Science and Technology for the Built Environment

ISSN: 2374-4731 (Print) 2374-474X (Online) Journal homepage: [Link]

Determine the absolute roughness of phenolic

duct (RP-1764)

Avinash Paruchuri & Stephen Idem

To cite this article: Avinash Paruchuri & Stephen Idem (2019) Determine the absolute roughness
of phenolic duct (RP-1764), Science and Technology for the Built Environment, 25:4, 409-420, DOI:
10.1080/23744731.2018.1539616

To link to this article: [Link]

© 2019 The Author(s). Published with

license by Taylor & Francis Group, LLC.

Accepted author version posted online: 03

Dec 2018.
Published online: 27 Jan 2019.

Submit your article to this journal

Article views: 247

View Crossmark data

Full Terms & Conditions of access and use can be found at

[Link]
Science and Technology for the Built Environment, (2019) 25, 409–420
Published with license by Taylor & Francis
ISSN: 2374-4731 print / 2374-474X online
DOI: 10.1080/23744731.2018.1539616

Determine the absolute roughness of phenolic duct (RP-1764)

AVINASH PARUCHURI and STEPHEN IDEM

Department of Mechanical Engineering, Tennessee Technological University, 115 W. 10th Street, Cookeville, TN 38505, USA

An experimental program was conducted to measure the relative and absolute roughness of phenolic duct systems connected with a four-bolt
flange and cleat joint. Ducts with seven distinct rectangular cross sections (internal aspect ratios 1  W=H  4) were investigated. In every
instance, the duct material thickness was 7/8 in. (22 mm), with smooth aluminum foil covering the internal surface between flanges. For
unreinforced phenolic ducts with 10 ft (3.0 m) sections connected by four-bolt flanges, the relative roughness e/Dh ranged from 0.0002 to
0.0005, with an average value of 0.0003. For similar ducts having 5 ft (1.5 m) duct sections (double the number of flanges per test section), the
relative roughness range and average doubled. It appears that the flanges are the dominant effect on relative roughness and that the ducts tested
have about 1.5 to 2.8 times more absolute roughness than the average value for galvanized steel ducts described in the ASHRAE Handbook.

Introduction panels butted together with no internal protrusions, other than

possibly not accurately aligning with each other.
Phenolic ducts are a new air distribution product that is entering SMACNA recently published the first edition of its
the North American market from Europe and China. A phenolic “Phenolic Duct Construction Standards” (2015). Those
duct system comprises phenolic panels, fabrication methods, cou- standards describe structural construction requirements, but
pling systems, and accessories to produce preinsulated rectangu- they do not include data to design phenolic duct systems. In
lar ductwork in sections up to 13 ft (4.0 m) long, with a thickness particular, technical data such as the effects of surface
of either 7/8 in. (22 mm) or 1-3/16 in. (30 mm) for different R roughness and flange connections on pressure drops are
values. Phenolic panels are made of a fiber-free rigid thermoset necessary to properly design duct systems. The objective of
phenolic insulation core faced with embossed aluminum foil this research was to determine the relative and absolute
or smooth aluminum foil on both the interior and exterior of roughness of phenolic duct systems connected with four-bolt
the duct. In addition, there are three transverse joint systems flange and cleat joints, using test and data reduction proce-
available to suit different installation and project specifica- dures provided in ANSI/ASHRAE Standard 120-2017,
tion requirements. They include (1) a four-bolt flange system, “Method of Testing to Determine Flow Resistance of HVAC
(2) drive cleats on four sides, and (3) butted ducts with fasten- Ducts and Fittings” (ASHRAE 2017a).
ers at exterior corners and sides. Of the duct joining systems, There are numerous references available in the literature per-
the four-bolt flange and drive cleat systems have the same taining to the experimental determination of pressure losses in
interior protrusions. The cleat system is simply two phenolic HVAC air duct systems; several of the most relevant studies are
herein discussed. Huebscher (1948) measured pressure drops in
round, square, and rectangular air ducts. The respective ducts
exhibited the same pressure loss per unit length when considered
Received May 25, 2018; accepted October 4, 2018 at the same air velocity, when each cross section had the same
Avinash Paruchuri is a PhD. Student. Stephen Idem, Member hydraulic diameter. The square and rectangular duct tests consid-
ASHRAE, is a Professor. ered only one duct section length, and all tests dealt with a single

Corresponding author e-mail: sidem@[Link]
absolute surface roughness. The concept of an equivalent round
Color versions of one or more figures in the article can be duct exhibiting the same friction and flow characteristics as rect-
found online at [Link]/uhvc.
angular cross sections was introduced; this analysis approach
ß 2019 The Author(s). Published with license by Taylor &
Francis Group, LLC.
remains in widespread usage among duct designers. Hutchison
This is an Open Access article distributed under the terms of the (1953) measured the surface roughness of aluminum ducts with
Creative Commons Attribution-NonCommercial-NoDerivatives longitudinal seams having 0.91 m (3 ft) section lengths. The
License ([Link] which joints were of the standard slip type, with the crimped end being
permits non-commercial re-use, distribution, and reproduction in slipped into the uncrimped end of the adjacent downstream sec-
any medium, provided the original work is properly cited, and is tion. The resulting absolute roughness of aluminum ducts was
not altered, transformed, or built upon in any way. determined to be of the order of 1/3 that of similar galvanized
410 Science and Technology for the Built Environment

