You are on page 1of 7

Building Serv. Eng. Res. Technol. 29,4 (2008) pp.

341–347

Equivalent diameters of rectangular and oval ducts


P Koch BSc MSc MCIBSE
Université Joseph Fourier, IUT1, Grenoble, France

To help engineers predict the pressure drop of a fluid passing along ducts of non-circular
cross-section, the concept of ‘equivalent diameter’ of a circular duct has been extensively
used. This paper re-analyses two earlier sets of research data for non-circular ducts, such
an analysis being easier now than at the time of the original research, 1948 and 1975.
This new work reveals that equations for equivalent diameter of non-rectangular ducts
lifted from the two prime sources are totally inapplicable; in short they are myths which
should no longer be reiterated. In analysing the experimental results as constructively
as possible to extract a more appropriate and correct relationship, the author finds that
the experimental results are not sufficiently consistent to draw any reliable quantitative
conclusions. Thus the experimental work needs to be repeated to verify whether
hydraulic diameter may be used as an ‘equivalent diameter’ or whether an alternative
relationship can be found.
Practical application: In showing that there are no reliable practical results to guide
engineers in calculating pressure drop along non-circular ducts, the paper demonstrates
that further practical research is necessary. In showing that the relationships used in some
texts and reputedly emanating from practical research, are both invalid, the author hopes
that their future use will cease. Until a proven relationship for ‘equivalent diameter’ is
forthcoming, there is no other choice than to use the hydraulic diameter.

List of symbols deqx diameter of an equivalent circular duct to give


c mean fluid velocity the same pressure drop, for the same volume
d diameter of circular duct flow, determined experimentally
dec diameter of an equivalent circular duct to dh hydraulic mean diameter
give the same pressure drop, for the same h height of a rectangular or flat-oval duct
velocity k surface roughness of the duct surface
l length
decx diameter of an equivalent circular duct to
p pressure drop
give the same pressure drop, for the same
w width of a rectangular or flat-oval duct
velocity, determined experimentally
qv volume flow
deq diameter of an equivalent circular duct to
give the same pressure drop, for the same A cross-section area of duct
volume flow K a constant
P perimeter of a duct
Re Reynolds number
l friction factor
Address for correspondence: P Koch, 10, Grange Vieille, 38660
 dynamic viscosity
Le Touvet, Isere, France.
E-mail: peterkoch@orange.fr

ß The Chartered Institution of Building Services Engineers 2008 10.1177/0143624408094631

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016


342 Equivalent diameters of rectangular and oval ducts
1 Introduction There is nothing empirical about
Equation (4). Its validity rests solely on an
Data for the pressure drop along the straight assumed validity of using hydraulic mean
for circular ducts is plentiful, and along with diameter in Equation (1). This validity has
pipe data was correlated by Moody1 in 1944 been of some doubt for ducts of cross section
into a non-dimensional graphical form, known having acute angles, and Huebscher2 (1948)
as the Moody diagram. For any relative rough- reported theoretical studies suggesting that the
ness k/d, and for any value of Reynolds number use of hydraulic diameter was inappropriate
(cd/), a friction factor l can be obtained for for ducts of large aspect ratios, but he did not
use in the D’Arcy-Weisbach equation: cite a reference.
p 11 Such empirical data as is available from
¼ l c2 ð1Þ Huebscher2 (1948) and Heyt and Diaz3 (1975),
l d2 are reassessed below.
which can be written in the form:
p 1 1 qv 2 2 Origin of two relationships for
¼l  ð2Þ
l d2 A ‘equivalent diameter’
For ducts of non-circular cross section, the use Huebscher conducted tests on rectangular
of Equations (1) or (2) necessitates the use of a ducts. In the descriptive part of his paper,
value for a ‘diameter’ characteristic of the non- reference is made, not to the Darcy equation
circular duct. For this, hydraulic mean diameter above (1), but to a pre-Moody equation of
has traditionally been used, defined by: Wright, namely

