it is
possible to obtain: the total floor area to be ventilated from the
lightwell, Ab; the cross-sectional area of the lightwell, Al; the
feasible maximum area of the outlets from the lightwell, Alo; the
area of the outlet from the plenum into the lightwell, Apo; the
gross area available for ambient air to enter the plenum, Api; and,
finally, the cross-sectional area of the stacks available to exhaust
the air from the building, Ast (see Table 3). These as-built areas,
expressed as a percentage of Ab, are compared with the target
areas intended for preliminary architectural design (Table 2) in
Table 4. The target areas are calculated using Qb values of 30 W/
m2 for the FLL and SSEES and 45 W/m2 for the HAWL, which
were the values adopted at the preliminary design stage.
It is evident (Table 4) that, in all three buildings, the as-built
cross-sectional area of the lightwells exceeds the target free
176 K.J. Lomas / Energy and Buildings 39 (2007) 166181

Table 3
As-built areas of main airflow apertures in the three case study buildings
FLL SSEES HAWL
2
Building Gross floor area of building (m ) 8161 3380 3468
Gross areas of floors 2228 m2 (G-2) 303 m2 (G) 665 m2 (14) 1165 m2 (G-2)
Lightwells Total building floor area serveda (m2) 4  1858 2936 2283
Cross sectional area (m2) 4  38 36 73
Perimeter lengthb 24.4 m (G-2) 12.2 m (3) 24 m (G-4), 19 (5) 34 m (G-2)
Total feasible maximum outlet areac (m2) 4  60 76 71
Plenum Depth at outlet into lightwell (m) 1.5 0.80 0.93
Gross area at outletd (m2) 4  18 19 h 32
Depth at inlet at building perimetere (m) 1.4 n/a 1.45
Gross area at inletf (m2) 4  36 15 55
Stacks Cross sectional areasg (m2) 160 33 47.4
2
Values in bold, rounded to the nearest 1 m , are used to calculate as-built statistics for comparison with preliminary design target values in Table 4.
a
Excludes the lightwell itself and all areas not ventilated from the lightwell(s), e.g. stair wells, mechanically ventilated areas (e.g. WCs), the basements and, in the
HAWL, perimeter offices directly ventilated from the facade.
b
Stated perimeter length includes curved corners of SSEES lightwell, length excluding corners is 18 m (G-4).
c
Presumes inlet heights are a maximum of 0.7 m (i.e. below desktop). Curved lightwell corners are not a feasible outlet location in the SSEES building. E.g. FLL
0.7  (3  24.4 + 1  12.2) = 60 m2; 24.4 m perimeter on G, 1 and 2 and 12.2 m perimeter on floor 3.
d
E.g. FLL = 1.5  12.2 = 18.3 m2 as only two sides of each lightwell served from plenum. The perimeter, excluding the curved corners, is used for SSEES.
e
Depth of plenum slot at the perimeter of the building for FLL and HAWL, not applicable at SSEES which has discrete air corridors and apertures as inlets.
f
Structural area, i.e. excluding any obstruction from louvers, dampers, grills, mesh, etc.
g
Area of all stacks at level of terminations, except HAWL which is area at entry to roof plenum, FLL includes area of the central lightwell.
h
Additionally there is 26 m2 of inlet at the head of the lightwell.

area by a considerable amount. For example, in the HAWL, the
target area for preliminary design is 1.1% of the floor area
ventilated, whereas the actual cross-section of the lightwell is
3.2% of the area ventilated. The target free areas presume that
there is no obstruction to airflow, however in the FLL, the heater
batteries across the base of the lightwell reduce the effective
free area by about 50%, i.e. to 1.0% of the floor area ventilated,
which is much closer to the target of 0.7%. Nevertheless, it is
evident that the cross-sectional area of these lightwells should
not, in practice, act as a constraint to the flow of ventilation air.
The air supply lightwells in the FLL occupy about 6.8%, of
the gross area of the ground to second floor, the SSEES
lightwell occupies 12% of the (reduced area) ground floor and
5.4% of floors one to four, and the HAWL lightwell 6.3% for all
above ground floors. In addition to providing fresh air, the
lightwells admit daylight which brings both functional and
physiological benefits to otherwise deep plan spaces.
The feasible maximum area of outlets from the lightwells
are much closer to the target areas, with the HAWL actually
having a maximum feasible opening area that is less than the