steel ducts. It was noted that the measured absolute roughness the same absolute roughness. Therein welded joints appeared
was influenced by the duct diameter under study. Jones (1979) to offer relatively less flow resistance, and drawband joints
performed further pressure loss measurements in round spiral exhibited the least resistance of the four joint types considered
seam galvanized steel ducts having 3.05 m (10 ft) sections joined in the study. In addition, they investigated pressure losses in
by slip couplings spacing. The results were presented in terms of one diameter of spiral sheet metal ducts with a ribbed construc-
absolute and relative roughness. tion and having a nominal joint spacing of 3.05 m (10 ft). The
Heyt and Diaz (1975) first measured pressure loss characteris- resulting absolute roughness data were found to be a function
tics of spiral flat oval galvanized steel air ducts. They did not of the number of ribs present in the spiral ducts. Griggs and
explicitly report the duct section lengths or connection methods Khodabakhsh-Sharifabad (1992) measured pressure loss in two
considered in the test program. It was shown that the absolute cross sections of rectangular galvanized sheet metal ducts
roughness was a function of duct aspect ratio. This was attributed attached using slip and drive connections; the duct section
to the strain imposed on the duct when stretched in its construc- lengths were not clearly noted in their paper. The absolute
tion. They likewise introduced an expression for the equivalent roughness of the smaller cross section duct was observed to
round diameter pertinent to spiral flat oval ducts. Townsend et al. fall below the hydraulically smooth curve on a Moody dia-
(1994) later confirmed that the equivalent round diameter gram, whereas the data for the large size duct resided above
expression derived from the data of Heyt and Diaz (1975) was that curve. The difference between these results was attributed
accurate to within 6.2% of a rigorous solution, based on pressure to the “S” slip joint construction, which protruded different
loss data obtained from pressure loss tests performed on galvan- amounts in the flow cross section. They also simulated the
ized steel spiral flat oval ducts having section lengths of either influence of transverse joints, cross breaks, and beads by means
2.09 m (6 ft) or 3.66 m (12 ft) joined by bead slip connectors. of pressure loss tests performed on rectangular plywood ducts.
Khaire and Idem (2010) considered the influence of upstream Those tests demonstrated that friction factors increased as joint
test section geometry on measured duct relative roughness, based spacing decreased, and that thicker joints consistently yielded
on pressure loss tests performed on straight flat oval galvanized higher friction factors than thinner joints. The principal conclu-
steel ducts having cross sections of 356  152 mm (14  6 in.), sion from their study was that absolute roughness measured in
and section lengths of 1.22 m (4 ft). The test ducts were con- rectangular ducts is a function of surface condition, duct fabri-
nected to a nozzle chamber by two methods, namely, a smooth cation, and joint spacing and construction.
bellmouth and a conical transition. It was observed that the meas- Kulkarni et al. (2009) measured friction factors in round
ured absolute roughness in a test section of fixed length was corrugated spiral seam galvanized steel ducts having three
influenced by the connection method used to attach the test duct diameters, and 3.05 m (10 ft) sections joined by beaded slip
to the nozzle chamber, unless the entrance length was increased connections. One particular corrugation profile having two
to at least 12 hydraulic diameters. It was later recommended that corrugations per seam was considered in their test program.
test section entrance lengths in ASHRAE Standard 120 (2017a) The presence of corrugations yielded greater surface rough-
be extended to 15 hydraulic diameters, based on a numerical ness compared to conventional noncorrugated spiral ducts,
study presented by Kulkarni et al. (2013). leading to an increased pressure loss ranging from 7% to
Swim (1978) measured pressure loss in straight rectangu- 34% at the same average air velocity. This was quantified in
lar sheet metal ducts when lined with spray-coated rigid terms of an equivalent increased corrugated duct diameter
board, and rigid board lined with a smooth facing material. that would exhibit the same flow and pressure losses as an
In addition, the study also addressed rectangular duct sys- uncorrugated duct. Therein Gibbs and Idem (2010) measured
tems that incorporated flexible fiberglass liners, and included the absolute roughness in similar round corrugated spiral
the influence of fasteners on pressure loss. It was noted that seam galvanized steel ducts exhibiting 3.05 m (10 ft) sec-
the addition of mechanical fasteners increased pressure loss tions joined by beaded slip connections. In this instance
by approximately 40%. Likewise, it was concluded that that there were four corrugations present in each individual spiral
the wide variety of installation practices and the quality of seam. It was determined that minor differences in corruga-
the transverse joints made it difficult to accurately predict tion profiles and seam pitches did not have a significant
the pressure loss of lined duct systems. Swim (1982) meas- influence on the measured pressure loss, since the resulting
ured friction characteristics for air flow in polyvinyl chloride absolute roughness values differed by less than 15% com-
(PVC) pipes. It was deemed that the absolute roughness of pared to the results presented by Kulkarni et al. (2009).
PVC pipes is approximately the same as for aluminum Abushakra et al. (2004) investigated the influence of
ducts. It was further demonstrated that couplers added little compression on pressure loss measured in wire-wound flex-
resistance to the flow, such that the measured pressure drop ible ducts. These ducts are typically used in residential and
occurring with one coupler per 18D of pipe was the same as light commercial applications. The study demonstrated that
that for a straight pipe with smooth butt joints. longitudinal compression of such ducts could increase the
Griggs et al. (1987) presented experimental pressure loss pressure drop by a factor ranging from 4 to 10, depending
data for airflow in round galvanized sheet metal ducts with a on the extent of compression. The increased pressure loss,
welded longitudinal seam and a nominal joint spacing of 1.22 relative to that for fully stretched ducts, was quantified in
m (4 ft). These tests investigated the influence of different joint terms of a pressure drop correction factor (PDCF) that
types, that is, beaded slip, Vanstone, drawband, and welded. It accounted for the compression fraction as well as the duct
was concluded that beaded slip and Vanstone joints exhibited diameter. Weaver and Culp (2007) contended that fully
Volume 25, Number 4, April 2019 411