dh ¼ 4
A
ð3Þ p 1  c 1:825
P ¼ 0:027 1:23 ð6Þ
l d 1000
Since most design problems have as their where the variables have a mixed set of
starting point, a desired volume flow qv, the imperial units, and the constant 0.027 incor-
most convenient form of equivalent diameter porates a constant value of l and has units of
is that which will give the same pressure drop in.wg in1.23 min1.845 ft2.845.
per unit length and for the same volume flow. Making a purely algebraic substitution
When hydraulic diameter is used in con- using the hydraulic diameter (3) with this
junction with Equation (2) to determine a equation, he obtained the following relation-
circular equivalent for a non-circular duct, to ship for equivalent diameter to give the same
give the same pressure drop for the same volume pressure drop for the same flow:
flow, the following equation is easily derived:
A0:625
A0:6 deq ¼ 1:30 ð7Þ
deq ¼ 1:453 0:2
P
ð4Þ ð0:5PÞ0:25
Since this relationship was obtained using
Such an equation is designed for use
Equation (2) it may only be used with that
with Equation (6), not with the D’Arcy
equation, for which it might be safer to
Equation (1). It is of no use today.
rewrite Equation (2) in the following form:
Nevertheless, it appears to have been lifted
p 16 1 1 2 out of context and republished as if
¼ 2 l 5 qv ð5Þ Huebscher had discovered this relationship
l  d eq 2 experimentally. He had not. He made no

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016


P Koch 343

attempt to use his experimental data to Since l is required before a value of equi-
determine a relationship for equivalent diam- valent diameter can be calculated, there is no
eter. He failed to emphasise that the value of choice but to use hydraulic diameter dh in
deq obtained using Equation (7) would be for Equation (11) for both Re and k/d. The
use in: roughness k of the galvanised ductwork was
taken as 0.05 mm. The paper of Huebscher
p 12:55 1  qv 1:825 fortunately includes the basic data for the
¼ 1:825 0:027 3:055 ð8Þ
l  d 1000 many tests he took: 9 for the circular duct,
25 for the square duct and 18 for the rectan-
Heyt and Diaz concerned themselves with gular duct. These are reproduced in Table 3 in
flat-oval ducts. In their preamble, they make SI units. The paper of Heyt and Diaz does not
the same algebraic substitutions of hydraulic include as much basic information, and
diameter, as did Huebscher. Inevitably this numerical values are now only obtainable by
led again to Equation (7) above, but this was reading from a log–log graph.
disguised by presenting it in the form: Some might consider the concept of an
equivalent diameter to give the same pressure
A0:625 drop for the same flow, deq, to be an unne-
deq ¼ 1:55 ð9Þ cessary deviation, and one which in this paper
P0:25
hides, there being no direct evidence as to
It is obvious that Equations (7) and (9) are whether the use of hydraulic mean diameter is
identical, yet they have been lifted by others valid. So values of decx for direct comparison
as if one applied to rectangular ducts and the with dh are also calculated, using
other to oval ducts. Furthermore, in using the
concept of a circular equivalent, each is for 1=2c2
decx ¼ l ð12Þ
application with Equation (8), and not with p=l
Equation (5) as would be modern practice.
If dec is found to be equal to the hydraulic
diameter dh then there is no need to go further
3 Analysis of the experimental data and look at deq since Equation (9) for deq was
merely obtained by algebraic substitution of dh.
Empirical values of equivalent diameter deqx However, for comparison with the Equation
for use in Equation (5) are calculated from: (9) which is similar to the equation attributed
" #0:2 to Huebscher and to Heyt and Diaz, a factor
16 l 1=2  q2v K has also been calculated, which would be
deqx ¼ 2 ð10Þ equal to the factor 1.453 in Equation (4) if the
  p=l
use of hydraulic diameter is valid.

Formerly values of l would have been


obtained from the Moody chart but the 4 Assessment of data of Huebscher,2
curve-fit by Haaland4 (11), now greatly 1948 for rectangular ducts
simplifies the analysis of the experimental
results. Huebscher tested for a circular duct, square
" duct and rectangular duct, the dimensions of
  #
1 6:9 k=d 1:11 which were chosen so ensure that all three had
pffiffiffi ¼ 1:8 log þ ð11Þ the same hydraulic diameter of around 202 mm,
l Re 3:71
and therefore the same relative roughness.