Table 4
Comparison of as-built areas and target structural areas for use in preliminary design
Gross areas as percentage of floor area served (%) a
FLL SSEES HAWL
b c b c
As-built Target As-built Target As-builtb Targetc
Lightwell Gross cross sectional aread 2.0 i 0.7 1.2 0.7 3.2 1.1
Feasible maximum area of outletse 3.2 2.4 2.6 2.4 3.1 3.6
Plenum Outlets into lightwellf 1.0 0.7 0.7j 0.7 1.4 1.1
Inlet from ambientg 1.9 1.4 0.5j 1.4 2.4 2.2
Stacks Cross sectional area at terminationsh 2.2 0.7 1.1 0.7 2.0 1.1
a
Areas rounded to nearest 0.1%.
b
Areas used to calculate as-built values taken from Table 3, methods of calculation are given below.
c
Target areas from Table 2 using whole building design heat loads of 30 W/m2FLL and SSEES and 45 W/m2 HAWL. These do not, necessarily, concur with the
heat loads used later in the design process when occupancy had been more accurately defined.
d
As percentage of floor area served, e.g. FLL 38/1858 = 2.0%.
e
Excludes curved lightwell corners in the SSEES, e.g. FLL 60/1858 = 3.2%.
f
Outlet from plenum to lightwell as percentage of floor area served by lightwell(s), e.g. FLL 18/1858 = 1.0%.
g
E.g. FLL 36/1858 = 1.9%.
h
Includes central lightwell in FLL, e.g. FLL = 160/(4  1858) = 2.2%.
i
But the horizontal heater batteries reduce the free cross-sectional area by 50%, to 1% of the floor area served.
j
Additional 26 m2 of inlet in the head of the lightwell for ventilation cooling giving an extra inlet area of 0.9% of the floor area served.
 K.J. Lomas / Energy and Buildings 39 (2007) 166181
target value (i.e. 3.1% compared to the target of 3.6%). This
arises because the single lightwell is supplying air to a large
surrounding floor plate. (Note, that in the HAWL the maximum
horizontal air inlet to outlet distance is 15.7 m, but in the other
two buildings it is only 12 m, see Appendix A.) The HAWL also
has a higher design internal heat gain, i.e. 45 W/m2 compared to
30 W/m2 in the other two buildings. To overcome the limitation
in available area, top-hung canopy windows operated by long
push bars are used at the outlets from the lightwell.
Additionally, the top-floor, a design studio, has operable
clerestory windows, so that it can be ventilated with cool air
directly from ambient (see Fig. 8).
Overall, these results illustrate the more general point that it
tends to be the length of the lightwell perimeter necessary to
achieve a desired outlet area, rather than the cross-sectional
area of the lightwell itself, that determines the overall lightwell
dimensions. Put another way, at the preliminary design stage
the size of the lightwell is likely to be determined by the area of
outlet needed around the perimeter (Eq. (8)) rather than the free
area required for the lightwell itself (Eq. (3)).
In all three buildings, the area of the outlet from the plenum
into the lightwell is close to, or in excess of, the target area. In
the SSEES building, the as-built area only just meets the target
value (0.7% of floor area), but this free area was difficult to
attain because the level of the basement and ground floor were
set by the site levels and, of course, head height had to be
retained within the basement.
The inlets to the plenum at the building perimeter were
larger than the target value in the FLL and HAWL but much
smaller in the SSEES building (0.5% as-built, compared to the
target value of 1.4%); again, this was due to the difficult site
configuration. The vehicle delivery route at ground level, which
loops round the back of the building (see Fig. 6), prevented the
use of a slot-type plenum inlet (as used in the FLL and HAWL);
so air corridors were used at the back of the building and four
large apertures at the front. The restricted area of inlet is
overcome when the building is in natural ventilation cooling
mode through the provision of air inlets at the head of the
lightwellthese inlets provide an additional 26 m2 of opening,
equivalent to 0.9%, of the floor area ventilated by the lightwell.
Thus the total inlet area in this mode of operation is 1.4%,
which is comparable to the target area. The area of inlets to the
SSEES plenum is, in fact, just sufficient to meet the fresh air
needs of occupants and it is during this winter time mode of
operation that the air needs pre-heating before delivery to the
lightwell.5
Although, in the HAWL, the as-built gross inlet area to the
plenum is marginally larger than the target value, not all the air
serves the lightwell; some ventilates the basement and some is
fed up to perimeter offices (Fig. 8). However, when the building
is in natural ventilation cooling mode, the clerestory windows
to the top floor studio and the operable windows to perimeter
5 outlet areas into the stacks can be further refined by, in effect,
Assuming a ventilation requirement of 10 l/s per person, and that each
