United States
                 Environmental Protection
National Risk Management
Research Laboratory
Research Triangle Park, NC 27711
                 Research and Development
EPA/600/SR-97/066     August 1997
SEPA      Project Summary

                 Phase  I  Pilot Air  Conveyance
                 System  Design,  Cleaning,  and

                 Douglas W. VanOsdell, Karin K. Foarde, and Roy Fortmann
                  Air conveyance system (ACS) clean-
                 ing, is advertised to homeowners as a
                 service having a number of benefits,
                 including the  improvement of indoor
                 air quality (IAQ).  Because ACS clean-
                 ing includes many procedures  applied
                 to many different duct systems, evalu-
                 ation has been difficult and the effec-
                 tiveness of ACS cleaning has not been
                 adequately measured.
                  The objective of this  project was to
                 develop and refine surface and airborne
                 contamination  measurement tech-
                 niques that could be used to evaluate
                 ACS cleaning. The research was in sup-
                 port of a field study to be conducted
                 later. To this end, a pilot air convey-
                 ance system (PACS)  using full-size resi-
                 dential heating and air-conditioning
                 (MAC) equipment was constructed and
                 operated to provide  a controlled, artifi-
                 cially soiled,  ACS  environment. The
                 PACS consisted of ducts, an MAC unit,
                 a dust mixing room, an instrument
                 room, and a dust generation and injec-
                 tion system. Three types of duct sys-
                 tems were evaluated with the proposed
                 measurement methods under unsoiled
                 and soiled conditions. Each duct sys-
                 tems was then cleaned by professional
                 ACS cleaners and reevaluated.
                  As a result of the pilot study, the
                 surface contamination measurement
                 methods were evaluated over a range
                 of conditions and improvements. Sur-
                 face contamination (microbial and total
                 dust) measurement methods and visual
                 inspection showed that the pilot unit
                 was  effectively cleaned by the meth-
                 ods applied during this study.  Submi-
                 cron and  larger particle counts were
                 reduced following ACS cleaning, and
                 respirable particle mass was  reduced
 for two of the three duct systems tested.
 The significance of these results in an
 actual residence was not determined.
   This Project Summary was developed
 by EPA's  National  Risk Management
 Research  Laboratory's Air Pollution
 Prevention and Control Division,  Re-
 search Triangle Park, NC, to announce
 key findings of the research project
 that is fully documented in a  separate
 report of  the  same title (see Project
 Report ordering information at back).

   The overall objectives of the Air Pollu-
 tion Prevention and Control Division's
 (APPCD's)  Air Duct Cleaning Program are
 to determine when and how to clean ven-
 tilation  systems, and evaluate  both the
 effectiveness of such  cleaning and its im-
 pact on IAQ. Little published research data
 are available to support the IAQ improve-
 ment claims sometimes  made  by ACS
 cleaning contractors,  and the data  that
 are available are difficult to  interpret. A
 two-phase  research program was under-
 taken  to develop evaluation  methods to
 achieve these objectives. This report de-
 scribes research conducted under Phase
 I of the program. The  research was a
 cooperative effort, led by RTI, with partici-
 pation by personnel from APPCD, Acurex
 Environmental,  Inc., and the  National Air
 Duct Cleaners Association (NADCA). Its
 objective was to develop and operate a
 PACS as a test bed suitable for:
    Development and testing proposed
    ACS cleaning evaluation methods,
    Comparison of IAQ instrumentation,
    intended for use in the  field, under
    controlled conditions that were  also
    as realistic as possible.

