United States       Office of Air Quality       EPA-450/4-85-001
            Environmental Protection  Planning and Standards      January 1985
            Agency         Research Triangle Park NC 27711

            Air
vvEPA      Dispersion of
            Airborne  Particulates
            In  Surface Coal
            Mines
            Data Analysis

-------
                              EPA-450/4-85-001
DISPERSION OF AIRBORNE
       PARTICULATES IN
   SURFACE  COAL MINES
           Data  Analysis
                    By
        TRC ENVIRONMENTAL CONSULTANTS, INC.
            Englewood, Colorado 80112
             Contract No. 68-02-3514
               EPA Project Officers:
                 J.L. Dicke
                 J.S. Touma
                 -al 1 Detection Ago
                 ^r (5PL-16)
                  ..'.-.not, Room 1670
       Chicago,
       U.S. ENVIRONMENTAL PROTECTION AGENCY
             Office of Air and Radiation
        Office of Air Quality Planning and Standards
           Research Triangle Park, NC 27711

                January 1 985

-------
                                        DISCLAIMER

This report has been reviewed by the Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, and approved for publication as received from TRC Environmental Consultants, Inc.
Approval does not signify that  the contents necessarily reflect  the views and policies of  the U.S.
Environmental Protection Agency, nor does mention of trade names for commercial products constitute
endorsement or recommendation for use. Copies of this report are available from the National Technical
Information Service, 5285 Port Royal Road, Springfield, Virginia 22161.

-------
                               TABLE OF CONTENTS

SECTION NO.                                                            PAGE NO.

  1.0     SUMMARY AND PURPOSE .....................    1

  2.0     BACKGROUND AND LITERATURE SURVEY  ..............    5
          2.1  EXPERIMENTAL STUDIES ..................    7
          2.2  SURFACE MINE MODELS  ..................    8
  3.0     DESCRIPTION OF FIELD WORK
  4.0     DATA REDUCTION  ........................    13
          4.1  DATA REDUCTION OVERVIEW  ................    13
          4.2  VCP OBSERVATIONS ....................    14
          4.3  FIELD OBSERVER LOGS  ..................    20
          4.4  METEOROLOGICAL DATA  ..................    20
          4.5  CALCULATED PARAMETERS  .................    22

  5.0     DATA ANALYSIS ........................    39
          5.1  COMPARISON OF IN-PIT VERSUS OUT-OF-PIT METEOPOLOGY ...    4C
          5.2  COMPARISON OF OBSERVED FLOW PATTERNS AND METEOROLOGICAL
               CONDITIONS .......................    45
          5.3  COMPARISON OF ESCAPE VELOCITY TO METEOPOLOGICAL
               CONDITIONS .......................    51
          5.4  COMPARISON OF ESCAPE FRACTION AND METEOROLOGY  .....    56

  6.0     SUMMARY OF FINDINGS .....................    61

  7.0     RECOMMENDATIONS FOR FUTURE WORK ...............    63

          APPENDIX 1  BIBLIOGRAPHY  ............  1  .....   A-l
                                    - 111 -

-------
1.0   SUMMARY AND PURPOSE

      During  the  summer  of  1983,  Air   Sciences,   Inc.   conducted  a  data
collection field study at  four Western surface coal mines  to characterize the
air flow within  the mine pits.   Smoke  puffs were  released at the bottom of the
pits, the  motion of the smoke puffs was  videotaped,  and  meteorological  data
were collected  both in  and  out of  the mine  pits  during  the  smoke releases.
The field study was designed to  allow visualization  of airflow within the mine
pits.

      TRC  Environmental   Consultants,   Inc.  contracted   with  FPA,  OAQPS,  to
reduce,  quantify, interpret, and analyze the  field data.   The purpose of TRC's
work, which is described in this report, was three-fold:

      •   Reduce and translate the  field data into a  computer compatible  data
          base.   This  involved  data  averaging,  computing  some meteorological
          parameters (e.g., stability classes), and data editing.
      •   Analyze  and  interpret the  field  data  to investigate relationships
          between  in-pit   and  out-of-pit   parameters,   as  well   as   other
          calculated   parameters   such   as  escape   fraction.    This   was
          accomplished  by   deriving  several   representative  parameters  (exit
          times,  net  exit   velocities,   escape  fractions),   and  analyzing
          statistical relationships.
      •   Perform a literature survey  to identify previous investigations that
          relate in-pit and out-of-pit  emissions, and  recommend future studies
          that  will  better   characterize  pit   retention.   Pesults  of  the'
          literature survey, and recommendations  for future work,  are included
          in this report.
      The  data  reduction  effort, discussed in Section 4.0,  utilized standard
National  Climatic  Data  Center  (NCDC)  and  FPA methods  and   conventions  to
compute  mean wind  directions,  wind  speeds,  and stability  classes  for  each
smoke release  episode.   Additionally,  data obtained  from the  VCR  videotapes
were  combined  with  pit geometry  data  to calculate  average smoke  puff  exit
velocities for each release episode.

      The  smoke  puff  observations  by themselves do not  provide a quantitative
measure of particle pit  retention.   However,  two  independent techniques  — one
based on a  simple  settling model,  the other based  on a  particle  deposition
model — were used in conjunction with  assumed particle size distributions and
relevant field data to infer values of escape fraction.

-------
      The  analysis  of  the   reduced   and  edited  data  base  yielded  several

findings:


      •   Computed escape  velocities   and  escape fractions  are  lowest  during
          early morning,  stable  atmospheres,  and  during  light  wind  speeds.
          This finding is in agreement with observed  flow  patterns  in the mine
          pits, as the  released smoke tracer was  frequently observed  to have
          stagnated within the mine pits during  these conditions.  Conversely,
          the greatest ventilation rates were observed during high  wind speeds
          and near neutral atmospheres.  It should be noted  that  the nature of
          the field  study  —  visual observations of  smoke puffs — precludes
          observations at  night  when  stable  atmospheres are  most  likely  to
          occur.   Hence the results of these  investigations are  biased toward
          non-stable, daytime  conditions.


      •   The computed escape  velocity was found to  be positively  correlated
          with measured  wind   speed,  temperature,   and  wind  direction,  and
          •negatively  correlated with stability category,  and  the width of  the
          mine pit.   However,  when  these parameters were  used  in  linear,
          multivariate regression analysis,  only  32%  of  the  variation  in
          escape  velocity  values  could  be accounted for.   The  linear  model
          could not be improved  through  the  use  of in-pit measurements rather
          than out-of-pit measurements, or by stratifying  the data  by mine,  by
          stability class,  or wind  speed  category.   It is concluded  that some
          processes or variables, not   accounted for in this analysis,  must  act
          in conjunction with the above meteorological parameters to determine
          the escape  velocity

      •   In-pit  winds are significantly different from out-of  pit  winds.   The
          in-pit  wind direction  differs  from  the out-of-pit wind direction  by
          about  60°.   Further,   no  correlation between  the   in-pit  versus
          out-of-pit   wind  direction   was  found  using  linear   regression
          techniques,  hence  the  in-pit  wind  direction can  not  accurately  be
          predicted from a knowledge of the out-of-pit wind direction.   In-pit
          wind speeds  are,  on  the  average,  25% smaller  than  out-of-pit  wind
          speeds.    Linear   regression   analysis  did   identify   a  significant
          positive correlation between  in-pit  and out-of-pit  wind  speeds.

      •   The value  of escape fraction  (that  portion of  the  dust   emitted  in
          the pit that leaves the pit) inferred  from  both  the  settling and  the
          deposition  models  is  greater  for  unstable and  neutral   conditions
          than for  stable  conditions,  as  shown  in Table  1.1.    This  suggests
          that  stable  atmospheres  may   suppress    vertical   motion  causing
          particulate matter  to be retained in the mine pits.   This finding is
          in  keeping  with  conventional  understanding of  atmospheric  motion,
          and agrees  with  the Winges   model of  pit  retention which  exhibits  a
          decrease in escape  fraction  with increasing  stability.

-------
                                    TABLE  1.1
                       ESCAPE FRACTION SHOWN BY STABILITY
PARTICLE SIZE                     SETTLING    DEPOSITION    WINCES    FABRICK
DISTRIBUTION      STABILITYC1)     MODEL        MODEL     EQUATION   EQUATION
UNIVERSAL


EDS


UNSTABLE
NEUTRAL
STABLE
UNSTABLE
NFUTRAL
STABLE
1.00
1.00
1.00
0.81
0.90
0.70
0.93
0.81
0.58
0.59
0.36
0.21
0.99
0.92
0.58
0.90
0.59
0.20
0.58
0.88
0.68
0.11
0.32
0.14
1. "A"  stability  class used for unstable;  "D"  used for neutral;  "F"  used for
stable.

      •  Reasonably  good agreement  is  indicated between  the  escape fractions
inferred from  the settling and  deposition models  and  those predicted  by the
equation  proposed by  Winges  when  the  data  are grouped  by  stability  class.
Relatively poor  agreement is  indicated  between escape fractions  predicted by
the  Fabrick equation  and  those  inferred from  the  settling  and  deposition
models when the data are grouped by stability class.

      •  The  magnitudes of  escape  fraction  inferred  from  the  settling model
and  from the  deposition model  differ  considerably,  as   shown  in  Table  1.1.
Because  the field data provide no  direct measure of escape fraction, it is not
possible to assess the accuracy  of  either the settling or deposition models as
means of computing escape fraction.

      •   Categorization of  smoke  release  flows into  characteristic  patterns
indicates that  the  smoke puffs  disperse  within the pit  before  exiting  during
stable,  low wind speed conditions.  This  finding is as expected.

      The literature search indicates  that  there  are  very few field  studies

that  have  looked at  pit  retention or  the  relationship between  in-pit and
                                                              •
out-of-pit concentrations.   There  are,  however, several models  that treat pit

retention and  dispersion in  pits,  ranging in  complexity from the simple Winges

and Fabrick algorithms  to  sophisticated  numerical models  that  could be used to
simulate pit dispersion.


      Because  the few  field  studies that  have attempted to  quantify  surface

mine  pit effects  to date have  only  inferred  values of escape fraction,  there

is  a  need   to measure  surface  mine pit  effects  directly.    Several  direct

measurement investigations  are proposed  in this report.   Additional effort in

deriving  and   validating   simple   escape  fraction  algorithms  that  can  be

incorporated into regulatory dispersion models  is needed.

-------
2.0   BACKGROUND AND LITERATURE SURVEY

      host air quality dispersion  models  which  are used to predict particulate
concentrations in the vicinity  of  surface mines simulate  emissions  as  if they
were  released   at   grade   level.   However,   many  different  dust-producing
operations at open pit mines occur  inside the pit,  sometimes  at depths  of many
hundreds  of  feet  below grade.   It  is  reasonable  to  suspect  that  only  a
fraction  of  the  fugitive  dust generated  at  the pit  floor  escapes  to  the
surface  where  it  then  may be  transported  to  the  mine  boundaries.   This
tendency for  particulate matter to remain inside  the pit has  been  called  pit
retention.

      There are probably two separate  mechanisms that cause particulate matter
to be retained in a mine pit.  The first  is a de-coupling  of  the wind field in
the  pit  from  the wind  field  at  the  surface,   inhibiting or  suppressing  the
vertical transport  of  particulate  from the bottom  of the pit  to  the surface.
This suppression  of vertical mixing  is obvious  to  anyone who has viewed a deep
surface mine pit on a calm  morning before sunrise.  A shroud  of diesel  exhaust
often hangs in  the  pit,  undisturbed by air movement  at the  surface.  This  pit
retention mechanism would be expected  to  be most pronounced  during stable,  low
wina speed  conditions,  such as  occur  at night.  The other mechanism by  which
particulate is  retained  is  through  deposition  and settling  at the mine  pit
surface and  along the  pit  walls.   Both  mechanisms occur simultaneously,  and
they are  linked.   When  the  air in the  pit  is   very  stable,  for  example,  the
residence time of a parcel  of air  in the pit  increases, and the deposition  and
settling processes have a longer time to act on airborne particulate.

      Pit retention is a  phenomenon  that  one  would intuitively expect to  occur
at a surface mine.  Similarly,  one would  expect that  the  presence of the mine
pit  itself  would disturb the  airflow over and inside the  pit,  so that  the
"plume" of oust might  not have  the  familiar  Gaussian distribution  imposed  by
many dispersion  models,  or might  have  a  significantly  different  trajectory
which would alter plume  location.  This altered  plume shape  or location,  while
technically different than  pit  retention, is certainly a related issue.   The
reason  that  these   surface mine  phenomena  —  pit   retention  and  plume
perturbation  —  are of  interest  is  because  most  air  quality  models  neglect
them.

