Professional paper

Modelling of auxiliary ventilation systems

I. J. Duckworth and I. S. Lowndes

In subsurface excavations, auxiliary ventilation systems matter. In order to better evaluate and design auxiliary
are an important and integrated component of the overall fan and duct systems, engineering software has been
ventilation scheme. Without adequate auxiliary venti- developed (called Ductsim). This program allows the
lation, it is impossible to provide sufficient air to working simulation of auxiliary fan and duct systems. Following
faces, regardless of the quantity or quality of air in the the specification of the required design in the form of
main airways. Extended development headings are input data, the user is able to construct models and
typically ventilated using fan and duct systems, which optimise them by considering a variation in the duct
either force or exhaust air through the duct. The optimum material and diameter, the shock losses, and the number,
design of such systems can be complex, depending on type and spacing of auxiliary fans. A second program,
factors such as duct resistance, leakage and diameter, fan Ductcon, an ancillary application for Ductsim, allows
selection and spacing, and shock losses. This paper the determination of important duct leakage and
describes the development and application of a compre- resistance constants from measured survey data.
hensive design process for auxiliary ventilation systems. In addition to software development, field surveys
Included is a summary of published theory and literature, were conducted to measure and catalogue the perform-
a description of software development, information and ance of auxiliary fan and duct systems within operating
results from detailed field measurements, and a mines and tunnels. Such measurements represent a
description of the application of the software to the design critical component of the overall project objective;
of an actual extended auxiliary fan and duct system. namely to develop a comprehensive design approach for
modelling existing and proposed auxiliary ventilation
Ian J. Duckworth is at P. T. Freeport Indonesia, Tembagapura, systems. These measurements enabled the development
Papua 99930, Indonesia and Ian S. Lowndes is at the of an empirical database of duct design parameters
Nottingham Mining and Minerals Centre, School of Chemical based on true values.
Environmental and Mining Engineering (SChEME), University
of Nottingham, University Park, Nottingham NG7 2RD, UK
(Tel: +44 115 951 4081; Fax: +44 115 951 4113; e-mail:
[email protected]).
LITERATURE REVIEW
The determination of frictional pressure drop in
© 2003 IoM Communications Ltd. Published by Maney for airways may be obtained from the original Darcy-
the Institute of Materials, Minerals and Mining in association Weisbach equation or, as is more commonly used in
with AusIMM. mine ventilation, the form adapted by Atkinson. 2 The
Keywords: Auxiliary ventilation systems, fans, ducts, Ductsim, Atkinson equation can be arranged to determine
Ductcon resistance (R, Ns2/m 8) from a typical Atkinson friction
factor (k, kg/m3), and known or proposed airway area
(A, m2), perimeter (Per, m) and length (L, m).

INTRODUCTION R = kL Per3 (1)

A
Auxiliary ventilation systems are generally used to
ventilate development sections or dead-end headings. The Square Law is an important relationship that is
The air is moved from a fresh air source, normally a used to establish resistance from measured pressure (p,
trunk intake airway, to a point close to the face of the Pa) and quantity data (Q, m 3/s).
heading and back out; hence two air routes are
p
required. Typically in mining, auxiliary fan and duct R= (2)
Q2
systems are used to provide a second air route within
the primary development airway/roadway.
In modern mining and tunnelling, face ventilation Losses in ducts
problems have been exacerbated by the installation of When air passes through a duct there is a frictional
large amounts of mechanised diesel and electrical pressure loss associated with the roughness and
equipment employed to achieve increased advance geometry of the duct, and dynamic (shock) losses
rates, and longer development headings. The use of caused by bends, junctions and sudden changes in
diesel engines results in heat, gas and diesel particulate cross sectional area. The magnitude of these losses is

DOI 10.1179/037178403225001665 Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112 A105
Duckworth and Lowndes Modelling of auxiliary ventilation systems

