KẾT CẤU MỚI THE ENVIRONMENTAL CONSEQUENCES OF A BUILDING WITH A WIDE SPAN

105
THE ENVIRONMENTAL CONSEQUENCES OF A BUILDING
WITH A WIDE SPAN
Max Fordham
Max Fordham & Partners
Fig 1 Model for a competition in Pottsdam designed by Straub & Vogler
This paper examines the environmental consequences of
a building with a wide span.
A bridge is wide span but it does not make the kind of
environmental impact which concerns me.
I take it that much of this symposium is concerned with
structures like the Dome :-
Sports arenas (Sydney 2000 Stadium), cricket schools,
greenhouses (the Great Glasshouse at the National
Botanic Garden of Wales), cities in Alaska, garden
centres (Manheim), the Albert Hall, the Crystal Palace,
the Pantheon, Gothic Cathedral. Enclosure of Manhattan
(Buckminster Fuller)
A wide span building is a building where the enclosing
envelope is the top surface and has a wide span. This
means that the top surface is light in weight and is
designed to carry only the minimum load.
The wide spanning surface can hardly carry the load of
another storey of accommodation and so we are
considering single storey buildings.
I am not a structural engineer but I suspect there are
reasons for wide span structures to be tall at least in
places. Even though the enclosure of a wide span
structure may be light, gravitational forces will be
developed. These forces are vertical and a vertical
component of forces generated in the structure will have

106
at least
to
equal
the
weight. Members which
are
inclined
to
the
horizontal
can
generate vertical forces even
without bending moments,
so
most wide span structures
seem
to
be
carried with arches
and
catenaries.
I dare
say the
structural papers will expand
on
this theme
and give some credit
to
Frei Otto

who
developed ideas
for forming
and
finding shapes which could support wide
span enclosures with bending moments only needing
to
be developed
for
perturbing forces.
The
wind
is a big
perturbation.
So we get
single storey, wide, high spaces
with lightweight cladding.
At
a
smaller scale there
are
cyclones which
are
turbulent
and chaotic
and
provide
the
variability
in our

weather.
The scale here
is
about 1000km.
Cumulo nimbus clouds
are
limited
by the
height
of
the
atmosphere where
the
plumes
of
buoyant
gas fan out
when they reach
the
tropopause.
The point about this introduction
is
that
the air in
very
large spaces
is
mixed
by
turbulent convection currents.

The other point
to
notice
is
that
at
the
tropopause there
is
a temperature inversion
(the
temperature increases with
height),
and the
atmosphere
is
stratified.
VENTILATION
The world
is
ventilated
by
natural movement
of air.
Inequalities
of
heat distribution drive
the
climatic
air

movements.
The
convection currents
in
the
atmosphere
are generally powerful enough
to
dilute
the
pollutants
which
we
generate. However, pollution
due to
industrialisation
is not
always removed
by the
wind.
The
great
fog in
London
in
1952,
which killed
in
excess
of

4000 people
who had
vulnerable lungs,
and the
photochemical smog
in
Los
Angeles
are
examples where
the natural
air
movements
in an
anticyclone
are not
strong enough
to
ventilate
a
city.
When considering
the
ventilation
of
individual parts
of a
town, there
is the
concept

of
canyon streets where pollution
is dispensed slowly
and the
concentration
of
Carbon
Monoxide
is
a
problem.
The thermal equilibrium
of
any
part
of
the world
is
affected
by ventilation.
We
do
not
particularly think
of
the
hottest
parts
of the
world

as
being places with inadequate
ventilation,
but the
hottest parts
of
the
world
are
generally
places subject
to
anticyclonic weather with
low air
movement, where
the
temperature builds
up.
A wide span structure clearly modifies
the
ventilation
of the
space
it
encloses. Ventilation
is
needed
to
disperse pollution
and

to
control
the
thermal conditions
in a
space.
The
ventilation
of a
fire
is a
particular case where these
two
requirements
are
combined.
I believe
we
have
to
understand some basic principals about
fluid flow
in
enclosures
so
that
we
accept
the
possibility that

ventilation
can
look after
itself.
Imagine looking
at a
plan
of
Paddington station
and
wondering
how to
provide adequate
ventilation
in
this deep structure with combustion engines
inside.
In
fact,
the
enclosure
is
adequately ventilated
by the
openings
in the top and at one end.
We live
in
a
very large enclosure.