-------
      Although  both  pit  retention  and  plume  perturbation  will  influence
particulate  concentrations  downwind  of  a surface  mine,  the  nature  of  the
influence is different.  Pit retention, which  removes  particulate  from  the air
before  it  leaves  the  mine  pit,  will  always  decrease   downwind   ambient
concentrations.  If an otherwise accurate model  simulation  of  a  mine  ignores
the  influence  of  pit  retention,  and  if  there  is  indeed  some  retention  of
particulate in  the  pit of  the mine,  then  the  model will  overpredict  downwind
concentrations.  The  error  will  be systematic  and  persistent.   That is,  even
in  predicting  long  term   (annual  average)  concentrations  when  the  random
emission  factor  and modeling  errors  cancel  out, the  error  due to  neglect  of
pit  retention  will still be  present  and  will produce modeled  concentrations
larger than  those  that would  actually occur.  On the  other  hand,  model errors
caused by  alteration  of the plume  shape  or  location  (plume  perturbation) due
to  the  presence  of  the  pit  could  conceivably  cause  overpredictions  or
underpredictions,  depending  upon how  the pit  is simulated.   If  the  pit  is
simulated  as  a very large  area  source, but the  dust  from the pit  exits  in a
narrow and  coherent plume,  then  the maximum concentration at the  top  edge  of
the  pit   could  be  underpredicted.    Conversely,  if  the   pit dust  completely
disperses  in  the  pit  before exiting,  a model  that  treats the pit  as a small,
discrete  area source  may  overpredict  concentrations  near the  pit.   However,
the  effect  of  poor approximation of the mine  pit  source  would be expected  to
diminish with  increasing time  and with  increasing  distance from the pit.   Over
long  time  intervals the plume of  dust  from the  pit certainly  approaches the
assumed  Gaussian  shape,  and  at   large  distances  from   the pit,  the  plume
perturbation caused by  the pit might be expected to be insignificant.

      The  remainder of  this  section  of  the  report  examines previous,  and  in
one  case  proposed,  investigations concerning pit  retention and  pit airflow.   A
distinction  is made  between  experimental studies of  pit retention  (of  which
there  are very  few),   and  models   of surface  mine  pits   (of which  there are
several).   The  discussion  in  this  section is  based on a  literature review of
meteorological  and  air quality  journals,  as well as  the  authors' discussions
with  investigators  working with  surface mine pits.

-------
2.1   EXPERIMENTAL STUDIES


      Although field studies in  the  vicinity  of surface mines have undoubtedly
been influenced by pit  retention,  very few studies have specifically addressed

pit  retention.   Furthermore,  the recent  interest  in  plume model  validation

typified  by  EPA and  EPP.I funded  investigations  (Bowne, 1982;  Lavery,  1982),

has been  confined  to more conventional stack  releases,  and  to  date there have

been no  rigorous  validations of  models applied at  surface  mines.   There  are

two reported  studies in which  the investigators detected discrepencies between
measured  and modeled   concentrations  at  surface  mines,  and  attributed  the

discrepencies to pit retention (Cole and Fabrick, 1984):


      •   WYOMING  EMISSION  FACTOR  STUDY.    In a  year-long  emission  factor
          development  study  conducted  at  two  surface  coal  mines  in Wyoming,
          Shearer,  et   al  (1981)  derived  site  specific particulate  emission
          factors.   These factors were  then  used  in  conjunction with  mine
          operational  data to  estimate the  total  particulate  emission  rate
          from  the  pits.   Independently,  a  modified  PAL  model  (modified  to
          account for  deposition over flat terrain) was used to backcalculate
          from TSP  concentrations measured at  the  edge of  the  pit  to compute
          "effective"  emission  rates  from the pits.   The   two pit  emission
          rates differed  by a  factor  of three,  and  Shearer,  et  al hypothesized
          that  only one-third  of  the  particulate  emitted  in  the pit  was
          escaping.   It  should  be   noted that   the  Wyoming  study  was  not
          designed to  isolate  the  effects  of  pit retention,  and  the  factor of
          three difference in  predicted emission rates  could have  been caused
          by phenomena  other than pit retention.

      •   BERKELEY PIT STUDY.   Cole  and Kunasz (1982)  used a  hybrid receptor
          model to  estimate  effective particulate  emissions at  Anaconda's  550
          meter  deep   copper  pit  in  Butte,  Montana.   The  receptor  model
          indicated  that  emissions  from  the  pit,  as  detected  by  hi-vol
          samplers  at  the  pit  perimeter,  were   59  grams  per  second.    A
          conventional  emissions  inventory  identified  pit  emissions  at  125
          grams per  second.    The authors hypothesized  that only  one-half  of
          the particulate matter  emitted  in  the pit  escaped  to  the surface.
          As  with  the  Wyoming   emission factor   study,   the   Berkeley  Pit
          investigation was not  designed  to look  at  pit retention,  and  it  is
          possible  that the difference in calculated  emission  factors  could
          have been caused by emission  factor  errors  or errors  in the receptor
          model.

      The only  field study that  specifically examines  pit  retention  and flow

fields at a  surface mine  appears  to be the  EPA  funded work performed  by  Air

Sciences,  Inc.  (Hittman  and  Air Sciences,  1983),  which  provided  the  data
analyzed  in  this  present  report.  The  Air Sciences  data collection  effort  is
outlined in Section 3.0 of this report.

-------
      A number of studies in air  pollution  literature  have  looked  at transport
and  dispersion over  features  similar to  surface  mine  pits,  namely,  urban
street  canyons.   The  street  canyon  is  like   a  mine   pit   in  that  both
configurations involve  the  release of pollutants  at the bottom of a  cavity,
and  both  must account  for  wind flow  circulation within the  cavity.   The
particulate retention in  a  pit is slightly more  involved than  the  dispersion
of  gases  in  a  street  canyon  because of  additional  effects  of  particulate
settling and deposition.  A  recent street canyon  study includes:  comparison of
an  extensive   date  base  of  measured  CO  concentrations  against predicted  CO
concentrations  from the  Intersection  Midblock Model   (IMM)  showing that  the
model underpredicts frequently and severely (Zamurs and Piracci,  1982).

      The  wind  tunnel offers a logical and  convenient means of  studying flow
over  cavities.   There  has  been  only one  reported  wind   tunnel   study  that
specifically  looked at street  canyons (Wedding,  et  al,  1977), and  unfortu-
nately,  that  study considered  shallow canyons,  and  did  not incorporate  the
findings into  a dispersion modeling  framework.   A very ambitious  wind tunnel
study  has   been  proposed   for  funding  by  the  Federal Highway  Administration
(FHWA)  in  1985.    The  FHWA  wind tunnel  study,   which will  consider  street
canyons  and deep  cuts,  could  have  direct applicability  to the  surface mine
pits.   The goal  of the 20  month effort will  be  to  develop new,  simplified
algorithms that  can be incorporated  into  existing Gaussian highway models to
better  predict  pollutant concentrations.   The  FHWA   study will   investigate
neutral  and  stable atmospheres,  will consider   various  wind   directions  and
surrounding  roughness  lengths,  and will consider  canyon  width-to-depth ratios
that  range from 6:1 to 1:6.

2.2    SURFACE  MINE  MODELS

       There  are  two simple  models  which  attempt  to simulate pit  retention by
deriving mass  escape fractions:

-------
      •   FABRICK  EQUATION.   Beginning  with the  helical  flow  street  canyon
          equations used  in  the APRAC-1A model,  Fabrick (1982)  derived  a  pit
          retention  equation  that  depends  on  pit  width  and  surface  wind
          speed.  The  Fabrick equation, discussed  in detail in Section 4.5.6
          of this report, predicted an escape fraction of  0.16 for  the Wyoming
          data discussed previously (Cole,  and  Fabrick,  1984).

      •   WINCES  EQUATION.   The  ERTEC Mining  Air  Quality Model  includes  a
          simple  algorithm  to  account  for  bulk  pit retention  (Winges  1981).
          The equation, which is  discussed  in  detail in Section 4.5.7 of this
          report,  is  a  function  of deposition-settling  velocity,  pit  depth,
          and vertical  diffusivity.  In  an application of  the Winges  equation
          to  the  Wyoming emission factor  study  mentioned  previously,  the
          Winges  equation predicted an  overall escape  fraction of   0.50,  as
          compared to  the 0.33 escape  fraction hypothesized from  the Wyoming
          data (Cole and Fabrick,  1984).

      In the field of  fluid  mechanics,  study of  flow over cavities  is  a mature

discipline.   Analytical  solutions  to  the  Navier-Stokes   equations  of  fluid

motion  are  available  if  a  number  of  simplifying  assumptions  are  made (two-

dimensional,  laminar  flow,   incompressible,  constant  diffusivity,  flat  plate

driven).   As the  simplifying  assumptions  are  discarded,  the  simulation  of

cavity  flow becomes  more complicated,  and  the  Navier-Stokes equations  can  be

solved  only with numerical  analysis.   A very  recent investigation of  cavity

flow was directed specifically at  surface mine  pits:
      •   HERWEHE MODEL.  The Bureau of Mines  funded  development  of  a computer
          simulation  modeling  scheme  applicable  to  shallow  surface  mines
          (Herwehe,   1984).   The   resulting   modeling   scheme  simulates  the
          transport, diffusion,  and dry  deposition  of  fugitive dust  emitted
          from an  idealized open-pit  surface  mine  through the  development  of
          two 2-dimensional finite-element models:  a planetary  boundary  layer
          model and an  advection-diffusion model.   The  boundary  layer model  is
          used  to  generate quasi-steady-state  atmospheric  flow  fields  and
          diffusion  quantities  to  be used  as  input  data  to the  advection-
          diffusion  model,  which  simulates  the ultimate fate  of  the  non-
          reactive  particulate  matter.    Synoptic   wind  conditions,  surface
          roughness, complex  nonhorizontal terrain,  atmospheric stability,  a
          variety  of  pollutant  sources,  participate   terminal  settling  and
          deposition velocities,  and  particulate  accumulation  on   the  lower
          surface are   all  factors accounted  for  in these  two  models.   The
          Herwehe model  accounts   for varying  diffusivity  with  pit  depth.
          However,  the  model is  hydrostatic   (assumes  constant  pressure  with
          pit depth), and as  a  consequence would not be  applicable  to  stable
          atmospheres  nor  to  pit  walls  with slopes  greater than  about  35
          degrees from horizontal (Herwehe,  1984 a).

-------
      The hydrostatic  assumption  in the Herwehe model  may severely  limit  its
applicability  to  surface  mine   pits.    Other  non-hydrostatic  models,   not
specifically  structured  to  simulate  surface  mines,  could  be  modified  to
acconiodate mine  pits.   Over  the  last  eight years  a series of  Finite  Element
Models (FEM) have  been developed  that  could be modified  for use  with  surface
mine pits:

      •   FEM.   The   3-dimensional Galerkin  Finite  Element Model   (FEM)  was
          developed  by  Lawrence  Livermore  Laboratory  to  simulate  complex
          terrains (Gresho, et  al, 1976).   The model has  been  applied  to flow
          over ridges,  regional geostropbic flow, and heavy gas flow.  The FEM
          is    non-hydrostatic,    transient,    and    accepts    non-constant
          diffusivities.   Finite  element modeling  can  very easily  accomodate
          terrain  features, such  as a  mine pit, and  with  much  less difficulty
          than finite  difference  models.  The major drawback to the  FEM model
          is the large computer expense  in running the model.
      While numerical  models  such as  Herwehe's  or  FEM provide scientifically
rigorous  solutions  to  specific   pit   simulations,   their  use  in  regulatory
applications is  limited  because  they  have  not been  validated with field data.
Also, a  good  deal of  effort  is needed  to  prepare the numerical models  for any
given simulation,  and  it  is not a trivial  matter to apply  the  same model  to a
pit with  an even slightly different geometry.

      A   brief   bibliography  of   references  that  examine  surface  pit  mine
modeling,  pit  retention,  street  canyons,  flow  over  cavities,  and  related
issues is included in  Appendix 1.
                                        10

-------
3.0   DESCRIPTION OF FIELD WOPK

      From June  28,  1983 through August 6,  1983,  Air Sciences,  Inc. conducted
smoke  release  measurements at  four mines  in  Colorado, Wyoming,  and Montana.
The  smoke release  field  study,  performed for  the  Industrial  Environmental
Research  Laboratory  of  EPA,  is detailed  in Studies  Related  to  Retention of
Airborne  Particulates  in Coal Mine Pits — Data  Collection Phase (Hittman and
Air Sciences, 1983), and is briefly summarized in this section.