also a factor of the density of the air in the duct. in which is presented a qualitative and quantitative
Vutukuri 18 presented a tabular list of typical values for discussion on the concept of leakage coefficient. In this
Atkinson friction factors and the resistance of duct case, the leakage coefficient is defined as the volume of
leakage paths for both rigid and flexible unsupported air in m3/s that would leak from 1 km of duct under a
ducts. An average friction factor of 0·0038 kg/m3 is uniform pressure of 100 Pa. Values provided indicate
reported in the paper. Meyer15 presented Atkinson that a new duct has a leakage coefficient of 0·03, a duct
friction factors from a series of tests conducted in installed underground in good condition has a leakage
South Africa by the former Coal Mining Division of the coefficient of less than 0·25, a leakage coefficient of 0·5 is
Chamber of Mines Research Organization on different considered poor, whereas a leakage coefficient of greater
duct materials. The results varied from 0·00252 kg/m3 for than 1·0 is considered bad. Le Roux correctly noted that
the Atkinson friction factor of a smooth bore metal these standards apply no correction for differing duct
duct, to 0·01658 kg/m3 for spiral re-inforced flexible size. It is expected that larger ducts, which have more
exhaust duct. Gillies and Wu8 presented Atkinson surface area, will result in increased leakage. Hence, sets
friction factor results for duct measurements conducted of leakage coefficients are presented that are dependent
on 100 m fabric duct lengths, which vary from 0·0021 on the duct diameter. Calizaya and Mousset-Jones4
kg/m3 for light-weight large diameter duct to 0·0041 summarised research trends in the ventilation design of
kg/m3 for heavier small diameter material. In other pub- very long tunnels and presented the results from leakage
lished literature, a major duct manufacturer (Schauen- measurements conducted on three auxiliary ventilation
burg Flexadux Corporation17) suggest that the ideal systems. The following formula (termed Woronin’s
friction factors for mine ventilation duct are 0·0022 equation) is provided to determine the leakage path
kg/m3 for steel and fibreglass, 0·0050 kg/m3 for flexible resistance from measured data:
suction and 0·0033 kg/m3 for flexible force duct. 3

r d/ 100 L (3)
100
Leakage from ducts r1/ 100 = ___________ 2
It is inevitable that some degree of leakage will be Q1
associated with ventilation ducts. The quantification 3 - 1
Q2
and prediction of this leakage is complex, and has
been the topic of significant research effort. Metcalf14 where rl/100 and rd/100 are the resistance of leakage paths
presented formulae for determining the leakage; and duct per 100 m of length respectively (Ns2/m 8).
however, the equations require the definition of The term L (m) denotes the actual length of duct over
certain variables that are hard to measure or estimate. which the quantity change (Q1 – Q2) is measured.
Gracie et al.9 described a method developed by the Gillies and Wu8 presented a comparison of air leakage
former UK National Coal Board (NCB) to evaluate prediction techniques for ventilation ducting systems.
the leakage associated with ventilation ducts. The In a detailed section on the theory of duct leakage, an
NCB method quantifies leakage using a parameter equation was provided for the leakage coefficient
called the ‘leakage coefficient’. This leakage developed by the NCB. This approximation is
coefficient (Lc) is defined as the leakage volume which basically the same as that given by Browning,3 namely:
would take place from a certain length of duct under a 3 Q1 - Q2 p1 - p 2
uniform pressure difference. Browning3 presented an Lc = 100 1000 (4)
approximate method which uses a chart to calculate 2L p 3/ 2
- p 3/ 2
1 2
resistance and leakage constants for an auxiliary duct.
The resistance constant is similar to airway resistance,
with units of Pa/100 m length at 1 m3/s flow. The where the leakage coefficient is defined by the volume
leakage constant adopts the units of m3/s/100 m of air in m3/s that would leak from a 1000 m duct
length at a pressure of 1 kPa. An approximate solu- subjected to a uniform pressure of 100 Pa.
tion is provided which assumes that the total leakage
is small compared to the inlet flow (Qout/Qin > 0·85)
and that the outlet pressure is zero (i.e. it is at the BRIEF DESCRIPTION OF DUCTSIM
discharge of the duct). Browning also developed an ALGORITHMS
empirical relationship based on the use of a computer Many different approaches exist for mathematically
program to determine inlet and delivered airflows over solving the problem of leakage from ventilation ducts.
different duct lengths and a wide range of constants. For this study, the Hardy Cross and Series Parallel
Vutukuri 18 published a summary for a study on the techniques were selected to solve a network formed by
variables in auxiliary ventilation. Considered in the the ventilation duct with associated discrete leakage
paper are the fan characteristic curve, duct diameter, paths. These approaches allow for the modelling of
standard of installation and duct length, and the more complex fan and duct configurations than those
impact that these variables have on the resistance of possible using mathematical formulae (which are
the duct and the airflow delivered to the face. The generally only valid for a certain configuration, such
author presented a tabular list of typical values for the as a forcing duct with single fan). Because of the
resistance of leakage paths for various standards of nature of a duct system, a network can be developed
installation and sizes of duct. In his book, Le Roux11 that incorporates the duct, drift, duct leakage paths
included a section entitled ‘Leakage from Metal Ducts’ and the face (in most cases the resistance of the drift

A106 Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112
 Duckworth and Lowndes Modelling of auxiliary ventilation systems