It is
about
12km
high
and
7000km radius. Convection currents drive strong
air
movements.
The
major convection cells
(the
trade winds)
have
a
scale
of
about 10,000km, even though
the
atmosphere
is only
12km
high.
In
a
large span structure, convection cells
are
likely
to
develop, with
a

scale characterised
by the
size
of the
space
itself.
When
a
strong temperature inversion develops
at
the top
of
the
space,
a
stratified layer
is
likely
to
develop.
Hot gas<ts flowing
out
through vent
1>
V"
twfyl"
V
\
A;r
«ntrc.nad by

rising
%
y
strccm
ot
goscs
^LAl
Formation of a layer of hot gases
Fig
2
Design
of
roof-venting systems
for
single storey buildings,
Fire Research Technical Paper
No 10; 1964, Fig
2
page
4
Picture
on
Fire Research.
Warm
air
discharge
Fig
3
Heat source forming
a

plume
in a
dome
Fig
4
Multiple heat sources
and
plumes
in a
dome
107
If we need to ventilate a space so as to remove heat from
'h' the lower occupied section of a tall space 'H\ then the
amount of air rising in the plumes to height 'h' must be
extracted from the upper reservoir and replacement air
allowed to enter at the base of the space.
During sunny weather the temperature of the skin will
build up. The temperature build up depends on the wind
speed, the reflection coefficient to short wave radiation
- light, and the emission coefficient to long wave
radiation. It is likely to be 10°C to 20°C above ambient.
The entering air must not cause uncomfortable air
movements in the occupied zone and it must not disrupt
the stability of the hot air reservoir. Most of the envelope
of a wide span structure is likely to be subject to a low
pressure zone while the elevated pressure stagnation
zone will be developed at low level on the windward
face.
It may be possible to rely on input air flowing only from
the windward with discharge at the top. Often with

strong winds, too much air will come from the windward,
and it will flow out to leeward. If the resulting air
currents are tolerable then the wind driven ventilation is
a good solution. If the air currents are too strong then the
air inputs on the windward side have to be throttled off
so that the air enters on the leeward faces driven by the
thermo syphon effect of the reservoir of hot gas. This
pattern of air movement requires input around the
perimeter and air outputs at the top. I was tempted to
illustrate how this thinking might be applied to the
Millennium Dome.
It is important that the warm air is stratified and stable
above the occupied zone of the building.
Where a hot zone of air is lying above a cooler zone,
there is a region with a strong vertical temperature
gradient. If this region has strong air movements with
the air speed changing with height, then the
stratification and the turbulence will act in opposition.
The Richardson number (R S Scorer 1) is defined as :-
dd_
where, 6 =
z =
U =
dU
lz
temperature
height
horizontal velocity
acceleration of gravity
If Ri >

1
then the turbulence will die down and the horizontal
air flow will be restrained below the stratified layer.
Fig 5 Hypothetical flow for the Millennium Dome
The internal area is, say, 80m high. Strong air movement
can be tolerated round the perimeter, say 0.5m/s at the
centre.
The heat from the exhibition buildings will be given out as
driven ventilation plumes at a temperature of, say, 25°C. I
suggest these discharges should be ducted above the
reservoir.
It is tempting to apply the ideas of stratification and
displacement ventilation to the heat loads in the Dome, but
the following calculation shows how quickly a plume cools
down and how much input air is needed.
Say there are 50,000 people in the centre of the Dome
(25,000 people in the exhibition buildings, and 25,000
people clustered into 13m diameter arrays of 400 people
each, producing 53kW per array, then with the addition to
solar gain reaching the floor at, say, 100W/m
2
, the
temperature rise in the plume is much less than 1°C.
Qf
heatsource
r
0
=
5a
0

radius of heatsource
Fire Research Paper No 7 HMSO 1983 1968 reprlr
acceleration due to gravity 9 8" m-'sec'
temperature above ambient
:
C
temperature "K
heatsource kW
density of air at source temp 1 2
kg/rrf'
specific heat of
a>s
kJ/kg
height from point source to plane defined by "y
Fig 6 Fire Research Paper 7
108
The air flow into the reservoir would be over 4000m
3
/s.
In fact the ventilation requirement for 50,000 people is
nearer 1000m
3
/s and the plume is likely to recirculate
inside the building.
lOOOmYs of ventilation picking up a heat gain of, say,
100W/m
2
plus 50,000 people is 12,000kW and raises the
temperature of the air by 10°C.
If this stratifies in the top 30m of the Dome, the stack