Table 1. Dimensions of ducts tested with 3.05 m (10 ft) and 1.52 m (5 ft) section lengths.
Number of flanges in test section
Ratio of number of flanges
W  H, mm  mm Aspect Duct section Duct section in test section (short
(in.  in.) ratio (W/H) Dh, mm (in.) length 3.05 m (10 ft) length 1.52 m (5 ft) duct/long duct)

305  152 2 203 1 2 2.0

(12  6) (8)
305  305 1 305 3 5 1.7
(12  12) (12)
610  305 2 406 3 7 2.3
(24  12) (16)
559  279 2 373 4 7 1.8
(22  11) (14.7)
559  559 1 559 5 10 2.0
(22  22) (22)
1118  559 2 744 7 13 1.9
(44  22) (29.3)
1118  279 4 447 4 8 2.0
(44  11) (17.6)

instance the fabric duct included an internal frame that was

installed to maintain a round cross section in the duct when
no airflow is occurring. In addition, they considered the case
where no internal support was present. The presence of the
frame increased the relative roughness in the duct, compared
to the case where no frame was used. It was concluded that
the absolute roughness of a fabric duct was 1.69 mm
(0.0056 ft) when a frame was present, whereas the absolute
roughness was 0.11 mm (0.0004 ft) when the internal sup-
ports was absent. Leverette et al. (2014) presented the results
of a combined experimental and numerical study of pressure
and velocity distributions in round fabric duct systems that
had regularly spaced vents located along the flow axis. Once
Fig. 1. Test setup. again two cases were considered: In one instance a ring and
spoke skeletal frame was mounted internally to the duct,
stretched flexible ducts exhibited pressure drops per unit whereas in other cases there was no internal frame and the
length that fell within ±2% of rigid sheet metal having the duct cross section was maintained solely by static pressure.
same diameter, at a given volume flow rate. However, their Although there was a pressure loss penalty associated with
study confirmed that flexible ducts having compressions the presence of an internal skeleton frame, it was determined
ranging from 4% to 45% exhibited dramatically increased that its use promoted more uniform flow distribution along
pressure loss, relative to fully stretched conditions. Urgusal the flow axis. Likewise, it was contended that incorporating
and Culp (2007) described a combined numerical and experi- variable area orifices at the midpoint along the flow axis
mental analysis of pressure loss as a function of air flow rate yielded more uniform air dispersion.
for board-supported nonmetallic flexible ducts that had com-
pressions including fully stretched, 4% compression, 15% com-
pression, and 30% compression. They contended that Tested duct sections
computational fluid dynamics (CFD) modeling was capable of
predicting the resulting pressure loss per unit length to within Factory-fabricated ducts with seven distinct rectangular cross
5% to 10% of the measured values. Culp and Cantrill (2009) sections selected to provide a range of aspect ratios
presented pressure loss data for nonmetallic wire-wound flex- (1  W=H  4), as listed in Table 1, were investigated in
ible ducts having compressions ranging from 4% to 45%, this project. The width W and height H of the ducts were
which had the maximum allowable sag between joist supports. based on measured duct internal dimensions. The major and
Their data extended the PDCF correlations of Abushakra et al. minor dimensions were carefully assessed across three separ-
(2004) to larger diameter flexible ducts. ate planes using a tape measure, and then averaged to evalu-
Kulkarni et al. (2012) measured pressure losses in one ate the test section hydraulic diameter. In every instance, the
diameter of an impermeable fabric duct used in air disper- duct material thickness was 7/8 in. (22 mm), with smooth
sion systems. Two specific cases were considered. In one aluminum foil covering the internal surface between flanges.
412 Science and Technology for the Built Environment

Fig. 3. Self-adhesive foam gasket applied to flange.

of the flange protruded approximately 0.12 in. (3.0 mm) into

the air flow cross section. Figure 3 portrays the self-adhesive
foam gasket mounted on the exposed surface of each flange.
This was installed in the laboratory per manufacturer require-
ments to minimize air leakage between the assembled duct sec-
tions. The duct sections comprising the test section were
carefully aligned and secured using four bolts mounted at each
corner. The bolts are apparent in Figure 4, which depicts an
extra application of caulking that was applied to the four
Fig. 2. Retaining screws factory installed by manufacturer exposed corners of each flange. While not required by the
mounted around the periphery of the flanges. manufacturer for field installations, this procedure was followed
because the corners of the flanges were viewed as a possible
Figure 1 shows the test setup and total duct lengths site for air leakage. The caulking was allowed to cure for at
required by Standard 120. All of the duct lengths there are least 24 h before leakage and pressure loss testing began. As
expressed in terms of hydraulic diameter Dh. For each duct shown in Figure 5, cleats were utilized to pull the flange ends
cross section, two test duct section lengths were tested. The together to minimize any possible air leakage paths. This was
section length of the first test series was 10 ft (3.0 m), and accomplished using a hammer to drive the cleat over the flange
for the second series, the section length was reduced by connections. The final step in the installation procedure was to
half, that is, 5 ft (1.5 m) duct lengths. In each case, the total apply aluminum foil tape around the outside of the flange con-
test duct length (combination of multiple duct sections) was nection to the duct, in order to completely cover each of the
at least 44 hydraulic diameters, with the test section com- screw heads and further reduce possible air leakage at these
prising a minimum of 25 hydraulic diameters, per Figure 1 locations. Taping over the screw heads was likewise not
(excerpted from Standard 120). Table 1 also lists the number required by the manufacturer. The applied tape is also dis-
of flanges in each case, and indicates the ratio of the number played in Figure 5.
of flanges in the shorter duct section case to the number in
the longer duct section case. It is important to note that halv-
ing the length of the test sections essentially doubled the Test program
number of joints for the ducts tested.
The duct sections tested were assembled according to Duct lengths upstream/downstream of the test section, as
manufacturer specifications, and per the requirements in the shown in Figure 1, were chosen to conform to ASHRAE
SMACNA Phenolic Duct Construction Standards (2015). In Standard 120. As a cost-saving measure, transformation
every instance, the phenolic ducts were supplied with a four- pieces to attach the rectangular cross sections of the test
bolt flange preinstalled on each end by the manufacturer ducts to the nozzle chamber were not employed in this pro-
using retaining screws to fasten the periphery of the flanges ject. Instead, more phenolic duct of the same cross section
to the ends of the ducts, as depicted in Figure 2. The screws as the test duct section was used upstream of the test sec-
were generally spaced approximately 6 in. (150 mm) from tion, with a minimum length of 15 hydraulic diameters. The
each corner of flange, as well as at the center of the flange inlet of the upstream section was attached by flanges directly
sides (space permitting). In every instance the screws pro- to a plywood sheet mounted on the outlet of the nozzle
jected as much as 1 in. (25 mm) into the flow cross section. chamber described in the following, as permitted by
Likewise, a stiffening ridge present on the internal surface ASHRAE Standard 120.
Volume 25, Number 4, April 2019 413