      The  smoke  release program  was  designed  to  provide  data  concerning air
motion within  surface coal mine pits.  At each  of  the  four mines  shown in
Table  3.1 smoke  generators  at  the bottom  of  the  pits  were  used  to release
discrete  10 second puffs of diesel fuel smoke.  An  observer positioned  at the
top  of  the  pit  recorded each smoke  release on  a  video cassette recorder
(VCR).  VCR recording began with the smoke  release  from the generator, and was
terminated when  the  smoke puff left  the   pit, or when  the  smoke in  the pit
became so disperse  that it  was  no longer visible.   Roughly 8CO  such  smoke
release experiments, or  episodes, were conducted at  the four mines.
                                    TABLE  3.1
                            STUDY MINES  AND  LOCATIONS
      MINE                               LOCATION
Colorado Yampa Coal Company              Steamboat Springs, CO
Caballo, Carter Mining Company           Gillette, WY
Spring Creek Mining Company              Decker, MT
P.osebud Coal Company                     Hanna, WY
      Meteorological data  were measured both  in-  and out-  of  the  pits during
the smoke  release experiments,  and  were recorded at one-minute intervals along
with  date  and  time  on  cassette  tape.   The  meteorological  parameters  are
indicated in Table 3.2
                                       11

-------
                                   TABLE 3.2
                          METEOROLOGICAL MEASUREMENTS
      OUT OF PIT                         INSIDE PIT
      6 m.  ABOVE GROUND                  3 m.  ABOVE GROUND
      WIND SPEED                         WIND SPEED
      WIND DIRECTION                     WIND DIRECTION
      TEMPERATURE                        STD.  DEV.  OF  HORIZONTAL WIND DIRECTION
      VERTICAL WIND SPEED                TEMPERATURE
      STD. DEV. OF HORIZONTAL WIND
         DIRECTION
      STD. DEV. OF VERTICAL WIND
         SPEED
      At  the  conclusion  of each  smoke  release  episode  an observer  located
outside the pit  filled  out  a field data  log,  recording  start  time of release,
elapsed time until  plume  exit,  distance to point  of plume  exit,  direction to
exit,  and  in-pit temperature.   In  addition,   he  estimated and  recorded  cloud
cover and made a subjective assessment of smoke dispersion.


      These data — the  VCR cassettes, meteorological  data tapes,  and  field
logs — form the experimental data base that is analyzed in this report.
                                       12

-------
4.0   DATA REDUCTION

      A major task in  the  analysis  of the data collected by Air Sciences, Inc.
was  the  reduction  of  data   into  parameters,  averages,  distributions,  and
expressions which allowed  subsequent  examination  of  air flow,  meteorology, and
pit  retention.   This  section  explains which  data  were reduced,  what  methods
and  conventions were  used, and how the data  were  processed.   A brief overview
of  the  data reduction effort  precedes the discussion,  with a  description  of
VCR  observations,  meteorological  data,   and  special  calculated  parameters
following.

4.1   DATA PEDUCTION OVERVIEW

      The data reduction effort began with the quantification  of data from the
VCR  cassettes.   Video cassette recordings  of the smoke release  episodes were
analyzed to determine  smoke puff exit times,  and  to categorize the smoke flow
into  characteristic  patterns.   Additionally,  mine identification  and  episode
case  numbers  were obtained  from  the  VCR cassettes to  allow  synchronizing  of
data.

      Meteorological data  gathered  in the pit and  out of the pit were recorded
on  magnetic tape  during  the  smoke  release  episodes.   Generally,  these data
consisted of one-minute averages of meteorological parameters.

      Field logs,  containing  observations made on site, were  also recorded  on
magnetic  tape.   The  observations  included a qualitative  estimate  of  initial
dispersion, time and angle of initial  smoke puff exit, and episode case number.

      Data  from  all  three sources  (VCR  cassettes,  meteorological  data tapes,
and  field logs) were merged into a  single computer compatible  data file.  Data
were  checked  for  consistent  time  synchronization,  and anomalous  or  clearly
erroneous data were  discarded.  The  data base was  edited to  remove illegal,
spurious  characters  which  had  been introduced during  data  transcription from
cassette to magnetic tape.
                                       13

-------
      The  next step  in  the  data  reduction  effort  was  to  summarize  data
corresponding  to   discrete   smoke  release  "episodes."  Each  episode  was  an
individual smoke  release  experiment,  that  began with  the  smoke release  from
the generators at  the  bottom of the pit, and ended when  the  smoke  puff exited
the pit,  or when  the smoke  puff became  too  dispersed  to  be visible.   The
duration  of  the  smoke release episodes ranged  from  about  30  seconds  to  more
than  5 minutes,  but  the average  episode  lasted  about  two  minutes.   Time
dependent  data,   such  as  wind  speed   and  wind  direction  for example,  were
averaged  over  the duration  of the smoke  release episode.    Other  parameters,
such as mean exit  velocity  and escape  fractions, were  computed for each smoke
release  episode.    The summarized  data  were  written  on  a  magnetic  tape,
structured  so  that  each  smoke  release  episode  was  contained in one  data
record.   The data values in each smoke  release episode record are shown  in
Figure  4.1,  and  the derivation  of  each value is described  in  more  detail  in
sections  4.2 through  4.4  of  this report.   A FORTRAN  program developed  by TPC
was used  to summarize the data which  then were  the  basis for  all subsequent
statistical analyses.

4.2   VCR OBSERVATIONS

      As  a  preliminary  step  the  videotapes  were   previewed  in  order  to
facilitate extraction of data  that would be used  in the analysis.   It was  seen
from  the  previewing  of the  videotapes that some information  still existed  in
audio/visual form which would  have to  be translated to  digital  form for use  in
the analysis.  The name  of the mine,  the  episode number (called "case number"
by  Air Sciences),  and many  observations  about  weather or  operation  of the
smoke  generators,  were announced  verbally  on  the VCR  soundtrack.  Furthermore,
it  became evident that  the  behavior  of  the smoke releases  could  be  grouped
into  distinguishable patterns.   The  puff  behavior  in  the  pit appeared  from
this previewing effort to take one of  two forms  - either it  stayed  in the pit,
or  was ventilated out of  the  pit.    In  some  cases  it was not  possible  to
determine  what the behavior was.  When the puff stayed  in  the pit  it  did  so
because   it  was  recirculated  back  into  the  pit  by   an  active  circulation
pattern,  or  it simply did not  move  a  significant distance and dispersed to the
point  of  losing visual definition while still within the confines of  the pit.
                                       14

-------
                               FIGURE 4.1
                    DESCRIPTION OF VARIABLES REDUCED
                     FROM SMOKE RELEASE EXPERIMENTS
VARIABLE
MINE

JDAY

NTIME

IDNUM

IRES1

IRES2


DESCRIPTION
MINE ID number

Julian day of experiment

Time of day (hr-minute)

Experiment case number

Time in seconds of
initial puff exit
Time in seconds of
exit of entire puff

VARIABLE
WDINN


AWSOUT

AWSIN

ATOUT

ATIN
NACOV

ISGOUT
DESCRIPTION
AWDIN normalized with respect to
pit long axis.

Average out-of-pit wind speed (MPH)

Average in-pit wind speed (MPH)

Average out-of-pit temperature (F)

Average in-pit temperature (F)
Average cloud cover (tenths)

Out-of-pit stability based on
NDUR
AWDOUT
AWDIN
WDOUTN
FRAC 1
FRAC 3
FRAC 5
Internal program
counter-number of
1-minute met. observations ISGIN
required to describe
experiment
Average out-of-pit
wind direction for
duration of experiment

Average in-pit wind
direction for duration
of experiment

AWDOUT normalized with
respect to pit long axis

Escape fraction based on
VEL1 and universal size
distribution

Escape fraction based
on VEL2 and universal
size distribution

Escape fraction based
on Winges equations and
universal size distribution
IPG

IVRT

AVRT


VEL1


VEL2


FRAC 2


FRAC4


FRAC6
sigma-theta (invalid)

In-pit stability based on sigma-
theta (invalid)

Stability based on Pasquill-Gifford

Stability based on sigma-w (invalid)

Average measured vertical velocity
(invalid)

Escape velocity based on IRES1
(cm/sec)

Escape velocity based on IRES2
(cm/sec)

Escape fraction based on VELl
and EDS size distribution

Escape fraction based on VEL2
and EDS size distribution

Escape fraction based on Winges
equations and EDS size distribution
                                      15

-------
                          FIGURE 4.1 (continued)
VARIABLE         DESCRIPTION        VARIABLE             DESCRIPTION
FRAC7    Escape fraction based      FRAC8    Escape fraction based on Fabrick
         on Fabrick equations                equations and EDS size distribution
         and universal size
         distribution               PITA     Pit angle measured from North

WIDTH    Pit width (meters)         DEPTH    Pit depth (meters)

ITYPE    Observed flow pattern
         category
                                        16

-------
 When  the  puff  left  the  pit  it  appeared  to  do  so  under  three  different
 regimes.   In  some  cases  there  appeared  to  be  a  very  local  and  subtle
 circulation  pattern  caused by the  sun heating the pit  wall directly adjacent
 to the  puff  and  in turn circulating the puff up the wall and out of the pit by
 a  thermally   driven  circulation.    In  other   cases  the  ambient  wind  was
 sufficiently strong  to  advect the  puff out of the pit.  A third behavior was
 seen where the putf  simply lifted gradually from its point  of  formation as if
 by thermal buoyancy.  In  some  cases,  however,  it  was  not possible to determine
 the behavior  of  the  puff.  During some  smoke  releases  the  puff  became too
 diffuse and lost visual definition  which  made  it  impossible to determine if or
 when it left the pit.  In  other  cases  the Test  Director  erroneously terminated
 the tape  record  of the experiment  before it was  possible  to determine  if the
 puff had left the pit.  A  separate  category  to allow  for these  occurrences was
 devised.

       From these initial  observations  it  was seen that  determination  of smoke
 plume   dimensions   and  smoke  opacity  could not  be  determined  as  had  been
 anticipated.   There  were   several  reasons for  this.   The  smoke  puff was  not
 often   contained  in  a  steady-state  coherent  plume  that  could  be  assigned
 dimensions because the air motion  in  the  pit was  at  times  extremely  chaotic.
 Furthermore,  the single  camera viewing point  located at  the  top  of  the  pit
 provided only  a  2-dimensional view from  the  side,   and there were  no scale
 features in  the visual recording  with which to judge distances.   Smoke opacity
 was  also very  difficult  to judge because  of the  rapid  movement  of the smoke
 plume  and  because  of  the varied color of  the background.  Even  if it had been
 possible to  estimate  opacity, this  information  alone,  without  optical path
 length  (i.e.,   plume  width),  would  not have  been  sufficient  to  characterize
 particulate concentrations.

      In order  to  convert  the audio/visual data into  digital information that
could be analyzed, a data coding  form  was  devised.  The  form was  designed in a
way to  allow  the  analyst to note  the  puff  behavior  in numerical  code form and
thus translate  the  audio/visual data to digital characters.  Factors  noted on
the form included experiment  identification,  characteristic puff  pattern  (in
several  optional   forms),   duration  of  the  experiment   and  some   general
observations which  seemed  to  be   important  and  might be amenable  to  further
                                         17

-------
analytical  treatment.   A sample  form is  shown in  Figure  4.2.  The  numerals
appearing under  the headings  "Question"  and  "Card  Column"  are solely  for use
in keypunching the data.  The  elapsed time in  Item  5  of the form  is,  in most
instances,  the time  from initial release  of  the smoke  puff  until the  end  of
the puff  exited  the  pit.   This time was determined  by  the  VCP  analyst  using a
stopwatch.