  Phase II,  a field study to be reported
separately, was  undertaken to evaluate
ACS cleaning in actual residential use.
  Though commonly referred to as "duct
cleaning," ACS cleaning properly includes
the cleaning of all air-side components of
a ventilation  system:   air handler,  heat
exchanger, humidifier,  blower, and  duct
system.  It is  sometimes  advertised  to
homeowners as a service capable of pre-
venting and possibly mitigating IAQ prob-
lems and may  improve the energy effi-
ciency of the relatively dirty systems. ACS
cleaning is a broadly defined service, with
a wide range of cleaning apparatus used
by different contractors and different parts
of the  system cleaned  using different
equipment. Many combinations of clean-
ing procedures can be used on any given
system, and there are many types of sys-

Pilot Air Conveyance System
  A PACS was constructed in a high-bay
laboratory to  allow artificial soiling  of air
duct components using a reasonable test
aerosol. As shown in Figure  1, the pilot
system  included commercially  available
components expected to accumulate dust
in varying degrees (e.g., bends,  diffusers,
registers,  grills, blowers, heat  exchang-
ers, expansions, contractions, regions of
surface irregularity, and dampers) and was
designed  to allow  application  of all  as-
pects  of the proposed evaluation methods
with the exception of evaluation  of IAQ in
residences, including pre- and post-clean-
ing inspections and evaluations.  The MAC
equipment was  all scaled for a small resi-
dence. The PACS was constructed  of the
following modules to simplify cleaning and
allow substitution of new test components:
      1) supply and return  air ducts, 2) MAC
      unit including air-conditioning coil and heat
      exchanger, 3) dust mixing room, 4) instru-
      ment  room, and  5)  dust generation sys-
      tem.  The PACS was operated in two
         normal operation  with flow into both
          rooms, and
         bypass of the instrument room during
          dust injection.
        Standard, commercially  available MAC
      equipment was used  when possible.  So
      that evaluation methods could be devel-
      oped  for the three major duct materials,
      completely separate PACS duct systems
      were constructed of the three duct materi-
      als commonly utilized  in residential  MAC:
      galvanized steel, fibrous glass duct liner
      (FDL), and fibrous glass duct board (FOB).
      A new air handler was installed for each
      duct type.
           Supply air sample locations S1-S4

Supply air
Temp, RH, and flow sensors
           , S1
                        Dust mixing room
                                                              Supply air duct
                                                                 I I I I I I  I
                                R1  .
                                              Return air duct   R2
                      Return air sample locations R1-R4
                                  Return air temp
                                  RH, and flow sensors
Figure 1.  Elevation view of PACS.

>ACS Cleaning Measurement
  The measurement parameters and sam-
pling  and analysis methods evaluated
were:  1) total surface dust  loading, 2)
culturable microbial surface loading, 3) re-
spirable aerosol concentrations (PM10 and
PM25), 4) size dependent particle number
concentration, 5) airborne fiber mass and
counts, and 6) fungal bioaerosol concen-
  The  primary  sampling method   being
evaluated (and  hence,  the  parameter of
greatest  interest during the study) was
measurement of the mass of dust on sur-
faces of the ACS components prior to and
following ACS cleaning.  Several variations
of vacuum sample  collection were  evalu-
ated, the most  promising being  the Me-
dium Volume Dust Sampler (MVDS), which
drew 10 L/min of air through  a small brush
or a custom designed slit nozzle.
  A  second important  parameter mea-
sured was the  number  of culturable mi-
crobial organisms deposited on the duct
and MAC surfaces  within a 10 cm2 tem-
plate area, as measured with a swab tech-
nique and a 10 L/min  vacuum  method.
Because of the short time between inject-
ing the dust and sampling during most of
these tests, microbial growth on the duct
was not an issue during this research.
  Airborne particle mass was measured
with  integrated  size dependent impactor
samplers at 10  L/min. Continuous  optical
particle monitors were set up in the instru-
mentation room of  the PACS, measuring
number concentration  in three ranges:
greater than 0.5 urn, greater than  5 urn,
and over 16 differential channels between
0.1 and about 10 |jm Bioaerosol samples
were also collected in the room using a
28.3  L/min slit to agar impactor. Although
the data, particularly those collected with
the  continuous  optical  particle counters,
showed changes in  particle concentrations
during the test, the reader is cautioned
not to draw conclusions  about the  impact
of ACS cleaning on airborne particle con-
centrations based on the data presented
in the report due to the limited scope of
the  measurements and  the artificial na-
ture  of  the  dust deposits  and  physical
arrangement of the room  in  which the
measurements were made.