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016


344 Equivalent diameters of rectangular and oval ducts
Even this last concept is debatable. For a Table 1 Comparison of values of experimentally determined
values of decx and deqx, for the circular duct, for differing values
circular duct, the mainstream of the flow is of duct surface roughness k
equidistant from the duct surface in all
K (mm) dh (mm) decx (mm) deqx (mm)
directions. For a rectangular duct, such as
the one tested having an aspect ratio w/h of 8, 0.05 204 220 207
most of the centreline flow is a mere 50 mm 0.02 204 208 205
0.01 204 204 204
away from the two long sides of the duct and
not 100 mm away as the notion of hydraulic
diameter would suggest.
Table 2 Comparison of values of experimentally determined
He produced several log–log graphs to values of decx (in mm), with those using Equation (11), for
show that the pressure drop incurred by the differing values of duct surface roughness k. The rectangular
rectangular duct was more than that for the duct has an aspect ratio h/w ¼ 8
square duct, which in turn was more than that k 0.05 mm 0.02 mm 0.01 mm
for the circular duct, and that all the lines
were parallel and straight. A certain element circular (dh ¼ 204 mm) 220 208 204
square (dh ¼ 200 mm) 230 213 207
of reliability is implied. Although the newly rectangular (dh ¼ 202 mm) 204 196 193
calculated values of Reynolds Number agree
with those of Huebscher, the values of l do
not. It would appear that Huebscher obtained
his values of l by use of Equation (1), rather of decx were calculated for all three ducts at
begging the question as the validity of using each of three different values of roughness.
the hydraulic diameter. Even assuming that See Table 2.
he used the smoothest of galvanised ductwork Since the only roughness which can give
having a roughness of only k ¼ 0.05 mm, his any consistency for the circular duct, is
practical pressure drop readings were all lower k ¼ 0.01 mm, this value is used for the sub-
than our modern predictions would be. sequent analysis of the square and rectangular
For the circular duct, it is inevitable that d, ducts. These results are shown in Table 3.
dh, dec and deq must have identical values. The author carried out a check on the
However, Huebscher’s practical values of variation of decx with Reynolds Number, but
pressure drop are 10% too low to give found there was no discernable trend, there
this. Both decx, and deqx are affected, but the being a wide scatter at low Re, where exper-
effect on decx is more obvious. Table 1 shows imental readings would be smaller.
the values obtained. Since Huebscher con- In summary, Huebscher’s results, from the
author’s analysis give:-
ducted so many tests for the circular duct and
because d, dh, dec and deq must have identical  For the square duct, h/w ¼ 1, de ¼ 1.035 dh
values, he unwittingly provides the necessary  For the rectangular duct, h/w ¼ 8, de ¼
data to establish a value for the roughness k, 0.955 dh
of the duct. This value turns out to be
0.010 mm, rather smoother than our modern
data would normally suggest; see Table 1. 5 Assessment of data of Heyt and Diaz,3
Relative roughness k/d plays a part in the 1975 for flat-oval ducts
analysis of the practical results, through the
evaluation of l [Equation (11)], and thus in Heyt and Diaz tested a range of flat-oval
the evaluation of the experimental values of spirally-wound ducts of 6 sizes and having
decx [Equation (12)]. Knowledge of its value a range of aspect ratios w/h from 2.0 to 4.2. The
is therefore essential. To illustrate this, values air velocity used was between 10 and 60 m/s.

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016


P Koch 345
2
Table 3 Data of Huebscher for round square and rectangular ducts using air. Hydraulic mean diameter ¼ c. 200 mm. The seams
were longitudinal

A. Circular duct, d ¼ 203.7 mm

Experimental data k ¼ 0.05 mm k ¼ 0.01 mm

Density (kg/m3) pv (Pa) p/l (Pa/m) c (m/s) decx (mm) deqx (mm) decx (mm) deqx (mm) K
1.112 1242.7 82.62 47.28 233.4 209.3 200.8 203.1 1.449
1.116 784.2 53.36 37.48 231.7 209.0 202.6 203.5 1.452
1.113 356.0 25.58 25.29 227.2 208.2 203.9 203.7 1.454
1.123 164.5 12.67 17.11 220.8 207.0 202.8 203.5 1.452
1.132 62.3 5.197 10.49 217.2 206.3 204.3 203.8 1.454
1.137 33.4 2.975 7.67 213.6 205.6 203.4 203.6 1.453
1.153 25.5 2.256 6.66 219.9 206.8 210.4 205.0 1.463
1.139 17.2 1.667 5.48 207.1 204.4 199.5 202.9 1.447
1.140 8.2 0.842 3.80 211.4 205.2 204.1 205.5 1.456
mean ¼ 220.3 206.9 203.7 203.7 1.453
B. Square duct, 200.1 mm, k ¼ 0.01 mm, dh ¼ 200.1 mm, 21.18C.