      When  the  smoke  puff dispersed  in  the pit so  completely that it  was  no
longer  visible,  the  elapsed time  was defined  as  the  time from initial puff
release  until  the plume  could no  longer  be seen.  Defining elapsed  time  in
this manner introduces  a bias  in  the measurement since  the  parcel of  air  in
which  the smoke puff dispersed,  in some  cases,  probably remained in  the pit
even though it  could  not be seen.  However,  it was  not  possible  to  determine
how much  longer  the  puff remained  in the  pit  (because the  puff could  not  be
seen),  nor  was  it considered prudent to discard  these episodes from  the data
base  (because  this would completely  ignore  the  longest duration smoke puff
retention episodes).  The effect  of this  bias  is  to yield smaller smoke puff
retention times  than would  actually have  been measured had the smoke puff been
visible,  and for this  reason  the  elapsed  time associated with  smoke   release
episodes  that  dispersed completely within the  pit must  be considered  a  lower
bound.  Of  the  roughly  800  individual  smoke release  episodes, smoke puffs from
248 of  them dispersed completely within  the pit.   While this represents  only a
fraction  of  the  total  smoke  release  episodes,.  the   bias  is  nevertheless
important because  it  is  these  long duration retention  episodes  that  probably
account for the majority of pit retention.
            •
      The VCR  data  digitizing  effort involved  viewing  the  roughly 40 hours of
videotape while  filling out  the coding forms.   One form was completed for each
smoke  release  episode,   and  then the data from the forms  were keypunched and
merged  into the  reduced  data  base.   The   roughly 800  coding forms  have been
bound  in  a  separate volume.
                                        18

-------
                           COAL MINE VIDEOTAPE CODING FORM
                                     Figure 4.2
                                                                                  Card
                                                                       Question Column
1.   Mine Identification
        1.  Yampa
        2.  Caballo
        3.  Spring Creek
        4.  Rosebud

2.   Case Number
3.  Characteristic Pattern

        A.  Plume Stays in Pit
            1.  Recirculation evident
            2.  Disperses in pit
            3.  Other (explain) 	
        BA  Plume Exits Pit
            1.  Thermally driven up side wall
            2.  Advected out side wall
            3.  Exits center of pit
            4.  Other (explain) 	
        C.  Don't know
            1.  Sample invalid
            2.  Tape ends
            3.  Puff too diffuse
            4.  Other (explain)
4.  General Observations
 5.   Elapsed Time  (seconds)

1
2
o
3A
3B
3C
4
5


1
11
2
3
4

5
1
6

7
1
•
8
9
10
§S$XX&
11
12
13

*$
1

~7
3
1
Z-
1
x%*

If
1
1
I





$$
1
$
6
                                        19

-------
4.3   FIELD OBSERVER LOGS

      The Air  Sciences,  Inc.  field  observer logs  were  available on  magnetic
tape.  These  data included  smoke  release episode  number  (case number),  mine
name, date,  release time  of smoke puff, and duration of  puff.   In  addition,
the observer estimated the qualitative degree of dispersion of  the smoke plume
(digitized from  1  to  5),  distance that the plume was  in contact with  the side
wall  prior  to  exit,  compass direction  to  exit   location,  cloud  cover  and
ceiling,  and in-pit temperature.

      These data  were  merged with the  VCR  and  meteorological data to  make up
the digital data base.

4.4   METEOROLOGICAL DATA

      Meteorological data  collected  in  and  out  of  the mine pits  were  averaged
over the  smoke  puff  episode  in order  to yield  measurements  of  meteorological
parameters  that  influenced  each  smoke  puff experiment.   The methods  used to
average  wind  speed, wind  direction, Pasquill-Gifford  stability  class,  sigma
theta  stability  class,  and sigma  w  stability  class  are   discussed  in  the
following sections.

4.4.1 WIND SPEED AND WIND DIRECTION

      Individual one-minute  wind speed values both  in  and  out of the  mine pits
were  scalar  averaged  to   yield   mean  wind  speeds.   That  is,  the  sum  of
individual wind speeds was divided by the total  number of valid observations.

      Individual one-minute  wind  directions  both in and out  of the mine pits
were unit  vector averaged  to yield  resultant wind direction.   Resultant wind
direction, WD , was given by:

              WDr = arctangent (X/Y)
      where
              X = £sine (WD.)
              Y =£cosine  (WDp
              WD. is individual wind direction
                                          20

-------
4.4.2 PASQUILL-GIFFOFD STABILITY CLASS

      The Pasquill-Gifford stability  class was  determined  from cloud cover and
ceiling  appearing  in  the  field observer's  logs, combined  with  average  wind
speed during  the smoke  release episode.  The  procedures  used to  compute P-G
stability class were those used by  the  National Climatic Data Center (NCDC) in
deriving STAR data distributions.

4.4.3 SIGMA THETA

      Values  of  sigma  theta,  that  is,  the  standard  deviation  of horizontal
wind  direction,  were  measured  simultaneously  inside and   outside  the  mine
pits.   Air  Sciences programmed its Campbell Scientific  CR-21 data  logger to
print out  sigma  theta values  once  per minute,  with  a polling interval of 10
seconds  and  an averaging time  of  one minute (Cole,  1984).   The  sampling  time
for  the  data  logger is  equal  to  the print  out time.  Clearly this  choice of
time  parameters  was prompted  by a  desire to increase data  resolution,  and is
appropriate since some of the  smoke  releases  were visible  for only a minute or
so.   However,  the  short (one  minute) averaging  time  and  the inability  of the
data  logger  to accumulate sigma theta  values  longer  than the print  out  time
means that  some  data manipulation is required  to convert  the one minute sigma
thetas  to the  fifteen  minute sigma  thetas needed  to calculate atmospheric
stability  class  (Irwin, 1980).  The sequence  of  consecutive one-minute sigma
theta  values  corresponding   to each  smoke  puff  episode were   converted  to
representative fifteen-minute  sigma  theta values using a method  described by
Hanna (Hanna, et al,  1982).   Next,   the  fifteen-minute  sigma  theta values  were
categorized  into alphabetic  stability  classes  using  procedures  outlined  by
Irwin (1980).

      Examination of alphabetic stability  classes determined in  this  manner
revealed  that  the  vast  majority of the stabilities  were  class  "D",  both in-
and  out-of  the  pit.   The  Pasquill-Gifford  stability distribution  indicated
that only 10 percent of  the episode  hours were  class  "D",  which is a much more
reasonable distribution for summer  daylight  hours.   Given the  low confidence
in  the  accuracy and  representativeness  of  the  sigma  theta  measurements,
stability classes determined  by the sigma theta  measurements were not  used in
this study.
                                       21

-------
4.4.4 SIGMA-W

      Values of  the  standard deviation of  vertical  wind speed, sigma  w,  were
recorded by  Air  Sciences,  Inc. at  the  out-of-pit sensor.   It  was hoped  that
these measurements  could  be used  as  an indication  of  atmospheric  stability
class by relating  sigma  w to  sigma phi in  the  manner recommended by  the AMS
(AMS,1977).  However,  TRC was notified by  Air  Sciences,  Inc.  that the sigma w
data were suspect and probably in error  (Cole,1984).   The error was attributed
to  a  faulty  calibration  of  the  vertical  wind speed  instrument  that  was
discovered  by  Air  Sciences  only  after   the   data   collection   effort   was
finished.   Air  Sciences   did  not  know the  direction  nor magnitude  of  the
calibration error.   The sigma w data were not used in this study.

4.5   CALCULATED PARAMETERS

      Several  parameters  used in   this  evaluation  of  mine  pit  data  were
calculated  from  direct   measurements   of   observed   values,  from   average
meteorological conditions, from approximations of  actual  pit  geometry,  or  from
combinations of other measurements.   These calculated parameters include:
      •   Pit  dimensions.   Obviously the  geometry of  the four  mine pits  is
          somewhat  complicated,  yet  the Fabrick  and Winges  escape  fraction
          equations  require   that   the  pits   be   assigned   discrete  widths,
          lengths,  and  depths.   Hence the mine  pits  have been  "idealized"  by
          approximating their shapes with specific dimensions.

      •   Effective pit  width.   The path  length  from one edge  of the  pit  to
          the  other, defined  by  the  out-of-pit  wind direction.   This  effective
          pit width is used in Fabrick's escape fraction expression.

      •   Wind  direction and  pit  orientation.   In an effort  to discriminate
          out-of-pit wind directions with  respect  to  the  long-axis orientation
          of  the  mine  pits,  wind directions were  categorized  as  "parallel"  or
          "cross-wind" using  a calculated variable  "TACK".
                                       22

-------
      •   Smoke puff  escape  velocity.   The  pit  depth  divided  by  the  time
          required for the  smoke  puff to exit  the  pit.   This is a measure  of
          net  upward  vertical velocity.

      •   Smoke puff  escape fraction  determined  by settling.  The smoke  puff
          escape   velocity   minus   settling/deposition   velocity,    weighted
          according to  particle size  distribution.  This  escape fraction  is
          dependent upon assumed particle size distributions.

      •   Smoke puff  escape  fraction  determined  by  deposition.   The  mass
          fraction of  particulate matter that remains in  the  dust plume  at the
          point of exit from the pit as  determined  by the Van der Hoven  source
          depletion/deposition  model  (Van  der  Hoven,   1968).    This  escape
          fraction is  dependent upon assumed particle size distributions.

      •   Fabrick   escape  fraction.   The mass  fraction  of  particulate  that
          escapes   the mine  pit  as determined  by  Fabrick's equation.    This
          escape   fraction   is  dependent  upon   an  assumed  particle  size
          distribution,  wind speed, and pit width.

      •   Winges  escape  fraction.   The mass  fraction  of  particulate  that
          escapes   the mine  pit  as  determined  by Winges'  equation.  This  escape
          fraction is dependent upon  an assumed  particle size  distribution,
          pit depth,  and stability class.

Each of these calculated parameters is discussed in the following  sections.

4.5.1     PIT GEOMETRY

      In order  to assign discrete  values  of  pit  depth,  length,  and  width for
use in subsequent  computations,  it  was necessary  to  use  an idealized geometry
of  each  pit.   Idealized  pit   dimensions  are  those  identified  in   the  final
report of  the  data collection  phase  of  the  pit  retention study (Hittman and
Air  Sciences,  1983),  summarized  in Table  4.1.   Tests  were  conducted  in two
different  pits,  or   trenches,  of   the  Yampa  Mine,  so  that  two   different
geometries result.
                                       23

-------
                                   TABLE 4.1
                                  PIT GEOMETRY

1.
2.
3.
4.
5.
MINE
YAMPA
6/28-6/30
YAMPA
6/30-7/2
CABALLO
SPRING CREEK
ROSEBUD
LENGTH
(m.)
600
600
850
1130
400
WIDTH
(m.)
22
40
408
40
40
DEPTH
(m.)
19.0
20.5
33.0
45.0
40.0
ORIENTATION TO
NORTH(a)
(deg)
35
55
90
110
130
a.  Angle measured between North and long axis of rectangular pit.


      As seen from  Table  4.1,  each of the  pits  approximates  a rectangle, with

a length to width ratio ranging from 2 to 28.


4.5.2 EFFECTIVE PIT WIDTH


      The effective  pit  width, that  is,  the path length from  one  side of the
pit to  the  other (measured parallel to  the wind direction),  is  a  function of

the actual  pit  width, pit  length,  and wind direction.   If  the wind direction
is perpendicular  to  the  long axis of  the  pit  then the effective pit  width, is

equal  to  the pit length.   An  approximation to the effective  pit width,  W  is
given by:


                               W = W/sin (theta)                  Equation 1.
               Subject to      W -c L
               Where           W is actual pit width
                               L is actual pit length
                               theta is the angle between the out-of-pit wind
                               direction and the long axis of the pit.
                                      24

-------
Equation  1 is  an  approximation  that  introduces  a maximum  error  in W   of
about 10  percent  within a wind  direction band of + 3  degrees at  the Caballo
Mine.   At  all  other  mines,  and  at  all  other  wind   directions,   the  error
introduced by the approximation will be less than 6 percent.

4.5.3 WIND DIRECTION AND PIT ORIENTATION

      To   examine   the   influence   that   out-of-pit   wind  direction  and  pit
orientation have  on dependent variables,  all wind  direction/pit  orientations
were  divided  into  "parallel"  or "crosswind"  categories.  When  the  out-of-pit
wind direction was equal to the  long axis orientation of  the pit  +_ 45 degrees,
the  wind  orientation  was  deemed  parallel;  when  the  wind  direction  was
perpendicular to  the long  axis of  the pit,  + 45 degrees,  the  wind orientation
was deemed crosswind.

4.5.4      SMOKE PUFF ESCAPE VELOCITY

      For  each  smoke release  episode,  two  different (but  related)  smoke puff
exit  times were  determined  from  the VCR recordings and  the  field  logs:  the
initial  exit  time and,  where  appropriate,  the final exit  time.   Dividing the
pit depths by each of  these exit times  yields  a  measure of maximum and minimum
escape velocity for the  smoke  puff.