Application of the  Test Methods
in the PACS
  A complete test series for a single duct
system consisted of a number of operat-
ing periods: 1) installation of the new duct
system and checkout; 2) pre-soiling eval-
uation of the duct and MAC unit, 3)  duct
soiling and deposited dust conditioning, 4)
post-soiling, pre-cleaning evaluation of the
duct and MAC  unit, 5) post-soiling,  pre-
cleaning air sampling, 6) cleaning the ven-
tilation system,  7)  post-cleaning  evalua-
tion  before restarting  the MAC  unit, 8)
immediate  post-cleaning  air sampling,
9) post-cleaning air sampling for total par-
ticles after cleaning, and 10) post-clean-
ing  (24-hr) sampling.

ACS Cleaning Methods
    Several procedures  used  during the
cleaning phase  of PACS operation  were
common to all three duct systems: 1) the
supply and return  grills  were  removed,
power washed,  and replaced as one of
the  final steps; 2) the interior of the MAC
unit  and the  blower were cleaned  each
time;  3) the evaporator  coil  in the  MAC
unit was inspected and cleaned each time
using a commercial coil cleaner; 4) nega-
tive system pressure was achieved  each
time by placing  large portions of the  duct
system under vacuum so that the  dust
and debris loosened and entrained by the
cleaning devices were transported  to a
large vacuum blower/filter unit capable of
drawing 1  m3/s (2000 cfm).  It remained
running as long  as debris was being gen-
  The three duct systems were  cleaned
using similar  techniques. With the  duct
system isolated  and under negative  pres-
sure, the galvanized steel  duct was
cleaned using primarily  a stiff, abrasive-
coated, cylindrical rotary  power  brush to
loosen the  dust. The  test dust  adhered
well, and multiple passes were required to
clean corners and the stiffening beads in
the  duct. Following brushing, air washing
(configured to move debris through the
system toward the  main  vacuum  source)
was used to  entrain  and transport  any
remaining dislodged dust to the collector.
  Cleaning of the fiberglass lined  and the
duct board ducts was different primarily in
that the  rotary  power brush was  con-
structed of cloth  strips to loosen  the  dust
while minimizing damage to the duct ma-
terial. Frequent  inspection was  used to
prevent  over cleaning and  consequent
damage to  the  duct.  Hand-brushing  was
used where required.

Results and  Discussion
  An overview of the sampling conducted
during this research is provided  in Table
1. The discussion below presents selected

Total Duct Dust Sampling
  The MVDS brush method worked  well
on the galvanized steel duct surface. Be-
cause there was no concern about dis-
lodging surface materials, it could  be used
aggressively to obtain  maximum collec-
tion efficiency. The  method also worked
well on the  FDL used  in this study.  Al-
though  background  mass  was  collected
from surfaces of the new  FDL  prior to
loading  of the dust  into the system, the
amount of background mass from "clean"
FDL was not substantially higher than that
collected from the surface of flexible duct
and foil  liner in the same system.  The
amount of mass  collected from "clean"
FDL surfaces was also similar to that col-
lected from galvanized steel duct surfaces,
flexible  duct surfaces,  and foil liner after
ACS cleaning.
  The  precision of the MVDS/brush sam-
pling method was generally very good for
duplicate side-by-side samples in spite of
the variability of particle deposition in the
ducts.  Duplicate samples were  obtained
in  locations similar in apparent dust load-
ing. Over all the  sampling in the three
duct types and the flexible  feeder ducts,
the standard  deviation of duplicates was
between 8 and 40%  of the mean  mass for
the pre-cleaning  samples. This level of
precision is  probably adequate  for sam-
pling duct dust from ACS components be-
cause the dust loading at different loca-
tions in an ACS  can be expected to be
highly variable.
  The NADCA sampling method was used
only to collect post-cleaning samples from
galvanized steel duct surfaces. This is cur-
rently the only application for which the
method is recommended.  It is not an effi-
cient sampling method.
  The  average mass  collected on the
cleaned galvanized steel duct surfaces was
0.26 + 0.11  g/m2,  while prior to  cleaning
the average  of all samples was 7.0 + 4.4
g/m2.  On the flexible duct surface, the
bottom-of-duct average prior to  cleaning
was 4.3 + 4.0 g/m2 while  after  cleaning
the overall average was 0.27 + 0.09 g/m2.
The mass on the cleaned foil liner of the
air handler was 0.28 g/m2  in  the galva-
nized  duct system.  Similar results were
observed in  the FDL system  where the
average  mass on surfaces after cleaning
was 0.39 + 0.08 g/m2 on FDL, 0.30 + 0.06
g/m2 on  flexible duct, and 0.24  g/m2 for
the one foil  liner sample. When efficient
sampling methods  such  as the MVDS
brush method are used, a more appropri-
ate criterion  for cleaning effectiveness is
probably residual dust of less than 0.5 g/
m2  based on the  results of these tests.
The sampling results for the duct board
system  are  inconclusive  because  all
sample contained a large fraction of fibers
that confounded  efforts to measure  re-
moval of deposited dust.