effect is about lOPa, giving a velocity of about 3m/s
through the vents. So 300m
2
of roof vents, which need to
be controlled to prevent too much ventilation in cold and
windy weather, should be supplemented with, say, 400m
2
of vent round the perimeter.
Of course, keeping the rain out has to be addressed.
This crude analysis should be a precursor to more
modern techniques such as salt water modelling and CFD
(computational fluid dynamics).
Salt water modelling gives an accurate representation of
convective turbulence, but it is not able to model the
momentum of incoming air, nor the thermal capacity of
the bounding surfaces.
National
Theatre
warm stage
-
cool auditorium
smoke flows towards fly-tower
cool
stage
-
warm auditorium
smoke flows towards auditorium
Fig 7 Air flows in the National Theatre
Then at the Royal Exchange Theatre Manchester which
is a very large enclosure with a theatre inside it

We worked with Professor Manfredi Nicoletti on an
entry for the Cardiff Bay Opera House Competition
which was to be a large glass enclosure.
The modelling of a thermal plume has to be carefully
considered. A person is modelled as a 100W heat input.
The model should equate to 100W emitted from, say, a
450mm diameter source with an initial plume of, say, 20
1/s at a 5°C temperature rise. It should not be a 100W gls
lamp,
say, 100mm diameter with a plume of, say, 5 1/s at *
20°C temperature rise. The difference here is
represented by different flow rates of salt solution at
different densities.
Computational fluid dynamics cannot deal properly with
turbulence. Assumptions about the amount of turbulence
have to be built into the finite difference equations as
dummy transfer constants. The constants have been
developed for heat transfer in jet turbines, nuclear power
station boilers, meteorological forecasting, and other rich
applications. The constants for buildings certainly need
examining as far as I am concerned before I am happy to
have them used in very large and very small spaces with
very different Reynolds numbers.
Fig 8 Competition entry for the Cardiff Bay Opera House - Interior.
Designed by Manfredi Nicoletti
These ideas about ventilation and fluid flow are based on
experience. When commissioning the National Theatre,
the convection currents seemed to be preventing the
ventilation operating as required by the fire officer. We
turned up the thermostat controlling the stage fan

convectors and the convection currents reversed.
Fig 9 Competition entry for the Cardiff Bay Opera House
Thermal Balance
109
Fig
10
CFD for the Cardiff Bay Opera House
-
Velocity Vectors
LIGHTING
One
of
the immediate consequences
of the
single storey
property of wide span structures is that the spaces can be roof
lit
The requirements for light to enable people to see are not to
be defined too simplisticaUy.
The eye
is
like
a
camera with
a
variable aperture pupil
to
control the amount of light entering and with
a
light sensitive

membrane retina to generate signals which are focused and
transmitted
to
the brain. The signals are processed
by the
brain and the nerves in the retina to generate sensations which
we interpret as "seeing". The retina can integrate the photons
it receives in
a
variety of ways.
I
CflRDff
86V
OPERA
MOUSF
'eMPERATI
Fig 11 CFD for the Cardiff Bay Opera House
-
Temperature Profile
The air flow and temperature were modelled using CFD
as shown on the figure above.
The
air
flow model immediately suggests that
the air
entry slot should
be
above head level
and
slanted

upwards.
There was another issue about the temperature plot.
We
initially modelled the solar gain onto every node
of
the
lowest part
of the
floor level.
The
temperature plots
showed very high temperatures
on the
floor which
we
couldn't understand. However,
I
vaguely remembered
a
lecture during which
it
was explained that mirages
in
the
desert sometimes occurred when the morning sun heated
the desert surface
and the
invert layer
of air to a
high

temperature causing
a
very strong temperature gradient
close
to
the ground and making the mirage. The hot
air
did not rise as
a
convection current because there was
no
trigger
to
start
the
convection
at any
point.
The
relaxations calculation process
of
CFD would be similar
if all nodes were
at
the same temperature with the same
heat gain.
The
heat gain would
be
equated