1 Network representation for a fan and duct system (adapted from Vutukuri and Lama19)

and face are insignificant when compared to the determined (termed RED). If the quantity at the fan is
resistance of the duct). Fig. 1 shows a network known, then the fan requirement (total pressure) is
representation of a leaky duct. The duct has n joints or determined from application of the Square Law (Eq. 2).
leakage paths, each with a resistance denoted by Rp Using an initial quantity, and the resistance values
(Ns2/m8). This effectively divides the duct into n+1 determined from the previous formulae, the airflow and
segments of equal length, each of which has been pressures throughout the network are evaluated directly
assigned a resistance of RD (Ns2/m8). For both the without iteration (just one initial pass through the
Hardy Cross and Series-Parallel methods it is assumed network is required to evaluate the series-parallel resist-
that: (i) the ambient pressure outside the duct is the ance data).
same at all locations along the duct; and (ii) the face In the Ductsim program, the Hardy Cross technique
and drift resistances are negligible. is used whenever fans are identified. This technique is the
most flexible, and allows accurate solution for multiple
Hardy Cross iterative solution fans, even if the fans are operating on different curves or
For the Hardy Cross solution, the following approach is are unequally spaced along a duct. The Series Parallel
adopted to solve for quantity distribution iteratively: technique is used for the specific solution termed fixed-
(i) Develop a network schematic. quantity. This tool allows the user to either determine the
(ii) Estimate an initial airflow for each branch fan total pressure and quantity for a specified airflow
ensuring that the airflows balance at each junction. requirement at the working face, or determine the face
(iii) Traverse one mesh and calculate the Hardy Cross airflow and fan total pressure for a specified fan quantity.
mesh correction factor (DQm, m3/s) from the For fixed quantity, all fans are ignored, and a single
branch resistances, airflows and fan pressures and booster fan requirement is evaluated directly.
slopes (Pfan – Pa, and Sfan – Pa/m3/s) within that
mesh, as follows:
SOFTWARE DESCRIPTION
- RQ Q - P fan (5) With the advent of personal computers and user-friendly
Qm = operating systems, the development and use of
2 R Q + S fan ventilation simulation programs has become popular to
(iv) Traverse the same mesh in the same direction assist in the design, optimisation, upgrade and main-
and correct all airflows by DQ m adhering to sign tenance of primary ventilation systems (including for
convention. tunnels and mines). However, there have been few applic-
(v) Repeat steps 3 and 4 for all meshes. ations developed that allow specialised modelling of
(vi) Repeat steps 3, 4 and 5 until the airflows balance auxiliary fan and duct systems. At an early stage, it was
at all junctions within an acceptable error. recognised that the development of new software was
required to allow for the efficient simulation of auxiliary
Series Parallel solution technique fan and duct systems. Hence, a significant component of
In examining Fig. 1, from parallel resistance theory the study was the development of two computer
the equivalent resistance of the final mesh at the face applications called Ductsim and Ductcon. It was decided
is (denoted by Rn–(n+1)): that any software design tools would be developed to the
level of professional Windows-based programs. Such
RD R p applications were to possess user-friendly visual input,
Rn - (n + 1) = 2
(6)
R p + RD output screens, and provide comprehensive support
documentation in terms of integrated help menus, user
If the next mesh is added, then the equivalent manuals and validation reports.
resistance of the last two meshes, from n–1 to n+1, The execution of the Ductsim program is governed
becomes: by the use of seven main screen views – Input View,
1 1 (7) Fan Input View, Results View, Fan Results View, Fan
= + 1 Curve View, Graph View and Fan Database View.
R(n - 1) - (n + 1) R D + Rn - (n + 1) Rp
Data entry is achieved via screen prompts displayed
Hence, by repeating this process through all meshes, through the use of dialog boxes. The physical char-
the overall equivalent resistance of the duct can be acteristics of the duct are entered in the Duct Details

Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112 A107
Duckworth and Lowndes Modelling of auxiliary ventilation systems