Table 1. Measurements Conducted
Dust loading
Dust loading
Dust loading
Dust loading
Microbial loading
Sampling Method
EADS - brush
EADS - nozzle
NADCA method
High volume sampler
Pipettip sampler
Mattson-Garvin slit to
agar impactor
Analysis Method
Plate counting
Plate counting
Primary method
For duct board
Un-lined galvanized only
Cooling coils only
Applied to all ducts
1-hr integrated samples
MS&T impactor/filter
 and 20 Ipm pump

MS&T impactor/filter
 and 20 Ipm pump
                                                                     24-hr integrated samples
24-hr integrated samples
Particles > 0.5|im
Particles > 5.0|im
Particle count -
16 channel
10-min averages
10-min averages
60-min averages
Filter/SKC pump
Optical (scattered light)
Optical (scattered light)
Laser aerosol
Phase contrast
Optical fiber monitor
Recorded with IAQDS
and Climet
Recorded by Climet
Direct download to
laptop computer
NIOSH 7400 method -
24-hr integrated samples
PDL-10data logger
Microbial Surface Samples
  The microbial surface samples showed
that duct cleaning significantly reduced the
microbial loading (by factors of 10 to 20).
However, the loadings were low and not
amplification sites,  so the  results cannot
be applied to microbial problem ducts. The
nozzle technique gave results comparable
to but generally lower than the  swab on
the galvanized and duct liner system, and
higher  on the duct board. Overall,  the
nozzle system was preferable for all sys-

Aerosol Measurements
  The aerosol measurements were all con-
ducted as an instrumentation and proce-
dure shakedown study, and were not in-
tended to establish the effect of ACS clean-
ing on indoor air quality. Integrated PM25
and  PM10 samples were collected in the
instrument  room during pre- and  post-
cleaning nominal 24-hr periods  for each
duct  system. All concentrations were low
(1.8 to 11.8 |j,g/m3 while the National Am-
bient Air  Quality  PM10  Annual  Primary
Standard is  50 u,g/m3).  The effect of the
             ACS cleaning was not clear-cut, given the
             small number of samples and lack of con-
             trol  over the particle content of infiltrating
             air.  Following cleaning of the  galvanized
             duct system,  the inhalable particle  mass
             was lower;   it was higher for  the fibrous
             glass  liner;  and about the same for the
             duct board system. The instrument shake-
             down that was the study's major purpose
             was successful, as  pump and timer op-
             eration and  flow stability were verified.
               The  mean  optical particle counter re-
             sults (an integrated sample)  were  simi-
             larly inconclusive with  respect to the ef-
             fect of duct  cleaning but productive as an
             equipment shakedown test. Post-cleaning
             particle counts for particles >0.5  and
             >5.0um were mostly lower following clean-
             ing, but the differences are  probably not
               Examination of the particle  counter re-
             sults as a function of time during the tests
             gave additional  information.  During  most
             of the 24 hour period, particle counts  in
             the  instrument room actually were  lower
             following cleaning of each duct. However,
             a burst of particles was emitted on  start-
                                 up  that raised  the  after-cleaning mean.
                                 This decrease occurred with both particle
                                 size ranges, and was especially clear for
                                 the >0.5 um particles. The optical particle
                                 counter results  show clearly that particle
                                 concentrations  change from day to day
                                 even in a simple system such as the PACS
                                 and that  aerosol samples  must be taken
                                 over several days to make valid compari-