person occupies 10 m2 of building floor area, the target gross area of inlet
required is 0.4% of the floor area served; which is comparable to the area
provided of 0.5% (Table 4).
177
offices enable these spaces to be ventilated directly from
ambient, thereby overcoming any restriction imposed by the
plenum inlet.
Because the plenum inlet area was barely large enough in the
SSEES and HAWL, particular care was taken to reduce the
blockage caused by heater batteries, by positioning them at a
raked angle (e.g. Fig. 8). In the HAWL, the potentially severe
restriction of the insect mesh, rather than bird/rodent mesh, was
limited by folding the mesh (effectively extending the plan
length over which it was distributed).
In all three buildings, the cross-sectional area of the stacks at
the level of the terminations (and in the case of the FLL, the
stacks plus the central lightwell) are in excess of the target
areas; in the FLL by a factor of 3 (2.2% compared to a target
value of 0.7%). These larger cross-sectional areas are the result,
in part, of adjusting the areas in line with the stack ventilation
calculationsby enlarging the outlets, relative to the inlets, the
neutral pressure level can be encouraged to settle higher up the
stacks reducing the likelihood of back flow into upper floors.
This was a particular concern during the design of the FLL.
Whilst the stacks in the HAWL seem amply sized, they also
ventilate the basement and the perimeter offices, which are fed
with fresh air directly from the plenum rather than from the
lightwell.
Overall, it is evident that the as-built structural opening areas
in the SSEES building and the HAWL are rather closely aligned
to the target areas proposed for preliminary design purposes,
whereas in the FLL, the opening areas are conservatively larger.
This general difference results, in part, from the growth in
confidence of the design team with successive buildings. The
SSEES and HAWL also have supporting mechanical summer-
time cooling systems which can be activated when the natural
displacement ventilation cooling fails to achieve the desired
internal temperatures.
5. Design development
The forgoing sections have shown how to calculate the
overall sizes of the components of the ANV systems at the
preliminary design stage. However, as designs develop, other
factors must also be considered, for example: the provision of
air transfer ducts, or labyrinths, to enable air to flow into (at low
level) or out of (at high level) cellular spaces, as in all three of
the buildings discussed here (Figs. 4, 6 and 8); the introduction
of acoustic absorbers, especially between spaces with different
noise level expectations (e.g. in the stacks of the HAWL
between the design studio (floor 2) and the library and office
floors below); and, of course, the fine adjustment of the areas of
inlets and outlets, to and from, individual spaces to reflect their
individual design heat loads and the different stack heights.
These area adjustments can be made by recalculating the target
areas in Table 2 using the actual heat loads in the individual
spaces (in conjunction with the floor area of the space). The
using more realistic air flow velocities for the stacksthese can
be calculated using the well known stack ventilation equations
(e.g. [24]).
178 K.J. Lomas / Energy and Buildings 39 (2007) 166181
Other building, climate and client-specific factors might also
need to be considered, for example, in the HAWL: the desire for
operable office windows (which led to offices having dedicated
stack outlets and inlets directly from the plenum); and the need
to integrate the HVAC system, and thus the provision of a return
air path from the exhaust stacks to the plant room (which led to
the use of the roof plenum).
Once the design is rather well developed, the likely
temperatures and airflows need to be determined using more
sophisticated analyses methods. Typically, as with the three
buildings described here, this will involve dynamic thermal
simulation modelling, to understand the time-varying
behaviour of the building, and CFD analyses to evaluate
airflows and temperatures under chosen critical conditions
(e.g. [15], FLL; [21], HAWL). Other computer models might
be used to evaluate solar heat gains and daylight levels and
physical models might be used for wind tunnel tests (e.g. of
termination designs) or for water-bath modelling of internal
airflows. Such analyses can reveal critical design flaws, for
example in the SSEES building it was found that the larger
stacks at the rear of the building could draw air into the top of
the double facade and across the floor plates, thus turning the
double facade into an inlet rather than an outletwhich
would generate very cold draughts, especially in winter, when
the condition was most likely to occur. In the final design a
glazed screen separates the front of the building from the rear