requirement at the working face, or determine the face

airflow and fan total pressure for a specified fan
quantity.
Following the input of all required data, the simul-
ation may be executed. The predicted duct pressure,
flow distribution and fan operational characteristics
produced by the simulation may be viewed in tabular
or graphical form. As each set of data are input to the
application, they are checked to ensure compliance
with practical operational limits. An error message
box is posted, prompting re-entry of an invalid input
character, should any data be detected to lie outside
the specified range. Presently, the program is restricted
to modelling single ducts, and not split-duct installations.
Two different verification exercises were conducted to
corroborate the Ductsim analysis techniques. The first
was to compare directly the output results against a
proven commercial ventilation network simulation
program (VnetPC 2000), and the second was to
2 Dialog box for shock loss data compare the predicted results against actual data from
duct measurements (i.e. to determine how accurately
the software simulates real-world conditions).
dialog box. This consists of four tabbed pages, which The Ductcon program has been developed to
are termed resistance, leakage, zones, and general data determine specific design parameters from measurements
input formats. The Resistance page allows the user to of actual duct installations. In particular, Ductcon
enter the duct diameter, length, and Atkinson friction determines operational estimates for the duct leakage
factor. The duct diameter and friction factor may be and resistance constants. These constants may then be
entered directly, or selected from user-defined default used as input to the Ductsim program. The main
lists. The leakage sheet features a slider allowing the equations incorporated into Ductcon are Atkinson
user to select leakage from a qualitative assessment of equation (Eq. 1), Woronin’s Equation (Eq. 3), and
the duct installation (bad to excellent). The user may Browning’s formula (Eq. 4). The program consists of
also choose to enter a numerical value for the leakage two spreadsheet-type views, one for entering duct data
resistance directly. The zones sheet allows the user to and one displaying the results. The user is required to
split the total duct length into one or more zones, each enter the duct diameter and segment length, and
of which may have different physical characteristics measured airflow and pressure data for each end of
(resistance, diameter and/or leakage). The general page the duct segment.
allows the user to enter an average air density, to select
whether the duct is forcing or exhausting air, and to
change the default run-time parameters for the FIELD MEASUREMENTS
software. During the development of the design software, field
A separate series of tabbed dialog boxes allows surveys were conducted to measure and catalogue the
information to be input to define shock losses in the duct. performance of actual auxiliary fan and duct systems
These include entry losses (sharp, flanged or formed bell), in operating mines and tunnels. Ventilation survey
exit losses (abrupt, flanged or diffuser), smooth elbow data were collected from eleven different operations.
(radius/diameter; 0·5, 0·75, 1·0, 1·5, 2·0, 2·5) and angles These data sets were classified according to type of
(45°, 90°, 135°, 180°), mitred elbow (angles, 45°, 60°, 90°), duct (rigid or flexible), and usage (face ventilation, or
dampers (degrees closed: wide open, 10°, 20°, 30°, 40°, ventilation of extended headings.)
50°, 60°), screens and fixed resistance. The shock loss
factor is determined for each type of obstruction based on Measurement techniques
ASHRAE1 fitting data. The input dialog box for shock The flow rates and pressures were measured using a
loss data is shown in Fig. 2. pitot-tube connected to a calibrated micromanometer.
The fan performance characteristic data are The pitot-static device allows the measurement of
summarised and viewed through the Fan Input View, total, static and velocity pressure. The pitot-static tube
which contains a list of the manufacturer, model, consists of two concentric tubes. When held facing
setting, distance along the duct, air density for the fan into the airflow the inner tube is subjected to the total
curve, fixed fan pressure (if applicable), and tags. pressure of the moving airstream. The outer tube is
Ductsim incorporates a data archival tool for the perforated with a ring of small holes drilled
development, manipulation and storage of fan curves. perpendicular to the direction of the airstream. The
This feature is called the fan database. Ductsim also outer tube is, therefore not, influenced by the kinetic
contains a feature called the Fixed Quantity Tool. energy of the air, and registers the static pressure only.
This tool allows the user to either determine the fan A manometer connected across the two tappings
total pressure and quantity for a specified airflow reports the velocity pressure. The flow rate at a

A108 Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112
 Duckworth and Lowndes Modelling of auxiliary ventilation systems

Table 1 Log–linear traverse positions in a circular duct (after McPherson13)

Points (n) Location of measurement point

Fractions of one diameter measured from the side of the duct

6 0·032 0·135 0·321 0·679 0·865 0·968

8 0·021 0·117 0·184 0·345 0·655 0·816 0·883 0·978
12 0·014 0·075 0·114 0·183 0·241 0·374 0·626 0·759 0·817 0·886 0·925 0·986