Fig. 4. Bolt connection at flange corner with caulk applied in

addition to manufacturer’s instructions.

An existing nozzle chamber in compliance with

ASHRAE Standard 120 was used to measure the flow rate
through the test section. The nozzle board contained four
long radius spun aluminum flow nozzles. Flow settling
screens located inside the chamber upstream of the nozzle
board were used to make the flow more uniform before
entering the flow nozzles; unused nozzles were blocked by
vinyl balls. The “blow-through” mode of Standard 120
(Section 7.2 b) was employed in conjunction with a 30-hp Fig. 5. Four-bolt flange assembly with cleat and tape.
(22 kW) blower located upstream of the nozzle chamber.
Flow rates were varied by adjusting the motor speed and
relatively large cross section, coupled with limitations of
changing the combination of nozzles. For each pressure drop
the test apparatus to achieve higher airflow rates. In all
test, the blower motor was allowed to run for several
cases, the maximum air velocity did not exceed 36 m/s
minutes in order to obtain steady-state conditions.
Each time a new setup was completed, an air leakage test (7090 ft/min).
was performed with the terminal end of the test setup care- The ambient dry-bulb and wet-bulb temperatures were
fully sealed using a combination of a plywood cover plate measured using a compact lab pyschrometer that utilized
and cloth-backed tape. These tests were conducted to verify liquid-in-glass thermometers. The atmospheric pressure was
that the apparatus leakage was within acceptable limits, and measured using a mercury barometer. The nozzle chamber
that the pressure taps and tubing were leak free. Standard air temperature was measured using a general purpose
120 requires that leakage “not exceed 0.5% of the minimum liquid-in-glass thermometer. The static pressure drop across
test flow rate at the maximum pressure expected during the the nozzles was measured with a liquid-filled micromanome-
test.” In all cases considered in the present test program, the ter connected to piezometric rings located 1.5 in. (38 mm)
leakage flow rates (expressed in terms of makeup air) were upstream and downstream of the nozzle board. Each ring
negligible, and in many instances beneath the detection limit was connected to a micromanometer to measure pressure
of the procedure. Care was taken to ensure that the phenolic drop across the nozzles. The upstream piezometric ring was
test ducts were not overpressurized, either during the leakage also connected through a tee to a digital manometer.
tests or during pressure loss testing with flow through the Measurements of pressure loss and upstream/downstream
system, to prevent damage to the ducts. The manufacturer static pressure in the test section were performed using the
recommended that the internal pressure of the unreinforced static taps of pitot-static tubes. The static taps of the
duct should not exceed 4 in. wg (1000 Pa). upstream and downstream pitot-static tubes were located at
After leakage testing was completed, the test program Planes 1 and 2 of Figure 1, respectively. In every instance
was conducted as follows. Pressure drops for at least nine the pitot-static tubes were installed at the duct centerline,
test points, evenly spaced over the target range of test and they were aligned carefully with the flow direction. The
velocities, were obtained for each duct cross section. Per test section pressure loss was measured using a liquid-filled
ASHRAE Standard 120, in most cases, the minimum test micromanometer, whereas the upstream and downstream
air velocity was maintained at 6 m/s (1180 ft/min) or static pressures were measured using digital manometers.
greater. The measured air velocities for 44 in.  22 in. The test section temperature was not measured, but was
(1118 mm 559 mm) phenolic duct sections were gener- assumed to equal the nozzle chamber temperature in the data
ally less than 6 m/s (1180 ft/min), because of the reduction procedure.
414 Science and Technology for the Built Environment

Data reduction and measurement uncertainty In this instance, dxi is the measurement uncertainty (precision
error) of each independent variable. For example, the hydraulic
All data reduction calculations to evaluate the friction factor diameter of the rectangular ducts was determined as a function
f and relative roughness e/Dh complied with Section 9 of of the measured duct height H and width W as follows:
ASHRAE Standard 120, which uses SI units only in its cal- 2HW
culations. In particular, the static pressure loss per unit Dh ¼ (8)
HþW
length of straight test duct was determined as follows for
each airflow rate (N data points for each test duct): It was assumed that measured duct dimensions were sub-
ject to random measurement uncertainties. In that case, the
Dps;12 root-sum-square measurement uncertainty for the hydraulic
Dpf;12 ¼ (1)
L12 diameter is given by
The numerical subscripts refer to the measurement loca- ( )1 =
tions indicated in Figure 1. The corresponding “measured” dDh H2 dW2 þ W2 dH2 2