                                 Airborne Fiber Measurements
                                   Fibers  were generally found to be be-
                                 low detection limits using both the  inte-
                                 grated mass  sampler and an optical de-

                                    The bioaerosol concentration was low
                                 in the instrument room for all tests. Con-
                                 centrations high enough to allow pre- and
                                 post-cleaning  comparisons  were  only
                                 reached with  the galvanized duct. In this
                                 case, the overall culturable fungal  con-
                                 centration rose following AHU  start-up
                                 (from about 20 cfu/m3 during the back-
                                 ground, dirty duct,  and  during-cleaning

samples) to 104 cfu/m3 in the hour imme-
diately following system startup. However,
over 80% of that increase was sampled in
the first  15  minutes  of the 60-minute
sample. After 45 minutes, the fungal con-
centration was down to about 2 cfu/m3.

Conclusions and
  Overall, the PACS was successful as a
test  bed  for  sampling method  develop-
ment. That  is, dust could be injected and
conditioned, and the system cleaned such
that the PACS was a reasonable labora-
tory surrogate for  a residential ACS. Op-
erated for only short periods,  as was true
of this  work, it  was not suitable  for
biocontaminant studies  because  active
growth was not present. Conclusions from
this research  are summarized below:
1.  Previously collected duct  dust can be
    dispersed into  a duct  system and con-
    ditioned at high humidity to provide a
    realistic challenge to conventional ACS
    cleaning techniques.  The dust deposit
    was clearly artificial but, in the opinion
    of experienced ACS cleaning  practi-
    tioners,  had reasonable distribution in
    the duct system and  adhesion to  the
2.  A pilot ventilation system can be used
    to investigate  some  aspects of ACS
    cleaning  under controlled conditions
    and provide results that may be appli-
    cable to field ACS cleaning. Additional
    research  is  needed to  understand all
    the parameters involved in obtaining a
    suitable ACS  dust deposit,  including
    dust injection and conditioning.
3.  The medium  volume  dust  sampler
    (MVDS), when fitted  with a brush on
    the nozzle, was shown to be suitable
    for collection of dust  from bare galva-
    nized steel, FDL,  and foil liner  sur-
    faces of ACS  components. Collection
    efficiency of the MVDS  with the brush
    was higher than  the  MVDS with a
    slotted nozzle  or the  NADCA Vacuum
    Test Method. The MVDS with brush is
    recommended  to  sample  dust mass
    deposited on  surfaces  during  the
    Phase II field study. NADCA Standard
    1992-01  should be only  used as in-
4.  Neither the MVDS  with  the  slotted
    nozzle  nor that with the  brush was
    suitable for collection of  dust from FOB.
    The brush  dislodged  a  substantial
   amount of fibrous  material from  new
   FOB, while the nozzle did not effec-
   tively remove  deposited dust on the
   fibrous  surface. Accurate measure-
   ments of  dust on  FOB surfaces can
   not be made with  the vacuum meth-
   ods  used  in this study. If FOB clean-
   ing is to be evaluated, a suitable sur-
   face sampling  method must be devel-
5.  The  dust  loading on bare galvanized
   steel duct surfaces that were cleaned
   was  less  than 0.02 g/m2 when mea-
   sured with the NADCA Vacuum Test
   Method,  meeting the NADCA Stan-
   dard  1992-01  criterion  for effective
   cleaning.  Collocated  measurements
   with  the MVDS-brush were 0.26, 0.37,
   and  0.36  g/m2 at the three  locations,
   demonstrating  the  low collection effi-
   ciency of the  NADCA Vacuum Test
6  For microbial sampling of dust depos-
   ited  on the surface of various fibrous
   glass and galvanized  metal surfaces,
   the  vacuum method  provided  more
   consistently reliable results than the
   surface swab  technique  and should
   be used in future studies. It was par-
   ticularly superior on fibrous materials.