(see Fig. 5).
The building geometry and opening areas, as described
above, are determined by the volume flows of air necessary for
effective natural ventilation cooling in warm summer condi-
tions under normal occupancy. Under other circumstances the
openings required for airflow can be much smaller: e.g. in
winter when pre-heated air to heat only the fresh air
requirements is needed; at times when the internal heat loads
are low, outside of occupied periods; to provide appropriate
night-time ventilation; when the spread of fire and smoke must
be controlled; and, in hybrid buildings, to control mechanically
driven airflows. A building energy management (BMS) system
is, of course, the most appropriate system for effecting such
control. It would take inputs from air (and possibly thermal
mass) temperature sensors, CO2 (or volatile organic compound
(VOC)) sensors, smoke detectors, and, in some hybrid
buildings, possibly humidity sensors, and send output signals
to control the dampers, windows (and possible shading
devices): some insight into the control strategy from the
HAWL is given in [20]. The definition of a suitable BMS
control strategy, its programming, and its subsequent refine-
ment during the commissioning and early post occupancy
period, is an area of ANV building design which would benefit
from further research.
Preliminary data from the FLL illustrates the good
summertime cooling performance and low energy consump-
tion that is possible [3]. It also confirms the need for more
post-occupancy performance data collection and analysis.
And this exemplifies a rather more general pointthat there is
rather little post-occupancy performance data for ANV
buildings.
6. Conclusions
The attributes of two different forms of simple natural
ventilation and four generic building types for exploiting
advanced natural ventilation (ANV) have been summarised,
highlighting, for each one: the architectural implications; the
indoor air quality provision; the degree of protection from the
surrounding environment; and the likely tolerance to climate
change. ANV buildings, with a central air supply and perimeter
exhaust stacks, seem to offer benefits in each of these four
areas. Such centre-in, edge-out (C-E) buildings can, in
principle, be designed so they are essentially wind neutral,
that is, wind pressures will not hinder, or assist, the airflow; this
gives added reliability to predictions of their likely, as-built,
performance.
Three case-study buildings which use the C-E ventilation
strategy are described: the Frederick Lanchester Library,
Coventry, which uses ANV; the School of Slavonic and East
European studies, which uses ANV with passive downdraught
cooling to combat the warmer central London micro-climate;
and the Harm A Webber library, being built near Chicago, USA,
which integrates and HVAC system within the ANV concept.
They each have a central air-supply lightwell, fed with fresh air
by a low level plenum and exhaust stacks arranged around the
building perimeter. The sizes and other characteristics of these
components are tabulated, along with synoptic climate data for
each site.
Based on experience gained through the design of these
buildings, simple equations, for use at the preliminary
architectural design stage, to roughly size the lightwell,
plenum and stacks are presented. The sizes are determined
by the volume flows of air needed for summertime NV
cooling. The target structural areas to be provided at the
preliminary design stage, expressed as a percentage of the
total building floor area to be ventilated from the lightwell(s)
are presented.
Finally, the as-build structural areas in the case study
buildings are compared with the target values. These
comparisons illustrate that it is relatively straightforward to
design a central supply route (e.g. lightwell) of sufficient great
cross-sectional area but that it can be difficult, particularly with
deeper floor plans and densely occupied buildings, to achieve
the target structural opening areas for air supply around the
perimeter of such lightwells. On constrained sites it can also be
difficult to achieve the target structural opening areas for the
plenum inlets. With design ingenuity, however, such difficulties
can be overcome and strategies for doing this in two of the case-
study buildings are described.
The equations and tabulated structural opening areas are
only a rough target for use at the preliminary design stage: more
sophisticated analyses should be undertaken as the design
develops.
It is hoped that this paper will give architects and engineers
the added confidence necessary to embark on the design of
ANV buildings. Their low energy consumption, relative to
typical air-conditioned buildings, is valuable in attempts to
combat global warming.
 K.J. Lomas / Energy and Buildings 39 (2007) 166181 179