¼ 2
(9)
Darcy friction factor for each of the N data points was eval- Dh ðHWÞ
uated at actual test conditions by
The combined volume flow rate through the nozzle bank
Dpf;12 is expressed as follows:
f meas ¼ (2)
pv1 =Dh1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X  2Dp

n s;56
The calculated friction factor and relative roughness for Q5 ¼ YCA
i¼1 i i i
(10)
the N data points for each test duct was then determined q5 1b4
iteratively using the Colebrook equation: In this case the quantity n refers to the number of open
 
1 e=Dh1 2:51 nozzles, and q5 is the air density at the nozzle chamber inlet.
pffiffiffiffiffiffiffiffi ¼ 2 log þ p ffiffiffiffiffiffiffiffi (3) Likewise, in Equation 10, Ci and Ai are the nozzle coeffi-
f calc 3:7 Re1 f calc
cients and cross sections, respectively, and b denotes the
In Equation 3, the Reynolds number was determined at nozzle diameter to chamber diameter ratio. Furthermore, the
test conditions using Equation 4: measured pressure drop across the combined set of open
Dh1 V1 q1 nozzles is indicated by Dps;56 .
Re1 ¼ (4) Section [Link] of Standard 120 indicates that b ¼ 0 for a
l1
chamber approach; this value was utilized in subsequent data
The iterative procedure was as follows. Initially, the reduction calculations. For all test conditions in the present
quantity fcalc was determined using Equation 3 by employing study it was determined that the nozzle expansion factor Yi ¼
a relative roughness e=Dh1 ¼ 0:0001. Then fcalc was succes- 1, to a close approximation. Uncertainties for Yi were deemed
sively recalculated by increasing e=Dh1 in increments of to be insignificant. It was assumed that uncertainties associated
0.0001 until e=Dh1 ¼ 0:05. The appropriate value of e=Dh1 with the nozzle cross sections Ai were negligible. However, it
for each duct test sample was that which minimized the was assumed that the nozzle coefficients and pressure drop
square of the difference of the N measured and calculated measurements were subject to random experimental uncertain-
friction factors, per Equation 5: ties. Likewise, measurement uncertainties of the air density and
X
N viscosity were fully taken into account in the data reduction;
ðf meas f calc Þ2 ! 0 (5) these are described in greater detail in ASHRAE RP-1764
i¼1 (2018). Therein, the root-sum-square uncertainty for the com-
When a measured quantity depends on more than one bined nozzle volume flow rate was determined using
8" ! #2
independent measurement, it is necessary to consider how < qffiffiffiffiffiffiffiffiffiffiffiffi X 
n
errors combine to produce the net uncertainty in the overall 2Dp dDp
dQ5 ¼ 1 s;56 s;56
Ci Ai Dps;56
measurement. Consider a dependent variable u that depends : 2 q5
i¼1
on more than one independent variable xi, such that
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! 32
u ¼ f ðx1 ; x2 ; x3 ; . . .Þ (6) 1 2Dp X
n
dq
þ4 55
s;56
Ci Ai
In the present study, it was determined that bias errors of the 2 q5 i¼1
q 5
instrumentation (based on prior calibration) were negligible, and
2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 91=
that all measurement errors were random. Such random (preci- !32 > 2
sion) errors tend to compensate, being equally likely to be both 2Dps;56 X
n =
positive and negative. Therein, since all measurement uncertain- þ4 Ai dCi 5
q5 i¼1
>
;
ties were stated at the same confidence level, the root-sum-
square error in the measured parameter dur was estimated as (11)
( 2 )1 =2
X  of The volume flow rate in the test section Q1-2 was calcu-
dur ¼ 6 dxi (7) lated in terms of the nozzle chamber volume flow rate,
oxi
based on the continuity of mass flow equation:
Volume 25, Number 4, April 2019 415

q5 Q5 Table 2. Uncertainties in measured parameters.

Q12 ¼ (12)
q12
0.6  C
In that instance, the measurement uncertainty of the test Test section temperature (1.0  F)
section volume flow rate was calculated as follows:
Dry-bulb temperature 0.6  C
(      )1 =2 (1.0  F)
dQ12 dq5 2 dq12 2 dQ5 2 0.6  C
¼ þ þ (13) Wet-bulb temperature
Q12 q5 q12 Q5 (1.0  F)
Barometric pressure 0.01 in. Hg
Referring to Equation 4, the test section Reynolds number
(34 Pa)
can be expressed in terms of the rectangular cross section as Plenum chamber temperature 0.6  C
follows: (1.0  F)
q12 Q12 Dh Plenum chamber static pressure 0.1 in. wg
Re ¼ (14) (25 Pa)
l12 HW
Nozzle chamber pressure loss 0.01 in. wg
The procedure to estimate the density and its measured (2.5 Pa)
uncertainty at any plane in the test apparatus is described in Test section static pressure 0.01 in. wg
ASHRAE RP-1764 (2018). Likewise, the method used to (2.5 Pa)
evaluate the air viscosity and its measured uncertainty is Test section pressure loss 0.001 in. wg
provided in ASHRAE RP-1764 (2018). Based upon esti- (0.25 Pa)
mates of the measured test section cross section uncertainty, Duct height/width in.
the hydraulic diameter measurement uncertainty (Equation (2.5 mm)
9), and the test section volume flow rate uncertainty Test section length 0.50 in. (13 mm)
(Equation 13), the root-sum-square uncertainty of the test
section Reynolds number was calculated using using liquid-in-glass thermometers, barometric pressure
(       2
dRe dQ12 2 dq12 2 dl12 dDh measurement, or test section pressure loss measurements.
¼ þ þ þ Section 6.1 of Standard 120 dictates that pressure measure-
Re Q12 q12 l12 Dh
ments “shall be measured with accuracy equal to or better
 2  2 )1 =2 than 1.0 Pa and a precision equal to or better than 0.5 Pa.”
dH dW For measurements of both the nozzle chamber static pressure
þ þ (15)
H W and the nozzle chamber pressure drop, as well as the test
section static pressure measurements at planes 1 and 2, the
In the present study, the Darcy friction factor was charac- measurement uncertainties exceeded the measurement preci-
terized as sion dictated by Standard 120. This was due to random pres-
2Dpf ;12 ðHWÞ Dh
2 sure fluctuations that were inevitably encountered during the
f¼ (16) experiments. In those cases the time-averaging feature of the
q12 Q212 L12 previously calibrated electronic manometers was utilized. As
Each quantity in Equation 16 was deemed to be subject suggested by Section 5.3.3 of Hickman et al. (2015), the
to a prescribed measurement uncertainty. Therein the root- nozzle discharge coefficient uncertainty Ci (in Equation 11)
sum-square uncertainty for the friction factor was determined was assumed to be ±1% in this investigation. The measure-
as follows: ment uncertainty estimates were performed with a 95% con-
8 !2  fidence level.
< 2  
df dDpf ;12 dDh dq12 2
¼ þ þ
f : Dpf ;12 Dh q12
Results
   2  2  2 ) 1=2

