Environmental Applications of Computational Fluid Dynamics - Environmental Modelling: Finding Simplicity in Complexity - page 99

Indeed, Murakami (1990) was one of the first to
champion CWE, but tempered his enthusiasm with an
awareness of the problems associated with turbulence
modelling, accuracy of the numerical schemes and the
treatment of the viscous sublayer at walls.
A contemporary paper by Selvam (1990) used a flow
domain and inlet boundary conditions that would not
be unfamiliar to today's practising computational wind
engineers - a short upwind fetch, a long downwind fetch
and u , k and

− 1.5

MMK model

− 1.0

Experiment

− 0.5

specified at the inlet appropriate to a
logarithmic atmospheric boundary layer (ABL) profile
( u - velocity, k - turbulent kinetic energy,

ε

Standard
k-

Wind

e model

- rate of dis-
sipation of k ). It took Richards and Hoxey (1993) to
justify and formalize the specification of the logarithmic
profile at the inlet of the domain, as enshrined in the
following equations:

ε

ln z + z 0
z 0

1.0 0.5 0.0 0.0 − 0.5

u ∗
κ

=

u (6.11)

Figure 6.5 Comparison of mean pressure coefficient for a cube
(Reproduced with permission from Tsuchiya, M., Murakami,
S., Mochida, A. et al . (1997) Development of a new k

u 2
∗
C µ k = (6.12)

model
for flow and pressure fields around a bluff body. Journal of Wind
Engineering and Industrial Aerodynamics , 67-68, 169-82.).

and

u 3
∗

ε = (6.13)

κ ( z + z 0 )

use of unsteady CFD, it might be thought, could lead to
a better understanding of the effect of the variations in
the wind might have on a structure. However, the next
bigmoveinCWE,thatoftheuseofLESwasdrivenby
the need to improve the mean pressure predictions. In
the late 1980s, LES was applied to simple CWE prob-
lems (Murakami et al ., 1987; Ferziger, 1990), such as the
surface-mounted cube in order to ascertain the applicabil-
ity of this approach toCWE. DespitewhatMurakami et al .
(1987) described as 'good' agreement between the numer-
ical predictions and wind tunnel experiments, the general
view was that it would be some years before LES entered
the fray for commercial CWE. Indeed Ferziger (1990)
outlines a course of action for the gradual introduction of
LES into the CWE mainstream - a situation that has not
yet been reached.
Faced with the inaccurate, but quick RANS models
on the one hand and the accurate but computationally
expensive LES on the other, the focus shifted a little
towards the development of hybrid LES/RANS models.
Foremost among these is the detached eddy simulation
(DES) model of Spalart et al . (1997). These hybridmodels
were developed for aerodynamics applications, but have
found a less than willing audience in CWE - ventilation
modellers appear to be the main proponents of this
approach (e.g. Wright and Hargreaves, 2006; Hasama

where z is height above the surface, is the von Karman

constant, z 0 is the surface roughness, u ∗ is the friction
velocity and C µ is a constant from the k - ε model. Richards
and Hoxey's work has recently been revisited and gener-
alized (Hargreaves and Wright, 2007; Yi et al ., 2009).
The focus in the development in CWE during the
1990s would appear to have been in the tweaking of the
two-equation RANS turbulence models to improve pre-
dictions of the flow around and, in particular, pressure
distributions on the surface of simple building structures.
Several notable attempts were made including the MMK
model (Tsuchiya et al ., 1997) which was able to reduce the
overproduction of turbulent kinetic energy at the upwind
roof edge of an experimental scale cube - the overproduc-
tion being a typical, yet unrealistic, prediction of the stan-
dard k -

κ

model. Indeed, Figure 6.5 shows the improved
prediction of surface pressure coefficients that result from
the MMK modifications. None of the RANS models
developed during this period were able to prove general
enough and accurate enough to convince the wider wind-
engineering community of the 'coming of age' of CWE.
At best, RANS models can predict mean pressures on
buildings for a given mean wind speed. Using this result
and quasi-static theory, perhaps some indication of the
extreme loads on a structure could be ascertained. The

ε