Table 2 Acceptance criteria – ideal and typical Atkinson #1–#11, where operations #1–#7 and #9 are all
friction factors metalliferous mines, #8 is a civil TBM excavated tunnel,
#10 is a coal mine and #11 is a salt mine. In some
Duct type Atkinson friction factor (kg/m3)
operations, multiple sets of auxiliary fan/duct systems
Manufacturer Literature were surveyed, which is indicated by the number of the
Spiral-wound steel duct 0·0022 0·0037 operation appearing in the list more than once. In some
Flexible ventilation bag 0·0033 0·0041 cases, insufficient data were available to determine
Fibreglass duct 0·0022 0·0030 leakage accurately, and the relevant section of the table
Flexible exhaust duct 0·0050 0·0150 has been left blank. All resistances and Atkinson friction
factors are referenced to standard air density.
The range of measured Atkinson friction factors for
particular cross section in a duct can be determined by steel duct vary from 0·0015–0·0045 kg/m3. The values are
measuring the local velocity pressure at a number of slightly lower in those ducts of large diameter (greater
fixed points to establish the distribution then than 1 m diameter). The average value for a straight,
integrating over the area. The velocity profile in the spiral-wound steel duct is 0·0023 kg/m3 (large) and
duct is dependent upon the Reynold’s number, relative 0·0032 kg/m3 (small). These values compare closely to
roughness, and upstream disturbances. One of the published data (see Table 2). The measurements for the
most common methods to establish the locations of blower bag (force-type fabric duct) average 0·0073 kg/m3
each fixed point traverse station is to divide the duct for larger duct and 0·0136 kg/m3 for duct less than 1 m in
into a number of equal areas and take the measure- diameter. A significant variation is observed between the
ments at the centre of each segment. The technique measured friction factors for the smaller ducting. The
used in this study to locate the points for the field average values are significantly higher than the typically
measurements was the log–linear method, which is published reference for straight duct. The results
based on the velocity profile for fully turbulent flow. emphasise the importance of good installation practice
The locations are presented in Table 1. for fabric ducts. The measurements for fibreglass duct
In order to determine duct air density, it is necessary average 0·0047 kg/m3. This value is higher than that
to measure certain psychrometric air properties. For the provided in the published literature. The data for flexible
study, these consisted of the barometric pressure, dry exhaust duct were obtained for a single duct at one
bulb temperature and relative humidity. From these operation. Due to this low sample size, it is difficult to
values, many important air properties can be evaluated, conclude whether the results are indicative of typical
including the actual air density. sections of mine exhaust duct. The results do, however,
show/illustrate that it is critical that flexible exhaust duct
Reduction of data be correctly installed and adequately tensioned.
The Ductcon program was used to determine the Table 4 presents both the computed results and a
various duct design parameters from the measured data. qualitative classification of the leakage adapted from
These included the duct area, resistance of duct/100 m, previous studies for each set of duct measurements. It
Atkinson friction factor, the resistance of the leakage can be seen that, with the exception of #7, all of the
paths/100 m and the leakage coefficient. These values duct installations are classified as ‘poor’ if the
were compared against acceptance criteria developed computed leakage path resistance is used. If the
from a review of literature. Tables 2 and 3 list the leakage coefficient is used then the majority of the
acceptance criteria used for the field measurements (all duct sections measured are still classified as ‘poor’.
related to a standard air density of 1.2 kg/m3). The duct installations surveyed at operations #7 and
#8 would be ‘excellent’, while two other steel ducts
Summary of field ventilation survey measurements would be classified as little better than ‘poor’.
The results from the field survey measurements are The conclusions from the above analysis are that
summarised in Table 4. The operations are numbered either the majority of mine duct installations are

Table 3 Acceptance criteria – duct leakage

Description Excellent Good Fair Poor

2 8 18 2 2 2
Leakage path resistance (Ns /m per 100 m duct) after Vutukuri 40 000/d 10 000/d 4444/d 2500/d2
Leakage coefficient (m3/s per 100 m at 1 kPa) after Le Roux11 0·16 d 0·32 d 0·47 d 0·63 d

d, duct diameter (m).

Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112 A109
Duckworth and Lowndes Modelling of auxiliary ventilation systems

Table 4 Summary of results from auxiliary fan and duct ventilation surveys

Operation Diameter k factor rl/100 m Classa Lc/100 m Classb

(m) (kg/m3) (Ns2/m8) 3
(m /s at 1 kPa)

Large diameter (> 1 m) – spiral wound steel duct

#1 1·52 0·0027 201 Poor 2·2 Poor
#2 1·65 0·0015 719 Poor 0·7 Fair–Good
#3 1·07 0·0031 1,561 Poor 0·7 Poor–Fair
#4 1·52 0·0019 149 Poor 2·2 Poor
#5 1·07 0·0023 14 Poor 3·9 Poor
Average 0·0023 529 … 1·9 …

Small diameter (< 1 m) – spiral wound steel duct

#2 0·61 0·0045 … … … …
#6 0·91 0·0029 146 Poor 1·4 Poor
#7 0·76 0·0023 146,275 Excellent 0·1 Excellent
Average 0·0032 … … … …