dQ12 2 dH dW dL12 The pressure loss test data measured for 10 ft (3.0 m) and
þ4 þ4 þ4 þ 5 ft (1.5 m) long phenolic ducts connected using a four-bolt
Q12 H W L12
flange and cleat assembly are summarized in Table 3 For
(17) these cases, there were no internal reinforcements present in
Values for the random measurement uncertainties consid- the sample ducts. In every instance, the Darcy friction fac-
ered in this study are presented in Table 2 The measurement tor was calculated for each flow rate by Equation 2 and
uncertainty of the test section length equaled ±0.5 in. plotted against the Reynolds number (Equation 4) on
(13 mm). It was estimated that the measurement uncertainties Moody diagrams. Figures 6–12 depict the test data obtained
of duct height and width were ±0.1 in. (2.5 mm). In some for 10 ft (3.0 m) duct sections. Similarly, Figures 13–19
instances the uncertainty for these single-sample measure- portray the results for the 5 ft (1.5 m) duct sections. The
ments was limited by the scale readability of the instrumen- Colebrook equation was used to determine relative rough-
tation, for example, temperature measurements performed ness values by the least squares method; refer to Equation 5.
416 Science and Technology for the Built Environment

Table 3. Relative and absolute roughness values for phenolic ducts (no internal reinforcements installed).
e/Dh e, mm (ft)
W  H, mm  mm Aspect ratio Dh, mm 3.0 m (10 ft) Section 1.5 m (5 ft) Section 3.0 m (10 ft) Section 1.5 m (5 ft) Section
(in.  in.) (W/H) (in.) lengths lengths lengths lengths

305  152 2 203 0.0005 0.0010 0.102 (0.00033) 0.203 (0.00067)

(12  6) (8)
305  305 1 305 0.0003 0.0006 0.092 (0.00030) 0.183 (0.00060)
(12  12) (12)
610  305 2 406 0.0002 0.0006 0.081 (0.00026) 0.244 (0.00080)
(24  12) (16)
559  279 2 373 0.0002 0.0004 0.075 (0.00025) 0.149 (0.00049)
(22  11) (14.7)
559  559 1 559 0.0003 0.0007 0.168 (0.00055) 0.391 (0.00128)
(22  22) (22)
1118  559 2 744 0.0004 0.0004 0.298 (0.00098) 0.298 (0.00098)
(44  22) (29.3)
1118  279 4 447 0.0003 0.0007 0.134 (0.00044) 0.313 (0.00103)
(44  11) (17.6)
Average roughness values 0.0003 0.0006 0.136 (0.00044) 0.254 (0.00084)

Fig. 6. Moody diagram for a 305 mm 152 mm (12 in.  6 in.) Fig. 8. Moody diagram for a 610 mm 305 mm (24 in.  12
phenolic duct with 3.05 m (10 ft) sections connected by four- in.) phenolic duct with 3.05 m (10 ft) sections connected by
bolt flanges (no internal reinforcements installed). four-bolt flanges (no internal reinforcements installed).

Fig. 7. Moody diagram for a 305 mm 305 mm (12 in.  12 Fig. 9. Moody diagram for a 559 mm 279 mm (22 in.  11
in.) phenolic duct with 3.05 m (10 ft) sections connected by in.) phenolic duct with 3.05 m (10 ft) sections connected by
four-bolt flanges (no internal reinforcements installed). four-bolt flanges (no internal reinforcements installed).
Volume 25, Number 4, April 2019 417

Fig. 10. Moody diagram for a 559 mm 559 mm (22 in.  22 Fig. 13. Moody diagram for a 305 mm 152 mm (12 in.  6
in.) phenolic duct with 3.05 m (10 ft) sections connected by in.) phenolic duct with 1.52 m (5 ft) sections connected by four-
four-bolt flanges (no internal reinforcements installed). bolt flanges (no internal reinforcements installed).

Fig. 11. Moody diagram for a 1118 mm 559 mm (44 in.  22 Fig. 14. Moody diagram for a 305 mm 305 mm (12 in.  12
in.) phenolic duct with 3.05 m (10 ft) sections connected by in.) phenolic duct with 1.52 m (5 ft) sections connected by four-
four-bolt flanges (no internal reinforcements installed). bolt flanges (no internal reinforcements installed).