Acknowledgements fruitful working relationship, and without whose assistance this

paper would not have been possible. The environmental design
The architects of the buildings described were Short and analyses were led by Dr. Malcolm Cook of the Institute of
Associates, with whom the author has had a long-standing and Energy and Sustainable Development.

Appendix A. Comparison of key features of the three advanced naturally ventilated buildings

Building name Frederick Lanchester School of Slavonic and East Harm A Webber
Library (FLL) European Studies (SSEES) Library (HAWL)

Client and context

Client Coventry University University College London Judson College
Location Coventry, UK Bloomsbury, London, UK Elgin, Nr Chicago, Il, USA
Site City center City centre Green campus
ANV Type C-E, C-C C-E C-E, E-E
Cooling method Natural Natural and PDC Natural and HVAC
Number of levels Basement + Ground + 3 Basement + Ground + 5 Basement + Ground + 2
Completion date September 2000 November 2005 Winter 2006
Structure Steel frame Concrete Steel frame
U-values
Roof 0.18 W/m2 0.20 W/m2 0.25 W/m2 K
Wall 0.26 W/m2 0.30 W/m2 0.25 W/m2 K
Window 2.00 W/m2 2.00 W/m2 2.60 W/m2 K
Footprint 50 m  50 m 31.5 m  27 m 34 m  34 m

Gross floor areas 8161 m2 (G + 1 + 2 + 3) 3380 m2 (G-5) 3468 m2 (G + 1 + 2)

942 m2 (B) 695 m2 (B) 1192 m2 (B)
Floor to ceiling height 3.9 m (G, 1, 2, 3) 3.2 m (G), 2.9 m (14), 2.6 to 4.9 m (5) 3.35 m (G, 1), 3.86.5 m (3)
Window shading Perimeter stacks, metal fins Adjacent buildings External window reveals
Approximate cost 20 m 10 m $13.5 m
Publications
By designers [2,3,11,1315] [2,1113] [20,21]
By others [1619]