      One  correction was made  to the  computation of escape velocity to account
for  the  initial  plume  rise  from   the  smoke  generator.   The  pit depths were
decreased  by  three meters,   in  accordance  with  Air   Sciences'  observations
(Hittman and Air  Sciences, 1983):

      "...The  (smoke)  generators,  which  imparted  an initial horizontal exit
      velocity  of 3 meters per  second  to  the  smoke, were  generally oriented
      with smoke  exiting downwind. Although  entrainment was rapid,  under calm
      and  stable  atmospheric conditions the plume  rose  to about  3 meters above
      ground before stabilizing..."
                                       25

-------
      The exit  velocities  computed from pit  depth and exit  time  represent an
average upward velocity of  the air  in  which the smoke puff  is  dispersed.   The
snoke acts  like  a  tracer to allow  the  observer to see the motion  of  a parcel
of air in  the mine pit.   Because  the  settling  velocity  of oil smoke  is  very
small (see below),  the smoke probably works well as a tracer.

      As discussed  in section 4.2  of  this report,  smoke release  episodes in
which the  smoke puff dispersed  completely within the pit  before  exiting  were
assigned exit times equal  to  the  time  duration from puff  release until  the
puff  was no  longer visible.  This procedure  yields  artificially  small  exit
times, and artificially large exit  velocities,  for  episodes  in  which the smoke
puff dispersed within the pit so completely that it was not visible.

4.5.5 SMOKE PUFF ESCAPE FRACTION DETERMINED BY SETTLING

      The vertical  velocities  computed  by dividing the  depth  of  each  pit by
the elapsed time between  smoke  puff release  and smoke puff exit  represent an
escape velocity  for smoke  particles.   That is,  the  computed  vertical  velocity
is  characteristic   of  oil  smoke  particles,  which  are   perhaps  0.03  to  1.0
microns  in  diameter, and  have  a  maximum  gravitational  settling  velocity of
0.01 centimeters per  second (Lapple,  1961).  Real particulate  matter  found in
surface  coal  mine  pits  certainly  has  a mass mean  diameter much  larger  than
smoke  particles,  and  has  an  appreciably  larger  settling  velocity.   The
difference  in the   observed  behavior  of the  smoke puffs and the  hypothesized
behavior of actual  coal mine dust,  then, could  be attributed to  the settling,
or  deposition,   of  the  coal mine  dust.   Subtracting  the  downward  settling
velocity of coal mine dust  from  the upward escape velocity of  the  smoke puffs
will  yield  a  net vertical  velocity that  should  better  characterize real  coal
mine dust.  Furthermore, if  the  net vertical  velocity is upward  then  the  dust
particle will be expected  to escape from the  mine pit, and if the net vertical
velocity is downward then the particle will be retained in the mine pit.

      When  this reasoning  is  applied  to  a distribution of particle  sizes it
provides  a simple  means  of assessing  an overall  escape  fraction  for  the
particle size distribution  in  question.   This  is  accomplished  by  dividing the
particle  size  distribution  into  categories,   and  computing  the  settling
velocity for  each of the size categories.   Next, each settling velocity is

                                       26

-------
subtracted  from  the  smoke  puff  exit  velocity,  and  those  particle  size
categories which  exhibit a  net  upward  velocity  are  assumed  to  have  escaped
from the  pit.   Summing  the  mass  fractions of each  size  category  that  escapes
from the pit yields an overall escape fraction.

      Clearly  this   computation   depends  upon   the   initial   particle  size
distribution that  is assumed  to  characterize real  coal mine  dust.   In  this
computation, and  throughout  the  remainder of  this  investigation,  two separate
particle size distributions are adopted:

      •   UNIVERSAL  SIZE  DISTRIBUTION.   The so-called  "universal"  particle
          size distribution  is one  that was  chosen in a  previous  EPA study of
          western  surface coal  mines   (PEDCO  &  TRC,  1982)  to  represent  a
          composite,  or average,  western surface  mine.   The  universal  size
          distribution  considers  only particles  equal to,  or  smaller than, 30
          microns  in   aerodynamic  diameter.   The   universal   particle  size
          distribution  is shown  in  Table 4.2,   with  mass  fraction expressed
          cumulatively.   Table   4.2   shows   both  deposition  and  settling
          velocities,  indicating  that  deposition  velocity is   greater  than
          settling  velocity  for  all  particle  sizes  in  the  universal particle
          size distribution.
      •   EDS  SIZE  DISTRIBUTION.   As   part  of  a  privately  funded  Emission
          Development  Study  (EDS)  for  surface coal  mines in  the  Powder River
          Basin,   particle   size   distribution  was  determined   from  optical
          microscopic   examination  of   millipore   filters   (Shearer,   et  al,
          1981).   The  EDS study considered  all  particles in the  visible size
          spectrum,  from about 2 microns physical  diameter  to  greater than 130
          microns.   The  EDS  particle size distribution  is  shown  in  Table 4.3,
          with mass  fraction expressed  cumulatively.
      For  each of  the  particle size distributions,  equivalent  deposition and
settling  velocities  were  estimated.   Deposition  velocities  associated  with
each  size  category were taken  from curves presented by  Hanna,  et al,  (1982),
with  assumed  surface  roughness length  (z )  of  1.0 centimeter,  and  particle
density  of  1.0 grams/cm ,   These  values are approximate  choices  of roughness
length and  particle density that could  be expected to occur  at  surface mines
(see  Hogstrom,  1978).   Settling  velocities  were  calculated  from  Stokes'
equation, with appropriate corrections  for  particle density  and  shape  factor
(Hanna,  et  al,  1982).   These  deposition velocities and settling velocities are
shown  in   Tables   4.2  and  4.3  for   the  universal  and  EDS  particle  size
distributions.   In  each  case  it  is   assumed  that  the  larger   of  the  two
velocities,  deposition velocity  or  gravitational  settling  velocity,   is  the
dominant  removal  mechanism for any given  particle size  category.   For every
                                    27

-------
«

















w
^
M
H
^J
^
t^
S
5
U

1
1

2
O
tH
H

PQ
hH
Oi
§— (
CO
(— t
Q

W
CO

w
K-i
U
rH
H
S>
S

5j
CO
PS
U
rH
2
S





























O
(U
U)
2 E
M y
_i
H «
H 2
W O
CO 1 1

O CO
O
Cfii P-l
jjj pTj
O Q
r-4 2
•o
2 CJ
O 




^-s
o

""o •
IS O
M ty
(J CO
EH **^.
£— i g
W O
CO
#1
CO ,-J
o
H
CO





j"**s Jj2
4J o
^^ ^^
W H
> 0
M <|
EH OS
3 CO
§ °5
CJ S




CO

M
O •
< E
K U

2
•<3
W
|g




/^-\
CO
x»x W
PS C
W O
H -i
S -H
M S
°







CO O O O O 0
CN r~- co O ro ro
O O rH CN CN CO








fO O O O O O
CM r^ co o ro ro
• •••••
O O rH CN CN m
















vO "^ to -^ O **C
oo sr r>. en o r~
rH f^ CT\ ^O O r^
f~*i ^^ CN ^O O'^ vO
• •••••
O O O O rH CM












rH -^3 O OO CO O
CN CTS r^ rH CO O
O O CN ST LO O
rH




*^-
I -^  c. •
U G.
O
•~~- CN
O rH O
• *O rH
• rH
II O, ra 60
>> - w fe<
•U CN IU
•H CO
M ON crj CN
C rH C CO
01 C 0>
•a « to I-*
^ 0 SB
0"! ^
EH » iH

•H "O ^
Q C rJ U
CTJ 0)
O W
•H O  CC
>, OH CO S
•o
0 B E B
S-i O O O
0) &-l ^J i-i


CD J3 o T3
                                28

-------





















w
HH
H
CJ
1
1
f^
O
1 1

^ s
— 4 CO
H ^->
E-' E
U. O
CO
PT, (X(
O W
c
fc a
O fe
PH <
rJ
U
^-v (1)
•c to
*7^ E
O 0
rH
M H
CO M
o t->
PL, O
W rJ
Q (zj
^^
01
x-s o
U 0)
a "*-
2 B
M O
hJ

H H
U rH
m C }
VJ *J
o
CO rJ
CD ^*
H
CO



43~ 2
--^ O
M
W H
> U
gg
3r^
UJ
S co
3 <






•»
CO
3
Q B
^x
BH CO
U C
H O

s u
M B
Q




OOO^OCNt^OrH
o o o *>o o sr o *>c
OOOO^Ji— it~-CN
srwio^ot^vci-isr
i — icNsrmoocNf^cN
iH rH CN










O O O O
sr m o o
rH CM ST "~l









sri^r^vOCNI^OrH
O rH CO **O O ST O ^O
inoinovDi-Hr^cN
Oi-tcnmoocMr^cM
rH rH CM










r^** ^3 ***r co ^o c^ "^" fO
i— i\£>rO^HCNCMCMCM
OOrHCNCO-sTiO^O








*^ c*^ ro oo OT co fO co
i i i i i i i i
oooooooo

xxxxxxxx

ooinoinoino
u-li-HrHCMCMOOO-lsr













OOOOOOOO
rH CN oo sr m vo r**« oo






O oo sr r^ CTV
oo o CT. oo CN
00 sr ON u-i i— l
oo O oo sr CN
oo in CN o cr»
CM co sr uo m





















O co sr r-. &
ro O cr> oo CM
oo sr os in I-H
oo o oo sr CM
CO HO CN O G*»
CM oo sr "o m










O *^r ^ oo CT*
r-H O CTi H3 CT»
r^ oo oo c3^ c^








oo oo oo oo oo
1 1 1 1 i
O O O O O
rH rH rH rH rH
X X X X X

m o u~i O u")
NJ* in in vo ^o













0 O O 0 O
O1 O rH CN 00
^H ^H ^H ^~i





























CJ
O)
CO
•v^
M
i-H
in

II
±j
•rH
U
C
01 CN
•u oo
B rH
« CJ
r-. i-i «
VO rH
• II CO
a, o
N 4J
o * cu
CJ CN «
— oo o -
oo a> r-> co
l a r^
i" "1 * VH
o a c
. oo co
rH 00 rH « JE
• CO rH
II CL CO 4-!
4-1 O
>, » 01 4J
4-1 r-H 
CO O"> CO M •
C rH C CO Q,
QJ C C
t3 * CO C ••
rH BB CO CO
cO X CM
CU 4-1 ^ * rH
4J 0) CO ST II
O* r-5 ' 3W
g O O
Cfl (-1 CO i-H
•H 41 -
Q IH 01 0)
CO ^ rH T3
rH CU O 43 01
CO 43 4J CO 4_)
O CO CO H O
•H qj
CO E S B !-•
>, O O O U
43 M >-i U O
PL, fit PL, fc, U


X"^ ^"N X-N /^\ ^~S
CO 43 O T3 01
29

-------
particle  size  category  in   the   0-30   micron   universal  distribution,   the
deposition velocity exceeds  the  settling velocity (see Table  4.2),  whereas in
the  EDS  distribution,  which  includes  a much  larger range of  particle  sizes,
gravitational settling is the  dominant means of  particle  removal  for particles
larger than about 40 microns  physical  diameter.   This  conclusion  is consistent
with  experimental  data  which suggest  that  for particles   smaller  than  40
microns  in  size,  deposition  dominates,  whereas  for  particles larger than 40
microns, gravitational settling dominates (Hanna, et  al,  1982).

      Both Tables  4.2  and 4.3 express  removal velocity  (either  deposition or
settling) and cumulative  mass fraction as a function  of  particle  size,  so it
is a  simple  matter  to plot mass fraction  versus removal  velocity.   Figure 4.3
(applicable to  the  universal  distribution),  and  Figure 4.4  (applicable  to the
EBS  distribution)  show   the   graphs  of  particle mass   fraction  and  removal
velocity.  In Figure  4.3  curves of mass fraction versus  removal  velocity are
shown for both  particle  settling and particle deposition merely  to illustrate
that  deposition velocities  exceed  settling  velocities  for all particle  sizes
in  the  universal  particle  size  distribution.    Each  of   these  graphs,  Figures
4.3  and  4.4,  indicates what  fraction  of the  total  mass  of  particles  exhibit
removal  velocities  smaller  than  specified  values.   For example,  Figure  4.3
shows that,  for the universal particle  size  distribution,  86 percent  of  the
total mass  of  particles  have  a  removal  velocity less  than 3.0 cm/sec.   If the
escape  velocity  at a  mine  pit  were  exactly  3.0   cm/sec,   then   Figure  4.3
suggests  that  86  percent  of  the  total  mass  of  the  universal  particle  size
distribution would  escape.   Figure 4.4  shows  that,  for  the  EDS  particle  size
distribution,  13   percent  of the  total  mass  fraction  of  particles  have  a
removal  velocity  less  than 3.0 cm/sec.   If  the  escape  velocity at  a mine pit
were  exactly 3.0  cm/sec,  then Figure  4.4 suggests that 13 percent of the total
mass  of  the  EDS particle  size distribution would  escape.   Figures  4.3  and 4.4
can  be used to infer escape fractions in the following manner:
                                        30

-------
MASS
FRACTION1
    i.o ;
    0.9
    0.8
    0.6
    0.5
    0.4
    0.3
    o.:
    0.1
                                FIGURE 4o3
                 REMOVAL VELOCITY VERSUS MASS FRACTION
                           UNIVERSAL DISTRIBUTION
            -Vf
                       settling
                          ±
                          £
                                                                           -r-
                                                           deposition
                  0.5
1.0        1.5       2.0

   REMOVAL VELOCITY,  cm/sec.