to the
temperature rise
and the
relaxation process might
be
stopped before getting
a
proper answer. We changed the
input, putting
in a
double batch
of
solar input to half the
nodes and immediately the answers looked sensible like
the figure above.
I am sure
I
will raise
as
many questions
as I
answer
in
this
brief summary
of
seeing
but I am
presenting
you

with
a
working designer's basic knowledge.
In the absence
of
light the pupil
is
wide open and too few
photons are received to see anything. The sensitivity
of
the
eye/brain system is increased and
it
takes about half an hour
for full sensitivity to be developed.
The part of the retina most sensitive to low light levels is not
in the same location as the area most sensitive to detail, so at
night air force pilots are trained to look away slightly from
a
dim object.
At low light levels, the retina integrates different coloured
photons
and
integrates them over
a
period
of
time before
transmitting
a

signal
to
the brain. Thus
the
image seen
is
monochrome and in these conditions the retina cannot catch
fast moving objects.
The sensitivity varies with age, but
a
fully adapted night eye
can "see"
a
light coloured object under an overcast night sky
at
0.001
lux.
SECURITY
(;N'«IM;t.;K]Ni;
Ronge
of
'Vidicon'
EOffiWa
ojtenuW
by
doling
leni
iris
Operating theatre
Wcli

iH
chart
Drawing ofiice
Offices, shops
Stairs, corridor
Weil
EH
s»
Upp»f limit of
VFiion
tolerance
Approximate
rong« of 'Newvici
camera, using
auto ilia .«n*
10
?
•
10"*'
i
Clear
nsght
10"*-
i
OttKCOtt
slight Ay
10 •'•
(0-4t*|
f
Range

of
'V.dicorV
and
extended
b,
infra-red
*
loop*
|2r,lx|
|
102ml«|
1
iO Oorolsjj
Fig
12
Figure 4.8 from the CIBSE Applications Manual AM4:
1991
Relative light level chart showing operational ranges
of
several types
of
camera.
110
"See"
must mean differentiate between two levels of
brightness of objects of a certain size (not too small) so
the reflective properties of the field of view are
important. When I say "see" in quotes I mean the whole
eye/brain process.
As the general light level increases, the eye/brain adapts to

the changed signals. With more photons arriving the cells
can differentiate between different colours and can send/fire
off signals to the brain at shorter intervals and the brain can
"see"
colour and fast moving images.
There is a limit to the rate at which a cell can send signals to
the brain. After receiving light and sending a signal it has to
reset
itself.
So, as the light received by a cell increases, the
frequency of sending signals increases until it is saturated
and can simply report maximum brightness. Indeed, if too
much light is received the cell can overheat and die.
I was reminded of the ancient Egyptians who worshipped the
sun god Ra. A person accused of offending the god was tied
down on their back in the open and their eyelids cut off. At
the end of the day, if they could still see they were innocent
but usually the retina was burnt out by the bright sun and this
was taken as a proof of guilt. Bright sun at 100,000 lux
(1000W/m
2
) is definitely too much.
If there is an unevenly lit field of view but all the cells report
saturation, the field of view is not being "seen" very well.
The eye needs to adapt so that the bright part of the field of
view just fails to saturate any cells. Of course the dimmest
part of the field must provide enough light for good, fast
colour vision. Thus, there is a maximum ratio of brightest to
dimmest light for good vision. The eye adapts to the average
illumination.

If a field of view is uniformly lit then the eye simply receives
a uniform vision and there is very Utile differentiation
between planes and objects. I experienced this once when I
went very early to an exhibition at the Satehi gallery. The
gallery was painted white all over and lit with fluorescent
uplighters. Light was diffusely reflected off the ceiling in all
directions. There were very few exhibits and very few
people so everything was white and I felt very disorientated.
In order to "see" the field of view must not be uniformly
bright.
Hold a pencil vertically up on a sheet of paper and look at the
shadows. I hope there are some - usually several. A shadow
represents a place where one of the brightest light sources
cannot shine on the paper behind the shadow. The light in
the shadow comes from all other sources. The contrast
between the brightness of the paper generally and the
brightness in the shadow represents the contrast between the
general diffuse light and the directional light from the source.
Set this up in a more disciplined way and we can define the
vector/scalar ratio of an illumination field.
Imagine looking at a hemisphere on a table. In diffuse
light where every surface is uniformly illuminated the
hemisphere looks like a disc. In purely vector light, the
shaded part looks black (like a half moon). A mixture of
diffuse and directed light is needed in order to perceive
the shape of the hemisphere.
Horizontal
light from right
Generally
diffuse light