Air supply lightwells

Type 4 no Lightwells Lightwell Lightwell
Levels serveda G, 1, 2, 3 G, 15 G, 1, 2
Shape Square Triangular Square
Top Sealed, glazedb Operable/ETFE Sealed, glazedb
Shading Moveable blind None Moveable blind
Bottom Opaque-heater battery Clear single glazing Clear single glazing
Sides Clear single glazing Clear single glazing Clear single glazing
Air outlet type Dampers Bottom hung windows Top hung windows
Secondary heating Trench heaters Column radiators Linear finned emitters
Cross-sectional area 4 no 38 m2 36 m2 73 m2
Gross perimeter lengthc 4  [25 m (G, 1, 2), 12 m2 (3)]d 24 m (G, 1, 2, 3, 4), 19 m (5) 34 m2 (G, 1, 2)
Floor area servede 1858 m2 per lightwell 2926 m2 2283 m2
Gross area of air inletf 18.6 m2 per lightwell 19 m2 bottom and 26 m2 top 32 m2
Maximum airflow distanceg 12 m 12 m 15.7 m
Air inlet plena
Air inlet type Perimeter slots Corridors (side) and four apertures (front) Two perimeter slots
Inlet depthh 1.4 m n/ai 1.45 m
Outlet depth 1.5 m 0.8 m 0.93 m
Gross plenum inlet area 36 m2 per lightwell 8.7 m2 (corridors), 6 m2 (apertures) 55.2 m2
Air preheating Horizontal heater coils Raked heater battery Raked heating battery
180 K.J. Lomas / Energy and Buildings 39 (2007) 166181

Appendix A (Continued )
Air exhaust pathsj

Lightwell Stacks Stacks Double Stacks and

Facade roof plenum
Rear stacks Front chimneys

Shape Square Square Triangular Rectangular Rectangular slot Rectangular

Cross-sectional areask 81 m2 20 no 3.28 m2 10 no 2.1 m2 4 no 1.2 m2 1 no 7.2 m2 10 no1.64 m2
(G, 1, 2) (15)l (3 and 4)m (G, 1, 2) and 10 no1.05 m2
(G, 1)
4 no 3.28 m2 (3) 3 no 6.5 m2 (2)
Total outlet areas 160 m2 33 m2 47.4 m2 n
Minimum and maximum stack height 7 m (2) 4.5 m (3) 6 m (5) 3.9 m (3)
15.5 m (G) 18.5 m (G) 23 m (1) 12.5 m (G)