          31
2.5
3.0

-------
                                        FIGURE 4o4
                        REMOVAL VELOCITY VERSUS MASS FRACTION
                               EDS PARTICLE DISTRIBUTION
MASS
FRACTION
       1.0
       0.9
       0.8
       0.7
       0.6
       0.5   -
       0.4
       0.3
       0.2
       0.1
                     10       20      30       40

                                 REMOVAL  VELOCITY
   50     60

,  cm/sec.
                                            32

-------
      1.   Choose a particle size distribution,  either  universal  (0-30 microns)
      or  EDS (0-130 microns),  of interest.
      2.   Calculate escape velocity for a given smoke  puff  episode by dividing
      pit depth by smoke puff  exit  time,  as  discussed in section  4.5.4.
      3.   Enter Figure  4.3 for universal distribution,  or Figure  4.4 for EDS
      distribution, with escape velocity along  the  horizontal  axis.   Find mass
      fraction along  the  vertical  axis of  the  Figures.  The  mass fraction is
      the amount of particulate that escapes.
      In  practice it would have been  too  time consuming to enter the graphs in
Figures  4.3 and  4.4  manually for  each  of  the  roughly  800  smoke  release
episodes.  To  expedite  the computation of  escape fraction, the curves in the
Figures  were  approximated  with  analytical  expressions  (curve   fits),  and
computer  programmed to  calculate  two  escape  fractions —  one  corresponding to
the  universal  size  distribution   and  the   other  corresponding  to  the  EDS
distribution  —  for  each  of  the 800  smoke  release  episodes.   The  escape
fractions were  stored  on magnetic  tape  for  subsequent  analysis.   The results
of this analysis are discussed in Section 5 of  this report.

      It  is important to understand that  the computation of escape fraction in
the manner described in this  subsection is  an oversimplification of the actual
pit retention phenomenon.  The  computation  of escape fraction  outlined in this
subsection  balances upward  escape velocity,  determined  from the  smoke puff
behavior, against  downward deposition and settling  velocities,  without regard
to  the exact details  of  smoke plume  trajectory or  plume-ground  interaction.
It is  reasonable  to expect that this  simplification may  tend to  overestimate
the  true escape fraction, because the actual  deposition process  is  strongly
dependent upon  the plume-terrain  interaction when  the plume  is very  close to
the  pit  floor  and the  pit  walls.   It  may  be that  some  of the particulate
matter released in the  pit of a surface  mine is removed by deposition at the
pit  floor  even when  the  net escape  velocity (the upward  exit  velocity minus
the  downward  deposition  or   settling  velocity)   is  directed  upwards.   The
magnitude of  the   overestimation of the escape fraction will be  greatest for
small  particles  whose deposition  velocities are  consistently lower  than the
upward exit velocity.   Furthermore, as has been explained in section 4.5.4,
                                       33

-------
there is  a  bias in  the  computation of escape  velocity  for some of  the smoke
release  episodes  which  yields   artificially   large  exit   velocities,   and
correspondingly large escape fractions.

      Because of the  uncertainty  and  known biases in this  method of  inferring
escape fraction from the field data, a  second,  independent  means  of estimating
escape fraction was  adopted.   This second  means  of computing  escape fraction
is discussed in the next subsection.

4.5.6 SMOKE PUFF ESCAPE FRACTION DETERMINED BY DEPOSITION

      The previously discussed method  of  inferring  escape fraction balances
upward escape  velocity against downward  particle  removal velocity,  and  in so
doing  ignores   the   plume-terrain  interaction   which  influences   particle
deposition.   A  simple  model that  accounts for  both settling and  deposition is
the so-called source depletion model (Van der Hoven, 1968),  which is given by

                                     dx          1     -(2/7T)1/2(Vd/u)
                                                                     Equation 2

where
                        [xf           dx          1
                  exp     	  I
                        i  Oz  exp (h2/2oi)        J
    Q /Q      is the ratio of apparent emission rate at distance  x,  divided by
     X  O
              true emission rate at  the source.
      (j       is the standard deviation of vertical  concentration,  m.
      h       is  the   separation  distance  between  the  dust  plume  and  the
              ground,  m.
      V       is the larger of deposition or settling velocity,  m/s.
      u       is wind speed,  m/s

The value of Q /Q  is, of course,  equal to escape  fraction.
              X  O

      A  simple  FORTRAN computer  program was written  to  solve  Equation 2  by
stepwise   numerical  integration   for   both   the   universal   and   the   EDS
distributions.  The forward  step  size  used  in the integration was set  to five
meters,  and  the plume-terrain separation distance, h,  was arbitrarily  set  to
                                      34

-------
one meter  to simulate  ground  level sources  within the mine  pit.   Values  of
wind speed, effective pit width, and stability  class  determined  from the field
data were  substituted  into equation 2.   The  wind  speed  for -each  episode  was
set equal  to the horizontal wind speed  measured inside the pit,  and the values
of sigma-z  were  computed using the  Martin (1976)  curve  fits to the  familiar
Turner  dispersion  coefficients (Turner,  1970).   The  limit  of integration,  x,
was set equal to the effective pit  width defined by the in-pit wind  direction.

      Equation 2 was  solved  for each particle  size category (weighted by mass
fraction)  for  the  universal  and   the  EDS particle  size  distributions.   This
computation was  repeated for each of  the roughly  800  smoke release episodes,
and the resulting escape fractions were  stored  on  magnetic tape  for subsequent
analysis.   The  results  of  this analysis  are discussed in  Section 5  of  this
report.

      There  are  several  potential  sources  of  error  inherent in  the use  of
Equation  2 to infer  escape fractions.   The most  important of  these is  the
specification of deposition  velocity,  which  can vary by an  order of magnitude
for small  particles depending  upon  ground cover,  roughness height, and  other
parameters  (Hanna,  et  al,  1982).   Additionally,   the  source depletion  model
imposes  a  Gaussian  distribution  in  the  vertical,  and  assumes  that  the
particulate concentration  is depleted  uniformly throughout the entire  vertical
extent  of  the plume.   The effect  that  these  assumptions  have  on  predicted
surface mine pit escape fractions  is not known.

4.5.7 FABRICK ESCAPE FRACTION

      Fabrick (1982) derived a  mine  pit escape  fraction equation that depends
upon  the  width  of  the  pit,  the   wind  speed  at  the  top of  the  pit,  and  a
particle size distribution:
                       = 1-Vd[C/u(i+ln4 )]            Equation 3
where     £   is escape fraction
          u   is wind speed, m/s
          w   is pit width, m
         V^   is the larger of deposition or settling velocity,  m/s
          C   is an empirical dimensionless constant with a value of 7.
                                   35

-------
The Fabrick  escape  fraction was evaluated  for  both the universal and  the  EDS
particle  size  distributions for  each of  about 800  individual smoke  release
episodes  by  substituting into  Equation 3  values  of  effective pit width  and
out-of-pit wind  speed  from  the field  data,  and the  larger  of deposition  and
settling  velocity.    The particle  size  distributions  were  subdivided  into
discrete size categories so that individual escape  fractions  could  be  computed
fcr each  size  category.  Multiplying  these individual escape  fractions  by  the
mass fraction  for  each size category,  and  then summing  the  product over  the
entire  particle  size  distribution,  yielded escape fractions  for  each  particle
size distribution.  These  computations were performed  in a   subroutine  of  the
data  reduction computer program.   The  universal   and  the EDS particle  size
distributions  are  shown  in Tables  4.5 and 4.6.   Note that  the  distributions
appearing  in Tables 4.2 and  4.3  are  identical  to the distributions  shown in
Tables 4.5 and 4.6, although the former distributions  (Tables 4.2  and  4.3)  are
expressed  cumulatively with particle  size, whereas  the  latter  distributions
(Tables 4.5 and 4.6) are expressed by particle size category.

4.5.8 WINCES ESCAPE FPACTION

      Winges  (1981) developed  an  equation to calculate  the  particulate escape
fraction  from  surface mine pits.  The escape fraction is given by:
                                     K  I H                   Equation 4


where e       is  the escape fraction
       V,       is  the larger of deposition or settling velocity, tn/s
       K        is  vertical diffi
       z
       E        is  pit depth, m.
                                  2
K       is vertical diffusivity,  m /sec
 Z
                                  36

-------
                                   TABLE 4.5
              UNIVERSAL PARTICLE SIZE DISTRIBUTION — CATEGORIZED
DIAMETER (a)
RANGE
(microns)
0-2.5
2.5-5.0
5.0-10.0
10.0-15.0
15.0-20.0
20.0-30.0
MASS (a)
FRACTION

0.021
0.073
0.176
0.148
0.115
0.467
MID-POINT
DIAMETER
(microns)
1.25
3.75
7.50
12.50
17.50
25.00
DEPOSITION (b)
VELOCITY
(cm/sec. )
0.070
0.600
1.100
1.400
2.000
2.700
a.  PEDCo & TRC, 1S82, p. 16
b.  Hanna,  et al,  1982, Fig. 10.4, p.70  @  z<
1 cm.,  density = 1.0 g/cc
      The Winges  equation  was also  evaluated  for both  the universal  and  the
ECS  particle  size  distributions for  each  individual  smoke  release  episode.
Values of stability  class  and pit depth were  determined from the  field  data.
Values of  vertical  diffusivity  corresponding  to stability  class were  those
presented by Draxler (1977), shown in Table 4.7.
                                    TABLE  4.7
                          VERTICAL DIFFUSIVITY, m2/sec
P-G STABILITY
KZ
A
50
B
30
C
15
D
7
E
3
F
1
                                       37

-------































vD
-a-
3
CO
H



































































Q
W
SI
oi
o
12
"
1
§
M
H
«
1— 1
OS
H
CO
M
Q
W
pM
CO
w
M
t™2
Sr«
OH

CO
a



































o
W Z
rJ O
g_... !-_,_)
H H
W M
CO CO
O *"N
f*T^ £^ fj
O W Ol
O 03
X/*^ c
^j £5

M M CO
CO CJ --^
o o e
CL, rJ O
Q £> V"'





^^»
CO
^-i*
x«\
U
Z E~»
M W *-x
hJ O 03
H.__3 C3
*— ' J3
W bJ cj
CO > ^->-





H
^•r » **
H Sx-s
0 M g
^ 9 °
O 94
M
S


03
c
H o
Z u
M JE 0
O 
ON ON
CN O
.






o o
0 0
o m
-* NO








CN ^O
ON ON
CN O
-3- r^




•O CO
1 1
O 0
X X
m m
r^ CN
r-t CM









m in
CO >
4-1
•H
03
c
01
-o
r-T
\,N 0 O
& r4 rJ
O-l p&4 p£4


co ja cj




































CM
00
ON
i— t
U r-l
CO
^ jj
II 111

O «
N CO
-g
O C8

•
a, u-i
o

r-4 00
CO ^O
•
•u rv
o>
M
CO 00
C CN
C •
CO 1-1
33
II
f\
"*^" S
0
Ol,
H
r-t O
M-l
Ol
r-l T3
*o 
g n
O rJ
u o
ta u


T3 Ol
38

-------
5.0   DATA ANALYSIS

      The  substantial  data  base  collected  both  during   the  original  smoke
release experiments  and  from analysis  of  the video  tape  recordings  of  those
experiments  provided the basis  for the  statistical comparisons  described  in
this section.  The data  consist  of meteorological data, collected  both  in and
out of the  four  mine pits,  escape times of the  smoke  puffs  within each  of the
pits, and discrete categories  which describe the movements,  or  flow patterns,
of the puffs within those pits.

      Several  different  types  of comparisons  are  provided  from  analysis  of
these  data.   Out-of-pit   meteorological  data   were  correlated  with   the
meteorological measurements  performed  at the same  time within the  mine  pits.
In  addition,  the  out-of-pit  meteorological  data  were compared  to  the  flow
pattern  categories,  to determine  the  relationships between  the  observed  puff
movements and  conditions  occurring outside of the  pit.  The escape velocities
of  the  smoke puffs  described  in Section 4.5.4  were  then  correlated  with the
meteorological measurements  to  determine  if  a  predictive method  could  be
developed.   Finally, escape  fractions were  computed  and  compared to  escape
fractions predicted  by existing equations.