Fig. 12. Moody diagram for a 1118 mm 279 mm (44 in.  11 Fig. 15. Moody diagram for a 610 mm 305 mm (24 in.  12
in.) phenolic duct with 3.05 m (10 ft) sections connected by in.) phenolic duct with 1.52 m (5 ft) sections connected by four-
four-bolt flanges (no internal reinforcements installed). bolt flanges (no internal reinforcements installed).

To a close approximation, the friction factor data for each Typically, the root-sum-square uncertainty of the test sec-
duct cross section followed a single relative roughness tion Reynolds number was approximately 2%. The uncer-
curve; these data are also summarized in Table 3. The error tainty analysis performed in this study indicated that several
bars shown for the test section Reynolds numbers and Darcy parameters provided the greatest contributions to that value.
friction factors were calculated per Equations 15 and 17, It was observed that the assumed nozzle discharge coeffi-
respectively. They are based on the estimated random meas- cient uncertainty, the measurement uncertainty of pressure
urement uncertainties provided in Table 2. loss across the nozzle chamber and test section, and the
418 Science and Technology for the Built Environment

Fig. 16. Moody diagram for a 559 mm 279 mm (22 in.  11 Fig. 19. Moody diagram for a 1118 mm 279 mm (44 in.  11
in.) phenolic duct with 1.52 m (5 ft) sections connected by four- in.) phenolic duct with 1.52 m (5 ft) sections connected by four-
bolt flanges (no internal reinforcements installed). bolt flanges (no internal reinforcements installed).

uncertainty was dominated by uncertainties associated with the

flow rate measurement, that is, the nozzle discharge coefficient
and the flow nozzle pressure loss measurement uncertainties.
Referring to Table 2, conservative estimates of the duct interior
dimensions were employed in the uncertainty analysis, because
in some instances the duct walls were not perfectly square. It
is readily apparent that the friction factor uncertainty increased
at lower flow rates, and correspondingly decreased as the air
velocity increased. This phenomenon was observed very clearly
in the Moody diagrams for the 44 in.  22 in. (1118 mm
559 mm) duct sections, where at the lowest flow rates, the
estimated friction factor root-sum-square uncertainty
approached 10%. This large uncertainty is attributed to the
Fig. 17. Moody diagram for a 559 mm 559 mm (22 in.  22 inability of the test apparatus to maintain the minimum test
in.) phenolic duct with 1.52 m (5 ft) sections connected by four- section air velocity (6 m/s) required by Section 5.4 of Standard
bolt flanges (no internal reinforcements installed).
120. Consequently, the measured test section pressure loss val-
ues were reduced, thereby yielding a higher percentage meas-
urement error.

Conclusions

For phenolic ducts with 10 ft (3.0 m) sections connected by

four-bolt flange and cleat assemblies, the relative roughness
e/Dh ranged from 0.0002 to 0.0005, with an average value
of 0.0003. By comparison, for similar duct systems having
5 ft (1.5 m) duct sections, the relative roughness range was
0.0004 to 0.0010, with an average value of 0.0006. Based
on these results, it appears that doubling the number of
joints per unit length over the test duct section generally
Fig. 18. Moody diagram for a 1118 mm 559 mm (44 in.  22 doubled the measured relative roughness. Otherwise, these
in.) phenolic duct with 1.52 m (5 ft) sections connected by four- results were apparently independent of the aspect ratio of the
bolt flanges (no internal reinforcements installed).
ducts, over the span 1  W=H  4. The corresponding abso-
measurement uncertainty of the test section cross section lute roughness e for phenolic ducts connected by four-bolt
influenced the root-sum-square uncertainty of the test section flange and cleat assemblies was 0.00044 ft (0.136 mm) for
Reynolds number almost equally. The effects of other sour- 10 ft (3.0 m) section lengths, and 0.00084 ft (0.254 mm) for
ces of uncertainty were orders of magnitude smaller for the
5 ft (1.5 m) section lengths. Their corresponding absolute
conditions encountered in the experiments.
roughness was about 1.5 to 2.8 times greater than the
For most cases considered in this test program, the root-
“average” absolute roughness (0.0003 ft, 0.09 mm)
sum-square uncertainty of the Darcy friction factor was
described for galvanized steel ducts in Table 1 of the Duct
roughly 4%. The estimation of the Darcy friction factor
Design chapter in the 2017 ASHRAE Handbook of
Volume 25, Number 4, April 2019 419

Fundamentals (ASHRAE 2017c). That roughness value is e/Dh ¼ relative roughness,

  dimensionless
based on the study of Griggs and Khodabakhsh-Sharifabad kg
q ¼ density, m3 lbm
ft3
(1992). They reported absolute roughness values of
0.00027 to 0.0005 for “galvanized steel rectangular ducts m ¼ kinematic viscosity, Pas, (lbfs/ft2)
with various types of joints, with 4 ft spacing.” However,
they did not test the effects of galvanized ducts with actual
joints. Instead the ducts were constructed from plywood, Subscripts
and wood strips placed inside the ducts were used to simu-
late joints. It is recommended that roughness data resulting 1 ¼ test section upstream
from the present study be included in the ASHRAE Duct 2 ¼ test section downstream
Fitting Database (2017b), and in future editions of the Duct 5 ¼ nozzle upstream
Design chapter in the ASHRAE Handbook of 6 ¼ nozzle downstream
Fundamentals. It is also suggested that other phenolic duct x ¼ plane
joining systems be considered for further experimental
investigation, to ascertain their influence on the relative References
and absolute roughness.
Abushakra, B, I.S. Walker, and M.H. Sherman. 2004. Compression
effects on pressure loss in flexible HVAC ducts. HVAC&R
Nomenclature Research 10(3):275–89.
ASHRAE. 2017a. Method of Testing to Determine Flow Resistance of
A ¼ duct cross-section, m2 (ft2) HVAC Ducts and Fittings. Standard 120-2017. Atlanta, GA:
Ai ¼ nozzle area, m2 ðft2 Þ ASHRAE.
ASHRAE. 2017b. Duct Fitting Database (DFDB), Version 6.00.05.
Ci ¼ nozzle coefficients, dimensionless
Atlanta, GA: ASHRAE.
Dh ¼ hydraulic diameter, m (ft) ASHRAE. 2017c. Handbook of Fundamentals. Chapter 21, Duct
f meas ¼ measured Darcy friction factor, dimensionless design. Atlanta, GA: ASHRAE.
f calc ¼ calculated Darcy friction factor, dimensionless ASHRAE. 2018. RP-1764. Determine the absolute roughness of
H ¼ height, mm (in.) phenolic duct. Final report. Atlanta, GA: ASHRAE.
L12 ¼ total test duct length, m (ft) Culp, C., and D. Cantrill. 2009. Pressure losses in 12”, 14”, and 16”
ṁ ¼ mass flow rate, kg/s (lbm/s) non-metallic flexible ducts with compression and sag. ASHRAE
Transactions 115(1):622–28.
n ¼ pressure exponent, dimensionless
Gibbs, D.C., and S. Idem. 2010. Measured and predicted pressure loss