30%
diffuse
Fig 13 From Table 10 - Relationship of vector/scalar ratio to
assessment of directional qualities of the lighting. IES Code
for Interior Lighting 1977
The overcast sky is very diffuse and the design minimum
figure is taken at about 5000 lux.
Under an overcast sky the field of view reflects back a
diverse field of light because of different reflection
coefficients and different colours. The lighting/seeing is
pretty adequate despite being viewed in very diffuse
light.
Inside a building the outside light is usually introduced
from a transparent window. The light then has a vector
component, and shapes and shadows can be seen even on
an object with a uniform surface. Compared to outside
the light level is reduced but the "seeing" is improved
because the light has a stronger vector component.
As an illustration of the need for diffuse light, think back
to the difficulty of seeing anything in the region of a matt
black motor car engine by the light of a single torch, even
when it doesn't wobble.
The reason for this discussion is to try to get you away
from the view that the amount of light needed in a space
can be defined by a single, simple figure of the amount
of light.
The contrast between inside and outside is important.
The shading over a motorway underpass assumes that the
speed limit is being kept and the light level can be halved
every 3 seconds.

In the Mediterranean one walks from the bright sunlight
outside at 100,000 lux into a room with the shutters
closed. It takes some time before you can see anything.
I guess the shuttered room allows about 0.1% of the
diffuse sky light (10,000 lux) into the room so that the
light level is about 10 lux.
So on entering the room the light level drops from bright
sunlight at 100,000 lux to 10 lux or
1:10,000,
ie 213.3 =
10,000. It takes 3 seconds for the eye to adapt to a
halving of the light level and therefore takes 40 seconds
(3 x 13.3 + 40) to adapt to a light level of 10 lux.
In the UK, well designed rooms with fixed windows
keep most direct sun out and then provide a daylight
factor of about 1% or 2%. I believe this should be
increased for new buildings so as to reduce the amount of
fossil fuel and electricity used for lighting. However, this
aim of mine carries an increased risk of buildings getting
too hot in summer.
The indoor cricket schools at Lord's and Edgbaston try to
exclude most direct sun and to provide a daylight factor
of 5% to 6% and so give 1000 to 1200 lux on an overcast
day.
These buildings do not fit my idea of wide span
enclosures but the cricket school at Lord's was won in a
competition where we expected the opposition would
offer air supported or other lightweight solutions. We put
forward a case for natural lighting.
A diffuse skin with a transparency of 20 to 25% provides

a light level of 1000 to 1250 lux as required.
However, in strong sun the internal light level rises to
20,000 to 26,000 lux and the direct solar gain rises to 200
to 260W/m
2
as well as long wave radiation from the
translucent skin.
Another disadvantage of the overall diffuse skin is that it
gives a very diffuse light inside. The lighting inside a
tent or marquee is very diffuse and gives poor figuring to
three dimensional shapes.
In saying this I am offering opinions which could inform
the development of the design of lightweight, wide span
enclosures. I realise that diffuse skins are provided for
indoor tennis centres, millennium domes and so on. As
structural engineers gain confidence in making
lightweight wide span structures using glass as the
membrane, then I think the design, for example, of
ventilating roof lights will be able to be developed. The
thinking about the type of lighting needs also to be
developed.
A solution with 75% to 80% opaque area with 20% to
25%
horizontal, transparent area supplies the same level
of light as a diffuse skin. It is then possible to insulate
the opaque area.
100,000
Lux
in
direction

of
sun
Light
Cloud
Blue
Sky
Overcast
Sky
20,000
Lux
10,000
Lux
£,000
Lux
No
to
option
one -
dark building fitted with fluorescent lighting.
Poor ambience,
and
lighting consumes
450KWh/m
2
a
year.
No
to
option
two -