      The statistical  tests  or methods used  to  facilitate  the  above-described
comparisons varied depending on  the  type  of  data being used.  Single variable,
linear   regression   analysis   and   direct   comparisons  of   mean  values  were
performed on  the meteorological  measurements  from in and out of  the mine  pits
to  determine not  only  relationships  between  the two  data  sets,  but also  to
ascertain if  in-pit  conditions could be predicted from knowledge of out-of-pit
conditions.   Comparisons  of  flow  patterns  to meteorological measurements  were
facilitated  through  the use of  frequency distributions,  given  the  fact  that
these   patterns    are   described   by   discrete,   and  arbitrarily  defined,
categories.   Finally,  the  escape  velocities  developed from  the  smoke release
observations  were   compared   to  meteorological  measurements   using   both   a
comparison   of  mean values,  and  through   multi-variate   linear  regression
techniques.   A predictive  equation  which  relates the escape velocity to  each
observed  meteorological  parameter  was   developed  and  tested   using   these
techniques.
                                       39

-------
      The  "Statistical  Analysis System"  (SAS) computer  package was  employed
for all  statistical  comparisons  described in this section.   It  provides means
for not  only analysis  but  for  data  programming, storage  and  retrieval,  and
file manipulation.

5.1   COMPARISON OF IN-PIT VERSUS OUT-OF-PIT METEOROLOGY

      A  comparison of  mean values of  meteorological parameters measured  both
in  and  out  of  the  mine  pits  is presented  in Table  5.1.   The  means  were
computed for all data combined,  for each  mine  individually,  for each stability
class, and for each wind speed category.

      Variations  in  wind  direction between  the  inside  and  outside  of  the pit
have  been  quantified by  taking  the difference  between the two values (wind
direction  (in)  - wind  direction  (out),  not  to  exceed 180°)  for  each smoke
release  experiment.  From  Table  5.1,  in-pit wind directions  varied  from those
measured  out-of-pit  by  almost  60°  for  all  data  combined.   Measurements
performed  at  the  Rosebud Mine varied  the most — about 87°,  while  those  from
Carter's  Caballo  Mine  varied  the least  —  36°.   Trends  in wind  direction
differences  between  the  inside  and  outside   of   the   mine  pits  are   not
discernible when  compared by Pasquill-Gifford stability  categories.  However,
when  viewed  by  wind   speed   categories,   a  very   marked  pattern  emerges.
Consistently greater differences between  in-pit  wind directions and out-of-pit
wind  directions occurred during light  wind  speeds.   In fact,  such variations
are  two to  almost  four times  greater during the  lightest wind  speeds (less
than  4 mph) than during  the high speed categories (greater than 12 mph).  This
result  is  not unexpected  as wind directions  are  typically more variable during
light winds versus high winds,  so  comparisons of measured  directions  from two
different  locations,  even in   close  proximity,  are  usually not  favorable.
However,  these  results  do  indicate that wind  directions  within  a mine  pit
probably cannot  be well represented by measurements performed  outside  of the
pit.  This finding will be explored further  later in this section.
                                       40

-------
o
cc. m
0
O
H
< sr


o
tn
u en
ft,
oo

Q
2. CN
M
3

U
C i-i
u
2

o
o
h4
o
«l
H *"*
< H 42 sr
> S EM
r, W H
S3 f] OT M

* 1^ i OP ^"^ CO
m x i pg
8 <; <
W Fn 0 H H
tj o eL «c to
CQ Q CN
2§B ?
W .- PH
M W
3 ^ IH

a w
£>
U H
PL '^
I
z

«^-

cd
a
M
ro
u
55
H
CN




i— 1








i_J
<



Pi

H
W
5£
3
c2
^
PM

CT. 00
i— 1 •
CN
f"H

o c"*

in






00 O
m
C
C 0
•H 11
4J

.
CN

O
sr

^°.
CN

%

O
in

m
.
cn

m
•
co

o

xO


VO

a\

m
•
o
i— i

m
•
m






00



1
*X3 *~*
01 or
01 O-
a e
CO ^
TJ
C u
•H 3
3 0








^x
CM* ^^
Q co
U 01
Z. CJ
^ C
0>
Pb h
pg (-1
E- 3
2. u
W CJ
U 0
co K BE
<; w K PH ft, o
t- M K PH 2 2 C
<; pi P-i J; ^- '
o o s: o oo
CJ» mj CT\ CTt rH
O [14 PxC?\O • 'X
M H s sr . oo sr PM
H < • CN iH CN S
< U en l~^ rH 1 1
£ O 1 1 O O CT\
M & • sr o r-t o 1-1
i-i w sr o m ' . •
o w I • • CM cr\ sr
P-i o sr r~ i— i i— l CN
<
OB i-tcNensrmvo
^
^
CJ













^
w
w
P5
c^
SQ
CJ 3
< < >j a «
PL, CL, < M Cd
2 2 PQ pa co
Q <3 < » >H cj co as

w
i— i i— i c\i cn sr m
J2
cfl



















00

iH
>. C
4-. O
•t-i
rH 01
•H CJ
pO C
Cfl -H
U CO
CO
in
O
1 ^%
P-i M
O
X 00
<4H 01
•H 4J
4J CO
C O
Ol
T3 O
i-l 1
PH
0
i-l l-i
O
Ol Ol
CO "O CO
3 01 CO
U J2
CO 3
•H O. CO
8 xj
C 0 cfl
O CJ T2
•H
4-1 4-1 0)
C 0 U
Ol C -rt
> 4-1
C 01 C
O M Ol
U 01
3 01
00 —
C W w
•H C
O 01 -H
iH 32

O 01
U-l • >
U >-i
0) 4-1 Ol
4= O) CO
^J J?)
- O
m o
II 01
c cn M
O 0)
•n « IS
4-1 m
U II CO
4) CN Ol
c/3 a
« c
4-1 < (U
3 II !*
O iH ^
J3 3
hfl • • O
3 co o
O 0) O
(-1 CO
43 CO J3
H cfl 0
iH 3
43 CJ CO













41

-------
      Average wind speeds occurring  inside  and  outside the mine pits  are also
listed in Table 5.1.  Averages were  computed  for  all  data combined,  as well as
for  each  mine,   stability   class,   and  wind   speed  category.    Since  the
categorization of  wind  speeds was  accomplished using out-of-pit  wind speeds,
the  mean  out-of-pit  values  listed  by  category  represent  essentially  the
midpoints  of  those  categories.    However,   comparing  these  means   to  those
occurring within the pit is nevertheless instructive.

      From Table  5.1,  it is  evident  that in-pit wind speeds  are  consistently
less  than  those  occurring   out-of-pit   at   the  same  times.   This  is  true
regardless  of  which mine  is  analyzed,  and  is  essentially  true  for  all
stability categories as  well.  The one exception  to this  observation is during
the  most  unstable — Class  A — conditions, where  only  15  observations were
available for analysis.   The  largest differences between  wind  speeds  measured
in and out of the mine pits occurred under the highest wind speed categories.

      Measurements of sigma-theta atmospheric stability both  in and  out of the
mine  pits  were  performed  during   the   original smoke   release  experiments.
Comparisons  of  such measurements  would be useful, however,  examination of the
data  revealed  that the  vast  majority of the stabilities  were  class "D", both
in and  out  of  the pit.    The  Pasquill-Gifford stability  distribution  indicates
that  only  10 percent  of  the  episode hours were class  "D", which is a much more
reasonable distribution  for  summer  daylight  hours.   Given the  low  confidence
in   the  accuracy  and  representativeness  of  the  sigma-theta  measurements,
comparisons  of  in-pit versus out-of- pit stability categories  thus  determined
are  not provided.

      Linear  regression  analysis   was   employed  to   determine  if   in-pit
meteorological  conditions  could   be  accurately   predicted  from  observations
performed  outside  the  pit.   Regression analysis  essentially  determines the
"best fit" of a linear expression of  the  form:
where   y   is  the  dependent  variable  (or  in-pit  measurement),  X  is   the
independent  variable  (or out-of-pit  measurement), b   is the  y-intercept of
the  "best-fit"  line  determined  using  the  least-squares method,  and  b,  is
slope  of  the best fit  line.

                                        42

-------
A  correlation  or relationship  between  the dependent  (y)  and  independent  (X)
variables is  indicated  if  the  slope of  the  "best-fit" line  is not  equal  to
infinity  or   zero.   A  measure  of  the  "goodness-of-fit"   of   the  linear
expression,  or model,  to the observed  data is described  by the  variation  of
the actual measurements  from the "modeled" values.  This variation is usually
                               2
expressed  in  terms  of  the  r   parameter  — a  dimensionless  number  which
                                                             2
ranges in value  between 0 and 1.  The  larger  the  value of  r  ,  the better the
model's fit.
      Linear regression  analysis was applied  to  determine whether or  not  the
wind direction within  a  mine pit could  be  determined  from a knowledge  of  the
wind direction  outside of the  pit.   Because of  the  discontinuity represented
by  north  winds  (0°  versus  360°),  out-of-pit  wind  directions  (WD)  were
expressed in terms  of  the long  axis  of  the pit by defining a  variable  called
"TACK" constrained to values between 0 and 90 degrees:

          •   0  <  WD  < 90; TACK = WD
          •   90 <  WD  < 180; TACK = 180-WD
          •   180 < WD  < 270; TACK = WD-180
          •   270 < WD  < 360; TACK = 360-WD

The  variable  TACK then  represents  the  relation of  the  wind direction  to  the
long axis  of  the pit.   A value  of  TACK  equal to 0  represents  a wind parallel
to  the  long  axis  of  the  wind,  while a  value  of  90°  represents  a  wind
perpendicular to  the pit's  long  axis, or a  crosswind.   To avoid the  northwind
discontinuity in  the  in-pit wind directions, the variable TACK was correlated
with the difference (AWD) between the in-pit wind direction and the out-of-pit
wind direction as used previously in Table  5.1.   The results of the regression
analysis:

where
and
y
y
X
= b +
o
= AWD
= TACK
are  shown in Table  5.2  The model  was evaluated  for all data  combined,  for
each mine individually, and  for  each stability class.  The Student's  -  T test
was  applied  to  evaluate  whether  the  slope,  b.. ,  of  the  best-fit  line  is
                                 43

-------
different from zero  (ie., whether  y  is  a function of X).  The parameter listed
as  PR   T in  Table  5.2  evaluates  whether  or  not the  computed slope,  b.. ,  is
statistically different from zero; values  of PR T that  approach zero indicate
good correlation between y and  X,  whereas  values  that approach unity represent
poor correlation.
                                    TABLE  5.2
                RESULTS  OF  REGRESSION ANALYSIS  FOR WIND  DIRECTION
MODEL
PARAMETER
ALL
b0 62.3
t>i 0.1
PR T .2248 .
r2 .002 .
DATA BASE
MINE
1
46.6
0.4
0091
063
2
14.2
1.3
.0001
.441
3
68.3
-0.5
.0017
.050
4
106.7
-0.9
.0001
.216
5
44.7
0.9
.0001
.284
1
36.5
0.5
.1340
.164
P-G STABILITY
2
69.1
-0.1
.7055
.001
3
57.5
-0.2
.1193
.007
4
27.7
0.5
.0081
.091
6
108.9
-0.7
.0528
.063
      The  results  in Table  5.2  indicate that the  out-of-pit  crosswind angle,
 TACK,  is a poor predictor of  WD,  the  difference  between  in-  and out-of-pit
 wind  directions.    In  other  words,  the  out-of-pit wind  direction is  a poor
 predictor  of  the  in-pit wind direction.  For all data  combined,  the  slope  of
 the  regression  line can be considered different from zero only at  a confidence
 level  of about 78% (from PR  T,  1.0  -  .2248 =  .7752).  Hence  variations  in
 out-of-pit  wind directions   characterize  less  than 1% of  the   variation  of
                                     2
 observed in-pit variations   (from  r  ,  .002 x  100 =  0.2%).  The regression
 model is improved somewhat when  the  data are separated  for each mine, but even
 at  Mine 2  (Yampa,   location 2), only 44%  of  the  variability of in-pit wind
 directions  can be  explained.   Stratification of  the  data  by stability  class
 does  not improve the  fit  of the model either.   From this  analysis,  it can  be
 concluded  that in-pit  wind  directions  can  not  be adequately  predicted  from
 knowledge  of  out-of-pit  wind directions.