Climateo

Latitude/longitude 52.378/1.338W 51.488/0.08 42.038/88.278W

HDDp (10 8C) 765 656 1745
HDD (15.5 8C) 2163 1896 1274
CDDq (18.3 8C) 13 69 426
CDD (15.5 8C) 77 229 776
Working hours
Over 25 8Cr 0.6% 2.9% 15.2%
Over 28 8C 0.0% 0.6% 7.3%
Mean diurnal swing
Springs 7.2 K 7.8 K 9.6 K
Autumn 5.6 K 6.4 K 10.1 K
MDMat 19.4 8C 22.4 8C 28.7 8C
MDMau 7.2 8C 7.3 8C 0.4 8C
Thermal analysis
Summer design targetv 5%/27 8C 5%/25 8C Comfort envelope
Weather filew Kew 67 London DSY Chicago TRY
Dynamic thermal Sim x ESP-r ESP-r ESP-r
Ventilation analysisy CFX CFX and Water-bath model CFX
Solar gain/daylightingz None Radiance Radiance
a
Basements either independently mechanically ventilated (FLL) or ventilated from plenum (SSEES, HAWL).
b
Ventilated greenhouse arrangement the bottom of which is sealed.
c
Length around the lightwell, in SSEES curved corners are not useable for supplying air length of straight sides is 18 m (G-4).
d
Only two sides of lightwell adjoin floor 3.
e
Excludes the lightwell itself and areas not ventilated from ite.g. stair wells, mech vented areas (e.g. WCs), and directly ventilated perimeter offices (HAWL).
f
Area of inlet from plenum to lightwells, and for SSEES also inlet at top.
g
Ie maximum air flow distance from lightwell to a perimeter stack.
h
Ie free height between insulation layers, in HAWL occurs at restricting downstand at lightwell edge.
i
Inlet is air corridors and apertures, not a plenum slot.
j
Excludes basement exhaust stacks (SSEES, HAWL).
k
Plan areas at termination of stack/lightwell/double facade (SSEES, HAWL), at entry to roof plenum (HAWL).
l
Internally divided stacks: Floors 1 and 2 combine, floors 3 and 4 combine and floor 5 linked separately.
m
Floor 3 offices and floor 4 offices internally partitioned chimney
n
Excludes exhaust from perimeter offices fed from perimeter (E-E) rather than the lightwell.
o
Data for FLL is Manchester TRY; for SSEES, London TRY; and for HAWL, Chicago TRY.
p
E.g. heating degree days to base 10 8C.
q
E.g. cooling degree days to base 18.3 8C.
r
E.g. annual percentage of hours between 8:00 and 18:00 over 25 8C.
s
Average of daily temperature swing (maximum to minimum) in March/April and October/November.
t
Mean of daily maxima for month of July for all climates.
u
Occurs in February for London (SSEES) and Manchester (FLL) and in January for Chicago (HAWL).
v
E.g., no more than 5% of occupied hours over 27 8C, for ANSI/ASHRAE comfort envelope, see eg [26].
w
See reference: [27] for Kew 67; [23] for London DSY; and [28] for Chicago TRY.
x
For all buildings, combined thermal and airflow modelling was used, for ESP-r, see [29].
y
CFX is a CFD code, see [30] for the water bath modelling see e.g. [4,5].
z
See e.g. [31] for description of radiance.
 K.J. Lomas / Energy and Buildings 39 (2007) 166181
References
[1] B. Bordass, R. Cohen, M. Standeven, A. Leaman, Assessing building
performance in use: energy performance of the PROBE buildings, Build-
ing Research and Information 29 (2) (2001) 114128.
[2] K.J. Lomas, M.J. Cook, Sustainable buildings for a warmer World, in:
Proceedings of the World Renewable Energy Congress, Aberdeen, May
2227, Elsevier, (2005), p. 26, ISBN: 0-080-44671-X.
[3] B. Krausse, M.J. Cook, K.J. Lomas. Environmental performance of a
naturally ventilated city centre library, in: Proceedings of the Conference
Comfort and Energy Use in BuildingsGetting Them Right, Windsor,
UK, p. 12.
[4] P.F. Linden, G.F. Lane-Serff, D.A. Smeed, Emptying filling boxes; the
fluid mechanics of natural ventilation, Journal for Fluid Mechanics 22
(1990) 309335.
[5] C. Gladstone, A.W. Woods, On buoyancy-driven natural ventilation of a
room with a heated floor, Journal of Fluid Mechanics 441 (2001) 293314.
[6] R. Asbridge, R. Cohen, Probe 4: Queens Building, Building Services
Journal 18 (4) (1996) 3538, ISSN 0951-9270.
[7] BRECSU, The Queens Building De Montfort UniversityNew Practice
Final Report, vol. 102, Department of the Environment, 1997.
[8] M. Swenarton, Low energy gothic: Alan Short and Brian Ford at Leicester.
Architecture Today, No. 41, 1993, pp. 2030, ISSN 0958-6407