7.  Both the  results of the post-cleaning
   dust sampling and visual inspection
   indicated  that  the  ACS components
   could be  cleaned  effectively  by the
   methods  used  in  this study.  The
   amount of dust measured on ACS
   components after cleaning was com-
   parable to those made prior to soiling
   in the PACS.
8.  The  impact of ACS cleaning  of par-
   ticle concentration indoors  remains
   unclear because infiltration and filtra-
   tion effects confounded the results.
9.  The  importance  of collecting multiple
   integrated samples and the  need to
   measure particle concentrations for ex-
   tended pre- and post-cleaning  periods
   were evident in the indoor particle data.
10. No evidence was  obtained for fiber
   emission  from the  cleaned duct  sys-
   tems, but the  scope of this research
   was  too limited  to  allow a definitive
   conclusion on  fiber emissions from
11. A brief pulse of particles was released
   when the  galvanized ACS was re-
   turned to  service following  cleaning.
   This phenomenon was detected  by
   both the bioaerosol and  optical par-
   ticle samplers.
12. While not a focus of the study, as the
   research progressed  it became ap-
   parent that ACS  construction  quality
   was an  important variable  in both
   PACS operation and the "cleanability"
   of an ACS. While poor construction
   practices did  not  interfere with  this
   study, which focused on methods de-
   velopment  and not  measurements,
   they did affect the performance of the
   ACS and the ease and thoroughness
   with which it could be cleaned.
   With regard to the duct itself, the un-
   lined galvanized duct  installed in the
   PACS  had no  apparent construction
   flaws. The butt joints  between sec-
   tions  in the  FDL  system had  been
   sprayed with  duct liner adhesive but
   were not sealed with a mastic. A small
   piece of liner near the  return air inlet
   was found to be loose when inspected
   prior to cleaning. The cut edges in the
   FOB system did not  appear to  be
   sealed and were not coated.  These
   construction details, while not in ac-
   cordance with applicable construction
   standards,  were flaws that  the duct
   cleaning  professionals  considered to
   be very common.
   In addition  to duct quality shortcom-
   ings, the air handler, though it was in
   "as received" condition, was not per-
   fectly sealed,  and coil bypass and
   leaks occurred  at  several  points.
13. The study  of biocontamination in  an
   ACS must  be conducted  over longer
   time periods  than were available to
   the present research  so that active
   microbial growth can  become estab-
   lished in the ACS. Accomplishing this
   would present some risk of exposure
   for those working in the vicinity unless
   the PACS  was redesigned  for con-
   tainment to  prevent exposure, and may
   be  impractical.  Such  studies  are
   needed,   and  use  of   smaller
   biocontamination  study apparatus is
   thus recommended.
14. Biocides, encapsulants, and sealants
   are all  used in  residential ACS clean-
   ing   in    attempts   to   control
   biocontamination  without  replacing
   duct work.  The usefulness  of these
   practices and their potential threats to
   residents have not been determined
   and should be investigated.

 Douglas I/I/ VanOsdell and Karin K. Foarde are with Research Triangle Institute,
   Research Triangle Park, NC 27709; Roy Fortmann is with Acurex Environmental
   Corporation, Research Triangle Park, NC 27709.
 Russell N. Kulp is the EPA Project Officer (see below).
 The  complete report, entitled "Phase  I Pilot Air Conveyance System Design,
   Cleaning, and Characterization, "(Order No. PB97-189682; Cost: $25.00, subject
   to  change) will be available only from:
         National Technical Information Service
         5285 Port Royal Road
         Springfield, VA 22161
         Telephone: 703-487-4650
 The EPA Project Officer can be contacted at:
         Air Pollution Prevention and Control Division
         National Risk Management Research Laboratory
         U.S. Environmental Protection Agency
         Research Triangle Park, NC 27711
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati,  OH 45268

Official Business
Penalty for Private Use $300
   PERMIT NO. G-35