fabric/translucent roof
In
sunlight
24,000 Lux
internally
can
quickly
be
reduced
to 2400 by a
small cloud.
\
l;>>7
JJJJlJJJJJhiflHB
1 1200
-4800
Lux I
Fig 15 Interior of the Indoor Cricket School at Lord's Ground.
Photograph by Dennis Gilbert.
The transparent areas possibly need blinds or shades to
protect the space from direct sunlight. Or the transparent
areas can be diffusing with high transparency.
Yes
to
option three -north roof light. Only diffuse light admitted.
Light levels
in
excess
of 1200
Lux except after sunset

in
winter.
Fig 14 Sawtooth roof arrangement at the Indoor Cricket School at
Lord's Ground designed by David Morley Architects
112
JHK
Fig 16 Interior of Bespak Stage 1 showing diffusing roof lights.
Designed by the Cambridge Design Group.
Fig 17 The Menil Collection, Houston. Designed by Renzo Piano.
Photograph by Hickey Robertson.
The lighting solution for the Menil Gallery is a special
case where the external condition was generally strong
direct sun (100,000 lux) and the light level inside had to
be kept low for conservation purposes (50 to 100 lux), so
0.1%
of the light was required and multiple reflections
provided a really clever solution.
So a wide span, single storey building can easily provide
adequate light.
A light transparency of 2 to 25% is feasible and provides
adequate light.
A lighting strategy can be developed to provide shading
for direct sunlight and improve the overall thermal
efficiency of the skin at the expense of a less
homogenous solution.
HEATING
The air movement in a large, tall space is likely to be
violent. The subject of ventilation and air movement has
been addressed.
The roof of the structure is likely to subtend an angle of

2p steradians from a person so that its radiant
temperature will be an important factor.
One of the main issues with wide span buildings is that if
the skin is to be light and transparent then a single or
possibly double skin of fabric is unlikely to meet the
Building Regulations for energy conservation.
A justification might run :-
1.
A conventional building deemed to satisfy the
Building Regulations
We need a light level of 1000 lux and a building with
an opaque roof will need electric light at 30W/m
2
,
using fossil fuel at a rate of 90-120W/m
2
.
An opaque roof is allowed to lose 3W/m
2
of heat
generated by fossil fuel, ie a U-value of 0.3 x 10°C
mean.
Total energy requirement 93-123W/m
2
.
2.
Lightweight wide span skin
During a 24 hour mean day with a mean inside to
outside temperature difference of 10°C, we have :-
Opaque Conventional Roof Transparent Roof

single skin double skin
U-value
Temperature Difference
Heat Loss
Heat Loss in 24 hours
Electric Light (12 hours)
(at 30W/m
2>
Watt hours per day
0.3 6 3
10°C 10°C 10°C
3W 60W 30W
72Whr 1440Whr 720Whr
lOoOWhr
1152 1440
720
This argument can be developed for different conditions.
The light saved in the summer will improve the argument
but for a lower light level, say 500 lux, the electrical
energy saved is less impressive.
Of course if the space is unheated the insulation value is
unimportant.
For a competition entry for the Cardiff Bay Opera House
(shown previously in Figures 8, 9, 10, and 11), we had
postulated a roof of 50% double glazing and 50%
insulated panels. The enclosed space was a foyer so the
electrical energy saved by lighting to 100 lux or so was
not significant.
113
However spaces which formed the brief were huddled

together rather like a Greek village or the National
Theatre.
Fig 18 Sketch by Max Fordham for Cardiff Bay Opera House
competition with Manfredi Nicoletti.
The surface area of this convoluted shape implied a heat
loss through walls and windows with ventilation which
we evaluated and compared to the heat loss of the simply
shaped envelope. The envelope had a lower heat loss
than that deemed to satisfy the building so the Building
Regulations were satisfied. The internal buildings could
have simple, un-insulated walls which notionally helped
to pay for the whole scheme.
Fig 19 Buckminster Fuller Dome over mid town Manhattan
This argument follows Buckminster Fuller's for the
dome over mid town Manhattan where the extended heat
transfer of the buildings is replaced by the smooth,
reduced envelope of the dome.
In both cases there is a problem enabling heat and
pollution to escape from the inner layer of buildings and
this is basically the ventilation problem addressed in the
next section.
Heating large high spaces depends on several levels of
consideration :-
Heat Loss
Firstly we have to decide on the heat loss. The U-value
of the cladding is important. Next, the amount of winter
ventilation; how airtight will the enclosure be and what
stack effect is likely. How much temperature gradient
will there be to increase the heat loss at the top.
Ventilation air will tend to come in at the bottom and on