[9] R. Bunn, Will natural ventilation work? Building Services Journal 15 (10)
(1993) 4142, ISSN 0951-9270.
[10] BRECSU, A Performance Specification for the Energy Efficient Office of
the Future, BRECSU, for the DOE Energy Efficient Office of the Future
(EOF) project, General Information Report 30, 1995, 24 pp.
[11] M.J. Cook, C.A. Short, Natural ventilation and low energy cooling of
large. Non-domestic buildingsfour case studies, The International
Journal of Ventilation 3 (4) (2005) 283294, ISSN 1473-3315.
[12] C.A. Short, Q.M. Pop, M.J. Cook, D. Fiala, K.J. Lomas, Passive down-
draught evaporative cooling of a Central London institutional building, in:
Proceedings of the 20th Conference on Passive and Low Energy Archi-
tecture (PLEA), Santiago, Chile, 912 November, (2003), p. 5.
[13] C.A. Short, K.J. Lomas, A. Woods, Design strategy for low-energy
ventilation and cooling within an urban heat island, Building Research
and Information 32 (3) (2004) 187206, ISSN 0961-3218.
[14] M.J. Cook, K.J. Lomas, H. Eppel, Design and operating concept for an
innovative naturally ventilated library, in: Proceedings of the CIBSE 99
National Conference, Harrogate, UK, (1999), pp. 500507, ISBN:
0900953977.
[15] M.J. Cook, K.J. Lomas, H. Eppel, Use of computer simulation in the
design of a naturally ventilated library, in: Proceedings of the PLEA 99
181
Conference, vol. 2, Brisbane, Australia, (1999), pp. 597602, ISBN: 1
86499 3480.
[16] A. McDonald, Celebrating outstanding new library buildings. SCONUL
(Society of College, National and University Libraries). 2002, http://
www.sconul.ac.uk/pubs_stats/newsletter/27/ARTICL27.PDF.
[17] J. Field, Breeze blocks. Building Services Journal, December, Architects:
Short & Associates, Consulting Engineers: Environmental Design Part-
nership, 2000, pp 1822 + front cover.
[18] S. Pidwell. Deep heat in Lanchester Library by Short and Associates.
Architecture Today, No. 115, 2001, pp. 3849, ISSN 0958-6407
[19] T. Monk, 2001. Cunning Plan, Brick Bulletin, Summer, pp 810 and front
cover.
[20] C.A. Short, K.J. Lomas. Exploiting opportunities for natural displacement
ventilation in a US continental climate, Building Research and Informa-
tion, in press.
[21] K.J. Lomas, M.J. Cook, D. Fiala., 2006. Low energy architecture for a
severe US climate: design and evaluation of a hybrid ventilation strategy,
Energy and Buildings, in press.
[22] H. Graves, R. Watkins, P. Westbury, P. Littlefair, Cooling buildings in
London: overcoming the heat island, Report BR431, BRE Centre for
Environmental Engineering/DETR, 2001, 32 pp.
[23] CIBSE, Guide J; Weather Solar and Illuminance Data, Chartered Institu-
tion of Building Services Engineers, 2002.
[24] CIBSE, Ventilation and air conditioning, CIBSE Guide B2, Chartered
Institution of Building Services Engineers, 2002, , ISBN: 0 900953 96 9.
[25] REHVA, Displacement ventilation in non-industrial premises, in: H.
Skistad (Ed.), Guidebook No. 1, Federation of European Heating and
Air-conditioning Associates, 2000, 95 pp.
[26] ANSI/ASHRAE, Thermal Conditions for Human Occupancy, American
National Standards Institute (ANSI), American Society of Heating,
Refrigeration and Air-conditioning Engineers (AHSRAE), Standard 55,
2004 170 pp.
[27] M.J. Holmes, E.R. Hitchin, An example year for the calculation of energy
demands in buildings, Building Services Engineering 45 (1978) 186190.
[28] W. Marion, K. Urban, Users Manual for TMY2s, Typical Meteorological
Years, National Renewable Energy Laboratory, 1995 49 pp. approx.
[29] ESRU, ESP a building and plant simulation system, user guide. Energy
Systems Research Unit, University of Strathclyde, Glasgow, 1998.
[30] CFX International, CFX 4.4 User Manual, AEA Technology, Didcot,
Oxfordshire, UK, 2001.
[31] G. Ward Larsen, R. Shakespeare, Rendering with Radiance: The Art and
Science of Lighting Visualisation, Morgan Kaufmann, San Francisco, 1998.
[32] BRECSU, Energy Use in Offices, Energy Consumption Guide 19, Build-
ing Research Energy Conservation Support Unit, Energy Efficiency Best
Practice Programme, 2000, p. 24.