Large diameter (> 1 m) – blower bag

#8 1·30 0·0079 … … 0·2 Excellent
#8 1·30 0·0079 … … 0·1 Excellent
#9 1·07 0·0061 78 Poor 2·2 Poor
Average 0·0073 … … … …

Small diameter (< 1 m) – blower bag

#2 0·61 0·0139 1,247 Poor 0·7 Poor
#2 0·76 0·0117 … … … …
#9 0·76 0·0081 128 Poor 1·8 Poor
#3 0·91 0·0206 70 Poor 2·6 Poor
Average 0·0136 482 … 1·7 …

Fibreglass duct
#2 0·61 0·0042 161 Poor 0·5 Poor
#10 0·71 0·0051 16 Poor 5·8 Poor
Average 0·0047 … … … …

Flexible exhaust duct

#11 – tight 1·00 0·0034 206 Poor 1·2 Poor
#11 – slack 1·00 0·0103 5 Poor 9·8 Poor

a
Denotes Vutukuri18 – see Table 3. b
Adapted from Le Roux11 – see Table 3.

substandard, or the leakage classifications are too new, used once, and installed well-outside the
stringent to be applied to typical mine duct systems. It clearance envelopes of vehicles. Hence, the leakage
is suggested that both of these propositions may have can be very low. It is probably not practical, nor
merit. In general, little attention is given to the indeed necessary, to require a short-face ventilation
maintenance of ducts and the minimisation of leakage system to maintain the same standard of leakage
within short stope/face auxiliary ventilation systems. criteria as that associated with extended, high-
The only ducts determined by the study to be pressure installations. This does not excuse poor
competent installations were specialised extended installation and/or maintenance, but rather suggests
high-pressure systems, more representative of civil that separate design parameters and acceptance
tunnelling installations than mining operations. In criteria should be applied to the shorter lower pressure
such civil engineering applications, the duct is general duct systems used for stope/face ventilation.

3 Sketch showing the Homestake duct layout with measured data

A110 Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112
 Duckworth and Lowndes Modelling of auxiliary ventilation systems

4 Measured duct total pressure against duct length

APPLICATION AND VALIDATION pressures (gauge) that vary from positive to negative
To illustrate the potential use of the Ductsim software, a according to the location along the duct.
simulation exercise was conducted to model actual mine The Homestake duct was modelled using the Ductsim
auxiliary ventilation systems. This study consisted of design process. From the measured data, Atkinson
making a comparative assessment of the accuracy of the friction factor values were determined for each section of
predictions produced by the model against those of the duct, and the leakage path resistance and the leakage
actual field measurements. One of the case studies coefficient were evaluated for the duct section between
investigated was a duct installed in an extended drift at Fan 4 and Fan 3. The results obtained are shown in
the former Homestake Mine, located in the Black Hills Tables 5 and 6 (presented at the actual duct air density of
of South Dakota (US). The North Drift started from 1·16 kg/m3).
the 6800 Level (2070 m below the shaft collar) and was The Atkinson friction factor results provide an
ventilated using a 0·76 m diameter rigid exhaust duct average value of 0·0022 kg/m3. This is a low Atkinson
with 30 kW in-line vane axial fans.12 Fig. 3 illustrates the friction factor that is indicative of straight, well-
fan and duct layout for this drift. The spacing between installed rigid ducting. The leakage parameters also
each fan is indicated in addition to the airflows and fan indicate that the duct was in excellent condition with
total pressures. very little leakage. Visual observations confirmed
Fig. 4 is a graph showing the measured pressure this.12 These data were entered in the Ductsim
profile for the duct. The pressures are duct total program together with a specification that there was
both a sharp entry and abrupt exit loss to the duct line.
Table 5 Atkinson friction factor results From the simulation exercises, it was concluded that
the use of a single averaged Atkinson friction factor,
Segment k factor (kg/m3) and a single value for the leakage resistance, produced
Face – #5 0.0027 predicted airflows very close to those measured. These
#5 – #4 0·0025 results are summarised in Table 7.
#4 – #3 0·0020 With the exception of the face airflow, the relative
#3 – #2 0·0018 errors between predicted and measured values are all
#2 – #1 0·0018
within ±4%. Thus, the simulation tool gave pre-
dictions that were a good correlation to the measured
data, especially considering that the duct was not
Table 6 Duct resistance parameters
broken into zones of differing friction factor and
Segment rd/100 m rl/100 m Lc/100 m leakage. Furthermore, the face-measured airflow
(Ns2/m8) (Ns2/m8) (m3/s at 1 kPa) appears to be in error. On an exhaust duct, this value
#4 – #3 5·169 141,155 0·08
should not be higher than that associated with Fan #5
because the first section of duct is under constant