the windward side. The incoming cold, fresh air needs to
be heated before it can lead to discomfort. A 4 or 5m/s
wind speed coming through the windward cracks or open
doors must be heated.
Most 50m high buildings have lobbied entrances
A 50m high stack with a 20°C temperature difference
will produce air movement through openings of about
8m/s.
The temperature gradient in the space depends on
the types of heat source. It is difficult for any part of the
space to get hotter than any individual object inside.
Direct fuel fired warm air heating is designed to be cheap
by recirculating air into a space at around 70°C and using
heaters of 300 to 600kW capacity.
The air flow is of the order 6 to 12kg/s and the air has to
be supplied at a very high velocity to ensure that it mixes
into the room before losing momentum and drifting up to
the ceiling.
At the necessary velocity noise generation is the
problem. The parameters of air flow, heat load, noise
generation, and temperature gradient have to be
considered.
In working out warm air heating we have relied on a
hypothesis advanced by Holmes and Caygill [2] and
repeated by P J Jackman [3], that :-
if thermal forces are not to dominate the pattern of air
circulation.
This relationship was originally postulated for a specific
set of conditions but we have used it successfully in
much more extreme situations.

Where a heating system provides W kg/s of air at q°C
specific heat c kJ/kg = 1 at velocity V
then the momentum M = WV
and the heat load q = Wcq
The relationship, where H = height, becomes :-
114
wv
f 0.07
WOH
or V > qH 0.07
We have used the relation at Churchill College, St Mary's
Church Barnes, and the CZWG office in Bowling Green
Lane.
At St Mary's Church Barnes we deliver 4mVs of air at
lOm/s from a nozzle at 70°C into a 10m high space. This
does not generate a noise.
Fig 20 St Mary's Church, Barnes
At Churchill College, the space is 10.5m x 18.5m x 22m
and the air supply to a dining room is at 12m/s.
I have started with crude warm air heating because I
believe it is suitable for large open space of
indeterminate use.
Of course, radiant heat has its advocates for tall spaces. I
don't want to give a detailed case as to why I am not in
favour. Where competing design solutions coexist in a
market then the reasons favouring one rather than the
other are probably marginal.
Of course, if a group of open air dining spaces were
under a wide span canopy, radiant heaters to each space
would be a good solution.

The behaviour of the air in a space with heat sources
inside it is largely defined by the behaviour of the
plumes. A plume is a rising current of air which is
warmer than the surroundings. The behaviour of plumes
is described in the book "Environmental Aerodynamics"
by Scorer and it has become a very important topic for
fire engineers.
The plume is a particular case of jet flow. It is a hot jet.
Jets are also described by Scorer and are very important
to HVAC engineers in considering how air flows in space
and how grilles need to be sized.
The best visualisation of a jet which I know and which I
expect most of you can visualise is a stream running
under a humped back bridge.
I idealise the flow in the following figure.
W i, i i i i *
« V » i i
r
1
I
1
a n *
^—
Streamline flow
^—
Flow starts
to
converge
Flow remains radial
Velocity

on
circle/sphere
^—
Stagnant: weeds thrive
^—
Rapids form
at
centre
of
stream
Level drops
T—
Eddies feed flow into side
of jet
Jet expands
Velocity drops
Momentum
is
conserved
^—
Quantity
of
flow
in jet
increases
then reduces
as
flow
is
bled

off
to
serve
the
eddies
^—
Status
quo
reinstated
Fig 21 Idealised flow at a hump backed bridge at a quiet stream.
CONDENSATION
Moisture movement in buildings is not perfectly
understood. We should remember that the moisture
content of air and water vapour has an upper limit. The
upper limit is a function of temperature.
Air
and water
•
separate out
into
cloud/mist
, .
Water
content
Upper
limit
of
water content
Temperature
Fig 22