n ¼ number of open nozzles, dimensionless in corrugated spiral duct. ASHRAE Transactions 116(2):380–86.
N ¼ number of data points, dimensionless Griggs, E.I., W.B. Swim, and G.H. Henderson. 1987. Resistance to
pb ¼ barometric pressure, kPa (in. Hg) flow of round galvanized ducts. ASHRAE Transactions 93(1):
pe ¼ saturation vapor pressure, Pa (lbf/in2) 3–16.
pp ¼ partial pressure, Pa (lbf/in2) Griggs, E.I., and F. Khodabakhsh-Sharifabad. 1992. Flow
ps ¼ static pressure, Pa (in. of water) characteristics in rectangular ducts. ASHRAE Transactions 98(1):
116–227.
pv ¼ velocity pressure, Pa (in. of water)
Heyt, J.W., and M.J. Diaz. 1975. Pressure drop in spiral flat-oval duct.
Dpf;12 ¼ duct pressure loss per unit length, Pa/m (in. of ASHRAE Transactions 81(2):221–32.
water/100 ft) Hickman, C., B.T. Beck, and B. Babin. 2015. Effect of fittings on
Dps ¼ static pressure differential between planes, Pa (in. volumetric airflow measurements (RP-1245): Single-path duct
of water) disturbances. Science and Technology for the Built Environment
Q¼ volume flow rate, L/s (ft3/min) 21(2):190–206.
R¼ ideal gas constant, kJ/(kgdaK) [ftlbf/(lbm,da R)] Huebscher, R.G. 1948. Friction equivalents for round, square and
rectangular ducts. ASHVE Transactions 54:101–18.
Re ¼ Reynolds number, dimensionless Hutchison, F.W. 1953. Friction loss in round aluminum ducts. ASHVE
tdb;o ¼ ambient dry-bulb temperature,  C ( F) Transactions 59:127–38.
twb;o ¼ ambient wet-bulb temperature,  C ( F) Jones, C.D. 1979. Friction factor and roughness of United Sheet Metal
t1 ¼ test section dry-bulb temperature,  C ( F) Company spiral duct. United Sheet Metal, Division of United
t5 ¼ nozzle chamber approach dry-bulb temperature, McGill Corp., Westerville, OH (August). Based on data in

C ( F) Friction Loss Tests, United Sheet Metal Company Spiral Duct,
V¼ average air velocity, m/s (ft/min) Ohio State University.
Khaire, S., and S. Idem. 2010. Influence of test section entrance
W¼ width, mm (in.) conditions on straight flat oval duct apparent relative roughness.
Y¼ nozzle expansion factor, dimensionless ASHRAE Transactions 116(2):371–79.
Kulkarni, D., S. Khaire, and S. Idem. 2009. Pressure loss of corrugated
spiral duct. ASHRAE Transactions 115(1):28–34.
Greek symbols Kulkarni, D., A.N. Nalla, S. Idem, and K. Gebke. 2012. Laboratory
testing of a fabric air dispersion system. ASHRAE Transactions
b ¼ nozzle diameter to chamber diameter ration, 118(2):484–90.
Kulkarni, D., J. Cui, and S. Idem. 2013. Computational fluid dynamics
dimensionless studies of developing flows with various entrance geometries.
e ¼ absolute roughness, mm (ft) HVAC&R Research 19(1):38–52.
420 Science and Technology for the Built Environment

Leverette, J., K. Gebke and S. Idem. 2014. Pressure and velocity variation Townsend, B., F. Khodabakhsh, and S. Idem. 1994. Equivalent round
in a fabric air dispersion system. HVAC&R Research 20(8):862–74. diameter of spiral flat oval ducts. ASHRAE Transactions 100(2):
SMACNA. 2015. Phenolic duct construction standards, 1st Ed. 389–95.
Chantilly, VA: Sheet Metal and Air-Conditioning Contractors Ugursal, A., and C. Culp. 2007. Comparative analysis of CFD DP vs.
National Association, Inc. [Link] measured DP for compressed flexible ducts. ASHRAE
Swim, W.B. 1978. Flow losses in rectangular ducts lined with fiber Transactions 113(1):462–69.
glass. ASHRAE Transactions 84(2):216–26. Weaver, K., and C. Culp. 2007. Static pressure losses in nonmetallic
Swim, W.B. 1982. Friction factor and roughness for airflow in plastic flexible duct. ASHRAE Transactions 113(2):400–5.
pipe. ASHRAE Transactions 88(1)269–84.