Table 7 Results from Ductsim modelling

Parameter Fan #1 Fan #2 Fan #3 Fan #4 Fan #5 Face

3
Measured airflow (m /s) 6·1 6·4 6·4 7·0 6·9 7·2
Predicted airflow (m3/s) 6·0 6·2 6·2 6·9 7·0 6·7
Error (% of measured) –1·6 –3·1 –3·1 –1·4 +1·5 –6·9

Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112 A111
Duckworth and Lowndes Modelling of auxiliary ventilation systems

5 Graphical representation of the predicted results for the Homestake duct system

suction. The pressure trend output from Ductsim may fans, one key component is to minimise the total fan
be observed in Fig. 5. This shows the Graph View for power consumption while still achieving the ventilation
Ductsim, with the total and static duct pressures design criteria (which in the case of auxiliary ventilation
plotted on the primary axis and the duct airflow is typically the face airflow). There have been many
plotted on the secondary access. The locations of the studies and publications on the subject of optimisation
duct fans are indicated on the schematic (below the of main and booster fan systems in ventilation
graphical trend). networks. One method20 uses a genetic algorithm for the
optimisation of complex ventilation networks. This
approach allows the analysis of fan pressure and fan
FUTURE DEVELOPMENTS location simultaneously. In the process suggested by
The Ductsim application and associated empirical data Yang,20 a family of possible solutions is developed, with
provide an ideal foundation for further study and each solution representing a combination of fan
software development. Three potential areas include pressures and fan locations. The strength of these
recirculation, fan optimisation and heat flow modelling/ solutions is dictated by the total airpower consumed by
climate prediction. the fans, with the stronger solutions corresponding to
Recirculation is the practice of re-using a portion of the lower total airpower. As the process continues, the
ventilating air to pass back across a working area. When stronger solutions are allowed to ‘breed’, and hence
examining recirculation theory, it is apparent that produce even stronger and more optimal solutions. It is
carefully planned systems can be effective in reducing basically a ‘survival of the fittest’ approach in which the
face gas and dust concentrations. Incorporation of solutions evolve until a final optimal solution is
recirculation in the Ductsim software would require that produced.
the program be updated such that the duct/drift network Another area of potential application is to convert the
is no longer a closed system. The network would be program to a mass-flow based simulator, incorporating
adapted to allow the inlet air quality to be specified (in heat flow algorithms. This may involve the integration of
terms of gas and dust concentrations), and the outlet an existing process, or the development of a new
conditions to be calculated. The gas and dust generation approach. Kertikov10 presented a series of procedures for
in the heading would be specified, and the program predicting the change in air temperature and humidity
updated to allow overlap duct systems to be included during the drivage of headings. The University of
(perhaps ‘face-overlap’ or ‘push-pull’ systems). Nottingham has also conducted significant research in
Fan optimisation is another potential avenue for the modelling of heat flow in dead-end headings, which
future development. In ventilation systems with multiple has led to various publications. Ross et al.16 described the

A112 Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112
 Duckworth and Lowndes Modelling of auxiliary ventilation systems