115
The air in a building may be represented by point 'A' at
a certain moisture content and temperature.
If the air is cooled down until it is saturated - at the limit
line - moisture will deposit as condensation or dew. The
temperature is then a measure of the moisture content. It
is called the dew point.
The dew point of the air inside a building will be the
same as the dew point outside unless moisture is added.
Generally in well ventilated slightly wanned spaces
condensation rarely occurs even on single membranes.
For condensation to occur, the dew point of the air has to
be above the temperature of any surface. On a clear
night, heat is radiated to the sky and a lightweight surface
quickly cools down to below the air temperature and
often to below the dew point so that dew forms.
Of course, dew will also form on the underside of a
single membrane under this condition.
As moisture condenses on the surface, latent heat is
released so that the surface temperature tends to stabilise
a little below the dew point. The boundary layer on the
outside has a higher conductivity than the boundary layer
on the inside so more dew will form outside than inside.
The conductivity ratio is about 3 so there will be about 3
times more condensation outside than inside. As
condensation takes place, moisture is taken out of the air.
The slope of the saturation curve varies with temperature
but at 10°C ± 10°C the moisture content changes by
about 5g/kg for a 10°C change in dew point (E B H
Stevens and M Fordham

4
).
The volume of air associated with each 1 sq m of roof is
roughly the height of the space.
To reduce the dew point of 50m
1
of air by 1°C and so
prevent further condensation inside, we need to condense
25g of moisture per sq metre.
lOg/m* looks like this.
Fig 23
Condensation on the underside of lightweight sheeted
roofs - warehouses, dutch barns, and so on - is not a
problem which has been recorded in the literature.
In buildings which are enclosed, moisture is likely to be
produced.
A person produces 0.022g/s of water and if ventilated at
10 1/s raises the dew point by 3-4°C. The surface
temperature of a membrane has to be raised by this
temperature above outside to prevent condensation. A
person raises the temperature of 10 1/s of air by about 8°C
and the temperature of a single skin is raised by about a
quarter of the temperature difference between inside and
outside. It is rare for condensation to form on the inside
of single membrane surfaces unless there is a high degree
of moisture production inside, say, a laundry, kitchen or
bathroom. If the outside heat is transfer suddenly
increased or the outside temperature suddenly reduced by
a hail storm, condensation is sometimes seen on
otherwise clear building surfaces.

Condensation is more complicated in more complicated
constructions. In a double skin membrane the heat
transfer to the upper skin is reduced while the moisture
transfer through a porous membrane is hardly reduced.
Moist air is circulated through some constructions and
this has been analysed in some detail by Stevens and
Fordham.
Membrane structures need to be developed with
improved thermal resistance. The main complication
arises because of handling interstitial condensation
problems.
An insulating layer is typically going to be 100mm thick
and of mass 3kg/m
;
.
OPTION 1
In cold climates the insulating layer can be resistant to
rain penetration and placed on top of the waterproof
membrane. The membrane is then kept warm and
condensation is prevented.
Gravel, paving slabs
or
other
means
of
holding
insulation
in
place
Roof membi

Fig 24
116
In warm moist climates with air conditioned enclosures
condensation is likely to occur on top of the membrane
but that will be drained away as though it were rain.
OPTION 2
The insulation layer below the membrane. With this
option, the membrane is at or near the outside
temperature. Any moisture which flows through the
insulating layer has to be removed by a ventilation path
to outside.
Standard UK practice (Building Regulations
2
) gives as
guidance that vnetilation openings should be about 0.6%
of the roof area.
I think I came to the understanding while at the
conference that wide span structures really do have to be
designed to minimise the weight. On the other hand,
some resistance to wind uplift is a benefit.
For a target insulation resistance of R = (say) 3°C/m
2
W
the mass per unit area = p R/k
where p is the density and k is thermal conductivity.
So p/k is the significant property.
The special feature of wide span structures are that the
spaces tend to be tall, rooflit, and poorly insulated.
The size is not a reason for abandoning natural
ventilation because air movement is more likely to be

turbulent in a large space than a small one.
I would like to see light admitted as a series of discrete,
transparent areas rather than an overall diffusing
surfaces. Then the opaque areas can be properly
insulated.
REFERENCES
1.
Scorer, R S. Environmental Aerodynamics. Ellis Horwood,
Chichester, England. 1978
2.
Holmes and Caygill. B SRI A Lab Report 83. 1973
3.
Jackman, PJ. B SRI AT N 3/90
4.
E B H Stevens and M Fordham: "Interstitial Condensation in
Building Structures" in Building Services Engineering Research &
Technology, 1996.
5.
Building Regulations Approved Document F, 1991