Density (kg/m3) pv (Pa) p/l (Pa/m) c (m/s) decx (mm) deqx (mm) K
1.124 1055 70.44 43.32 202.6 221.0 1.457
1.145 869.5 57.53 38.97 207.0 221.9 1.463
1.153 634.3 42.99 33.16 206.7 221.9 1.463
1.148 436 30.40 27.55 206.8 221.9 1.463
1.155 255.4 19.61 21.03 196.0 219.5 1.447
1.150 244.8 17.73 20.63 208.6 222.3 1.465
1.134 301.3 21.90 23.05 204.6 221.4 1.460
1.144 177.8 13.57 17.63 203.6 221.2 1.458
1.131 81.8 6.693 12.03 203.8 221.3 1.459
1.134 47.1 3.914 9.115 211.5 222.9 1.470
1.118 73.5 6.007 11.47 206.4 221.8 1.462
1.147 9.70 0.956 4.106 209.0 222.4 1.466
1.137 25.6 2.37 6.714 201.9 220.8 1.456
1.108 642.7 42.99 34.05 209.9 222.6 1.467
1.115 609.8 41.51 33.07 206.9 221.9 1.463
1.108 563.2 37.84 31.88 211.0 222.8 1.469
1.073 375.3 26.89 26.45 204.8 221.5 1.460
1.112 265.1 18.88 21.84 211.4 222.9 1.469
1.116 137.5 10.46 15.69 209.3 222.4 1.467
1.112 1046.6 68.56 43.39 206.7 221.9 1.463
1.112 1000.1 65.95 42.42 206.0 221.7 1.462
1.115 846.4 56.31 38.97 206.6 221.9 1.463
1.115 856.4 57.45 39.20 204.7 221.5 1.460
1.113 795.3 52.55 37.80 209.0 222.4 1.466
1.123 500.1 34.08 29.84 209.7 222.5 1.467
dh ¼ 200.1 mm mean ¼ 206.6 221.9 1.463
C. Rectangular duct, 914.4 mm  113.3 mm, dh ¼ 201.6 mm, k ¼ 0.01 mm, 21.18C.

Density (kg/m3) pv (Pa) p/l (Pa/m) c (m/s) decx (mm) deqx (mm) K
1.120 206.1 16.10 19.19 196.2 321.1 1.446
1.110 236.2 18.63 20.63 192.2 319.8 1.440
1.131 55.4 4.92 9.901 194.4 320.5 1.443
1.126 131.6 10.79 15.29 194.3 320.5 1.443
1.118 69.5 6.05 11.15 194.2 320.4 1.443
1.129 40.8 3.743 8.505 193.9 320.3 1.442
1.102 1.3 0.181 1.523 184.7 327.2 1.428
1.115 9.8 1.030 4.192 196.3 321.1 1.446
1.128 5.3 0.604 3.064 193.4 320.2 1.441
1.131 2.9 0.351 2.275 197.3 321.4 1.447

(Continued)

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016


346 Equivalent diameters of rectangular and oval ducts
Table 3 Continued

Density (kg/m3) pv (Pa) p/l (Pa/m) c (m/s) decx (mm) deqx (mm) K

1.134 1.5 0.194 1.614 194.9 320.7 1.444


1.137 7.7 0.858 3.676 189.3 318.8 1.435
1.132 22.2 2.198 6.257 190.5 319.2 1.437
1.157 13.6 1.141 4.841 190.2 319.1 1.437
1.145 67.8 6.039 10.88 189.7 318.9 1.436
1.157 40.4 3.849 8.363 186.4 317.8 1.431
1.118 912.3 64.15 40.40 194.0 320.4 1.442
1.113 409.3 30.40 27.12 195.1 320.7 1.444
dh ¼ 201.6 mm mean ¼ 192.6 319.9 1.440

Somehow, the paper gives as results, very Table 4 Data of Heyt and Diaz4 for flat-oval ducts using air.
(density of air presumed to be 1.210 kg/m3). The data has been
different values of absolute roughness for each obtained from the extremities of straight lines on a log–log plot
duct tested, ranging from 0.004 mm to
0.140 mm. No explanation of this was given w (mm) h (mm) w/h c (m/s) p/l (Pa/m)
and it was not the purpose of the investigation. 300 150 2.00 5.08 1.55
Heyt and Diaz did not reproduce as much 50.8 107
147 817
of their basic data as Huebscher, but the 483 150 3.17 5.08 1.23
author has nevertheless tried to re-analyse 50.8 106
them, that their work be not lost. There were 147 817
635 150 4.17 5.08 0.98
difficulties in reading data from log–log plots, 50.8 81.7
so extreme values at either end of the straight 152 683
lines were read so as to minimise errors. 508 254 2.00 5.33 0.82
50.8 54.8
Table 4 gives this data, converted to SI units. 152 487
In order to make some comparison with 737 254 2.90 6.00 0.81
the above results for a rectangular duct, I 50.8 42.3
152 357
have had to assume a value of surface 1041 254 4.10 7.42 0.82
roughness, which, for spirally wound duct, 50.8 27.6
would be larger than for duct with longi- 152 275
tudinal seams. To show the sensitivity by
which the results are affected by the assumed
value of roughness k, results were produced Table 5 Comparison of values of experimentally determined
values of decx and deqx, for flat-oval ducts. (Assuming a surface
for both k ¼ 0.05 mm and k ¼ 0.04 mm. See roughness of 0.05 mm, 15.58C)
Tables 5 and 6.
w (mm) h (mm) w/h dh (mm) decx (mm) deqx (mm) K