development of a computer model called DRIVE to 7. I. J. DUCKWORTH and I. S. LOWNDES: ‘The modelling of fan
predict the climate in single entry headings. Crossley and and duct systems in extended headings’, 429–435; 2001, (ed.
Lowndes5 presented future development of the DRIVE S. Wasilewski), 7th International Mine Ventilation Congress,
application, with correlation against ventilation, climatic Krakow, Poland, Research and Development Center for
Electrical Engineering and Automation in Mining, EMAG.
and operational data obtained from measurements in
8. A. D. S. GILLIES and H. W. WU: ‘A comparison of air leakage
deep British coal mines.
prediction techniques for auxiliary ventilation ducting
systems’, 1999, (ed. J. Tien) Proc. 8th US Mine Ventilation
Symposium, Rolla, MO, University of Missouri-Rolla Press.
CONCLUSIONS 9. A. GRACIE, G. HOWARD and B. JOB: ‘Analysis of leaky
The overall objective of the study was the development of auxiliary ventilation systems by microcomputer’, Min. Eng.,
an accurate, but easily applied, process to assist engineers 1981, 140, 850–851.
with the design and operation of efficient auxiliary fan and 10. V. KERTIKOV: ‘Air temperature and humidity in dead-end
duct systems. At the commencement of the study, certain headings with auxiliary ventilation’, Proc. 6th International
tasks were identified to achieve this goal. These included a Mine Ventilation Congress, SME, Littleton, CO, 1997,
detailed review of existing literature and software, 267–275.
11. W. L. LE ROUX: ‘Le Roux’s notes on mine environmental
measurement of a variety of duct systems to determine
control’, Note 4.10, CTP Book Printers, The Mine Venti-
typical design criteria, and the design of specialised, but
lation Society of South Africa, 1990, 79–84.
user-friendly, software. This comprehensive approach has 12. J. R. MARKS: Personal communication, 1998.
resulted in the development of powerful Windows-based 13. M. J. MCPHERSON: ‘Subsurface ventilation and environ-
software, empirical design data and associated mental engineering’, London, Chapman & Hall, 1993,
documentation. These tools provide engineers with the 134–140.
means to rapidly, and accurately design, or troubleshoot 14. J. R. METCALF: ‘Leakage in ventilation tubing’, Min.
complex auxiliary fan and duct systems. Magazine, 1958, 98, 274–277.
15. C. F. MEYER: ‘Determining the friction losses of underground
ventilation ducting’, J. Mine Ventilation Soc. South Afr.,
ACKNOWLEDGEMENTS 1990, October, 191–196.
16. A. J. ROSS, M. R. TUCK, M. R. STOKES and I. S. LOWNDES:
The authors would like to acknowledge the financial
‘Computer simulation of climatic conditions in rapid
support given to the project by the European Coal
development drivages’, Proc. 6th International Mine
and Steel Community under Research Project Grant Ventilation Congress, SME, Littleton, CO, 1997, 283–288.
7220-AC/857. Thanks are also given to Mr John 17. ‘Designing a mine auxiliary ventilation system’, Product
Marks, formerly of Homestake Mining Company, for literature, Schauenburg Flexadux Corporation, US, 1–12.
the provision of data and reports. 18. V. S. VUTUKURI: ‘Study of variables in auxiliary ventilation’,
The Institution of Mining and Metallurgy, 1984, January,
REFERENCES A10–A14.
1. ‘Fundamentals’, ASHRAE Handbook, I-P edn, American 19. V. S. VUTUKURI and R. D. LAMA: ‘Environmental engineering

Society of Heating, Refrigeration and Air Conditioning, mines’, Melbourne, Australia, Cambridge University Press,
1989, 32, 27–32, 52. 1986.
2. J. J. ATKINSON: ‘On the theory of the ventilation of mines’, 20. Z. Y. YANG, I. S. LOWNDES and B. DENBY: ‘Genetic algorithm

National English Institution for Mining Engineering, 1854, optimization of a large UK coal mine ventilation network’,
1–118. 625–632; 1999, (ed. J. Tien) Proc. 8th US Mine Ventilation
3. E. J. BROWNING: ‘An approximate method for auxiliary venti- Symposium, Rolla, MO, University of Missouri-Rolla Press.
lation simulations’, Min. Eng., 1983, September, 129–134.
4. F. CALIZAYA and P. MOUSSET-JONES: ‘Estimation of leakage Authors
quantity for long auxiliary ventilation systems’, Proc. 6th Ian J. Duckworth holds BEng, MSc and PhD degrees in
International Mine Ventilation Congress, SME, Littleton, Mining Engineering. He is a registered Professional Engineer
CO, 1997, 475–478. in the US and Chartered Engineer in the UK and EEC.
5. A. J. CROSSLEY and I. S. LOWNDES: ‘Climatic and thermo- Presently Dr Duckworth is employed with P. T. Freeport
dynamic modelling of raid development drivages’, 57–61; Indonesia, where he works on ventilation related issues at their
2001, (ed. S. Wasilewski), 7th International Mine Ventilation underground copper and gold mining complex in West Papua.
Congress, Krakow, Poland, Research and Development
Center for Electrical Engineering and Automation in Ian S. Lowndes is a chartered engineer and holds MSc and
Mining, EMAG. PhD degrees from the University of Nottingham. He is
6. I. J. DUCKWORTH and I. S. LOWNDES: ‘Ductsim: the currently Senior Lecturer in Mine Environmental Engineering
development of a visual duct simulation program’, 2000 within the Nottingham Mining and Minerals Centre in the
AIME-SME Annual Meeting and Exhibit, SME, School of Chemical, Environmental and Mining Engineering,
Littleton, CO, 2000. University of Nottingham, UK.

Mining Technology (Trans. Inst. Min. Metall. A) August 2003 Vol. 112 A113
Copyright of Mining Technology: Transactions of the Institute of Mining & Metallurgy, Section A is the
property of Maney Publishing and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.