300 150 2.00 211 212 226 1.455


483 150 3.17 241 221 279 1.428
6 Discussion 635 150 4.17 254 291 331 1.493
508 254 2.00 353 381 383 1.476
737 254 2.90 393 485 473 1.515
Although values of deq were derived from the 1041 254 4.10 423 662 583 1.590
experimental results, these were only calcu-
lated for comparison with the thinking of
Huebscher and Heyt and Diaz. In order to It has to be remembered that Huebscher’s
test the validity of using hydraulic diameter paper (1948) was written very soon after
as a circular equivalent dec, only the values of Moody had co-ordinated and rationalised all
decx need to be considered. pipe and duct data in conjunction with the

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016


P Koch 347
Table 6 Comparison of values of experimentally determined For rectangular ducts,
values of decx and deqx, for flat-oval ducts. (Assuming a surface
roughness of 0.04 mm, 15.58C)
 For the square duct, h/w ¼ 1, dec ¼ 1.035 dh
w (mm) h (mm) w/h dh (mm) decx (mm) deqx (mm) K  For the rectangular duct, h/w ¼ 8, dec ¼
300 150 2.00 211 207 225 1.447 0.955 dh
483 150 3.17 241 215 277 1.420
635 150 4.17 254 284 329 1.486 These relationships were found not to vary
508 254 2.00 353 372 381 1.486 with Reynolds Number.
737 254 2.90 393 472 470 1.508 That the equivalent diameter for the
1041 254 4.10 423 669 584 1.593
rectangular duct is 4.5% smaller than the
hydraulic mean diameter, is very significant,
as the pressure drop for the same volume flow
D’Arcy equation. He may perhaps be for- would be 26% more than estimates obtained
given for not basing his pre-amble on this. using dh.
There is rather less sympathy for Heyt and More experimental work is required if a
Diaz in 1972 aping the same obsolete math- better relationship between equivalent diam-
ematics as Huebscher. eter and hydraulic mean diameter is to be
The results for a square duct and for a found.
rectangular duct (of aspect ratio ¼ 8) give The data for the pressure drop along flat-
dec4dh for the square duct which is not oval ducts is not consistent enough for any
necessarily inconsistent as the real area is confidence to be placed in them.
greater than the area of a circle of diameter
dh. For the rectangle of aspect ratio 8, dec5dh.
For oval ducts Heyt and Diaz did not test Acknowledgement
a similar spirally wound circular duct, so it is
impossible to ascertain what value of surface I am grateful to Ken Butcher (CIBSE) for his
roughness should apply. Not too surprisingly, probing questions which led to the investigation
the author’s analysis of the results does not detailed in this paper.
succeed in finding consistency. If the results
are to be believed, the comparison of the
experimentally determined dec with hydraulic References
diameter would suggest that there is no con-
1 Moody LF. Friction factors for pipe flow.
nection between equivalent diameter and dh. Trans ASME 1944; 66: 671.
2 Huebscher RG. Friction equivalents for round
square and rectangular ducts. ASHVE
7 Conclusions Transactions 1948; 54: 101–118.
3 Heyt JW, Diaz MJ. Pressure drop in flat oval
The equations for equivalent diameter, reput- spiral air duct. ASHRAE Transactions 1975;
edly resulting from experimental work by 81(2): 221–232.
Huebscher (rectangular) and by Heyt and 4 Haaland SE. Simple and explicit formulas for
Diaz (flat-oval) are nothing of the sort and the friction factor in turbulent flow. ASME
are identical. They are both invalid. Journal of Fluids Engineering 1983; 105(3): 89–90.

Downloaded from bse.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016

You might also like