      A single variable linear  regression  analysis was performed  to  determine
 if  in-pit  wind  speeds are  related  to  wind speeds measured  out-of-pit.   For
 this  test,  the  out-of-pit  speed  was  used  as  the  independent variable —  X,
 while  the in-pit  wind  speed was  used  as   the dependent  variable — y.   The
 results of this investigation appear in  Table 5.3.
                                      - 44 -

-------
                                    TABLE 5.3
                  RESULTS  OF REGRESSION ANALYSIS FOR WIND SPEED
MODEL
DATA BASE
PARAMETER MINE

bo
bl
Pte-T
r2
ALL 1 2 3 4 5 1
1.5 2.0 3.6 -0.1 2.2 -0.5 0.9
0.6 0.6 0.4 0.7 0.4 0.9 1.0
.0001 .0001 .0001 .0001 .0001 .0001 .0008
.61 .55 .58 .73 '.34 .75 .60

P-G STABILITY
2 346
1.1 1.2 -0.4 1.1
0.8 0.6 0.7 0.4
.0001 .0001 .0001 .0003
.56 .53 .63 .19
      The results  listed  in Table 5.3  indicate  that  the out-of-pit wind speed
is  a  reasonably  good  predictor  of  the  in-pit  wind  speed.    For  all  data
combined, the  small value  of PR=»T  indicates  that the  dependent  variable is
                                                                     2
strongly  correlated  with  the  independent  variable.   Further, the  r  value of
0.61 indicates that 61% of  the variation of  the  in-pit wind speed is  explained
by the regression model.   Stratification of  the  data by mine, and by  stability
class  yielded  essentially  similar results.  An interesting  exception occurs
when  the wind  speeds  are  analyzed  during  stable  atmospheres  (P-G class  6).
For  these conditions,  even  though a  strong correlation  is  indicated  by  the
Student's T  test,  only 19%  of the variation of  the  in-pit  wind  speed can be
explained by the  model.    It  is  not  clear why  the  model's   performance  is
reduced  during these conditions,  but  it is proposed that conditions within the
mine  pit are  sometimes  decoupled  from   the  atmosphere  above  during  stable
conditions.

5.2   COMPARISON OF OBSERVED FLOW PATTERNS AND METEOROLOGICAL CONDITIONS

      The videotape  recordings of  the actual smoke   release experiments  were-
reviewed  to  determine  not  only  the smoke  puff  escape time  from  which escape
velocities were computed, but  also  to  determine  the characteristic  patterns of
the  puff's transport and  dispersion within  the mine  pits.   The  categorization
of the observed  flow patterns was accomplished  through the use of  the coding
forms described in Section 4.0 of this report.
                                         45

-------
      The discrete categories obtained  from  analysis of the data  contained  on
the coding forms are  as follows:
                                   TABLE 5.4
              CATEGORIZATION SCHEME USED TO DESCRIBE PUFF BEHAVIOE
PUFF REMAINS                     PUFF EXITS
IN PIT                              PIT                 DON'T KNOW
100-recirculation            10-puff thermally up       1-invalid sample
     evident                    sidewall
200-puff disperses           20-puff mechanically       2-videotape ended
     in pit                     driven up sidewall         prematurely
                             30-puff exits at center    3-puff too diffuse
                                of pit
Definitions for each of the above categories were  provided  in  Section 4.0.   It
should  be  noted  however  that  two  of the  categories do  not  appear in  the
analysis.   Samples  considered  by  the observer  to  be invalid  (Category  1),  43
out  of  the  original  811  puff releases,  were  extracted  from  the  data  base
before  analysis.   Also,  puffs considered  to  have escaped  the pit  by thermal
processes  (Category 10)  were  combined with those driven mechanically  up  the
side    wall    (Category    20)   as    the    two   categories   were   virtually
indistinguishable.   Undoubtedly  both  processes must  occur within  mine  pits,
however,  it was found to  be  impossible to distinguish  between them  from  the
available visual record.

      Data  which are  comprised  of  discrete,  arbitrarily  defined categories,
such as  the flow patterns  described  above,  do  not  lend themselves  to the types
of  statistical  analysis   used  for  other   parameters.   It  is meaningless  to
compute an "average"  flow pattern,  just  as it  is impossible to  state  that a
flow  pattern  is  proportional  to  (or  correlated   with)  another  parameter.
However,  it  is  instructive  to   identify the categories which occur  most
frequently,  and  to  examine   the  frequency  of occurrence as  a   function  of
meteorological  conditions.  For this  reason,  the  comparison  of  observed flow
patterns  to measured   meteorological  conditions is comprised of comparisons of
                                        46

-------
the   frequency   of   occurrence   of   specified   flow   patterns   for   given
meteorological parameters.   These  frequencies,  expressed as  percentage  of all
occurrences  of a  given meteorological  condition,  are  listed  in  Tables  5.5a
through 5.5c, along with total  count  of occurrences.   Every vertical column of
frequencies in Table 5.5 sums  to  100 percent,  so that each individual entry in
the Table  represents the  frequency of occurrence of  a  specified  flow category
observed during a  specified  meteorological  condition.  For example, of all the
smoke  release  episodes during  which the  P-G  stability  class  category was  1
(left  most  vertical   column of  values  under  heading  P-C  STABILITY  1),  27
percent of  the episodes exhibited puff  dispersion  while the puff  remained in
the  pit.    Similarly,   of  all  the  smoke  release  episodes  during which  the
out-of-pit  wind  direction  was  cross-wind  to the long  axis of  the  pit  (right
most vertical  column of Table  5.5c), 5  percent  of  the episodes exhibited puff
recirculation.

      From  Table  5.5,   the  observed  flow patterns  appear  to relate better to
atmospheric stability  categories  than to either  the  wind  speed categories, or
to  wind  directions.   For example, during the most unstable  conditions  — P-G
Class  1 — 53% of the  smoke puffs were  observed to  exit  the  mine pit, while
only  27%  remained in  the  pit.  In  contrast,  during  stable  conditions  — P-G
Class 6 — only 16% of  the observed  puffs exited  the  pit while  58% remained in
the  pit.    However,   the  distribution   of   flow  patterns  remains  reasonably
constant  for  different  wind  speeds,  and  for  different  wind  directions,  at .
least  in  terms of the  primary categories  (ie.,  "plume  remains in pit"  versus
"plume exits pit").
                                        •
      Other  observations  are  possible  from examination  of the sub-categories
of  flow patterns.   Recirculation of  smoke  puffs  were observed  more frequently
during neutral atmospheres  (P-G Class  4),  and  during  the  highest wind  speeds
(wind  speed   Category  5)   than  during  other  stability  and   wind   speed
categories.   Evidently the  recirculation patterns observed  in the  mire  pits
were  most  frequently  associated  with  aerodynamic  wake  effects   (ie.,  smoke
puffs  trapped  within  an   aerodynamic  cavity  formed  alongside  the  upwind
sidewall).   Another  anticipated  recirculation  pattern, cellular  circulation
structures  caused  during  light   winds  by differential   surface   heating  or
cooling, were  also observed  but much less frequently.
                                       47

-------
                                   TABLE 5.5a
                PERCENT FREQUENCY OF OCCURRENCE AND TOTAL NUMBER
                OF OCCURRENCES (in parentheses) OF OBSERVED FLOW
                  PATTERNS FOR GIVEN METEOROLOGICAL CONDITIONS
FLOW CATEGORY
                        P-G STABILITY
                      2              3
PUFF REMAINS IN PIT
100-RECIRCULATICN
200-DISPERSED
CAT. TOTAL
0 ( 0)
27 ( 4)
27 ( 4)
5 ( 13)
25 ( 65)
30 ( 78)
2 ( 8)
32 (109)
34 (117)
11 ( 8)
33 (25)
44 (33)
4 ( 2)
54 (35)
58 (37)
PUFF EXITS PIT
20-EXITS SIDE WALL     20 (3)      22 ( 57)      23 ( 79)    18 (14)
30-EXITS CENTER        33 ( 5)	11 ( 29)	2(8)     0(0)
CAT. TOTAL             53 (8)      33 ( 86)
                                                      12 (  8)
                                                       4 (  2)
                                25  (  87)     18  (14)    16 (10)
DON'T KNOW
2-TAPE ENDED
3-TOO DIFFUSE
CAT. TOTAL
0 ( 0)
20 ( 3)
20 ( 3)
3 ( 7)
35 ( 92)
38 ( 99)
3 ( 11)
37 (127)
40 (138)
0 ( 0)
38 (29)
38 (29)
0 ( 0)
26 (17)
26 (17)
Column total
    100 (15)      100 (263)      100 (342)    100 (76)   100 (64)

(760 Valid P-G Stability Class  Observations)
                                       48

-------
                                   TABLE 5.5b
                PERCENT FREQUENCY OF OCCURRENCE AND TOTAL NUMBER
                OF OCCURRENCES (in parentheses) OF OBSERVED FLOW
                  PATTERNS FOR GIVEN METEOROLOGICAL CONDITIONS
FLOW CATEGORY
                    WIND SPEED CATEGORY
                      2              3
PUFF REMAINS IN PIT
100-RECIRCULATION
200-DISPERSED
CAT. TOTAL
2 (
39 (
41 (
I 76)
: 80)
5
29
34
(
(
(
9)
54)
63)
2 (
30 (
32 (
: 5)
: 77)
: 82)
9 (
26 (
35 (
10)
30)
40)
7
17
34
( 1)
( L)
( 2)
PUFF EXITS PIT
20-EXITS SIDE WALL     18 ( 35)     23 ( 42)      23 ( 59)
30-EXITS CENTER   	9 ( 18)     10 ( 19)       3(8)
CAT. TOTAL             27 ( 53)     33 ( 61)
                                            22  (26)
                                             0  (  0)
           33 ( 2)
            0 ( 0)
                                26  (  67)     22  (26)    33  (  2)
DON'T KNOW
2-TAPE ENDED
3-TOO DIFFUSE
CAT. TOTAL
4 ( 8)
28 ( 55)
32 ( 63)
3 ( 6)
32 ( 59)
35 ( 65)
1 ( 3)
42 (109)
43 (112)
3 ( 3)
41 (47)
44 (50)
0 ( 0)
33 ( 2)
33 ( 2)
Column total
    100 (196)     100 (189)      100 (261)

(768 Valid Wind Speed Observations)
100 (116) 100 ( 6)
                                       49

-------
                                   TABLE 5.5c
                PERCENT FREQUENCY OF OCCURRENCE AND TOTAL NUMBER
                OF OCCURRENCES (in parentheses) OF OBSERVED FLOW
                  PATTERNS FOR GIVEN METEOROLOGICAL CONDITIONS
FLOW CATEGORY
                           PARALLEL
                   OUT OF  PIT WIND  DIRECTION
                                        CROSS-WIND
PUFF REMAINS IN PIT
10C-RECIRCULATION
200-DISPERSED
CAT. TOTAL

PUFF EXITS PIT
20-EXITS SIDE WALL
30-EXITS CENTER
CAT. TOTAL
Column total
          3 (
         32 (
 8)
90)
         35  (  98)
         22  (  62)
          9  (  25)
         31  (  87)
 5 ( 24)
30 (146)
35 (170)
                            21 (102)
                             3 ( 15)
                            24 (117)
DON'T KNOW
2-TAPE ENDED
3-TOO DIFFUSE
CAT. TOTAL
2 ( 6)
32 ( 90)
34 ( 96)
3 ( 15)
38 (185)
41 (200)
        100 (281)

(768 Valid Wind Direction Observations)
                            100 ( 487)
                                      -  50

-------
      To summarize  the  findings of  this  investigation of  characteristic  flow
patterns it was determined that:

      •   smoke  puffs   remained   in  mine  pits  more   often   during   stable
          atmospheres than during unstable atmospheres;

      •   smoke  puffs   remained  in  mine   pits  more  often  during  light  wind
          speeds than during high wind speeds;

      •   recirculating  puffs  are mostly  associated  with  "downwash"  cavities
          formed alongside the sidewalls during  high winds;

      •   puff  exits at  the center  of  the  pit  (as opposed  to exits  along
          sidewalls)  are  most   frequently  associated  with   very   unstable
          atmospheres and light wind speeds;

      •   circulation  patterns  exhibit a greater  association  with  stability
          categories than with wind speed categories or with wind directions.

5.3   COMPARISON OF ESCAPE VELOCITY TO METEOROLOGICAL CONDITIONS
      The  videotape  recordings  of  each smoke  puff  release  were analyzed  to
determine  the  amount  of time,  termed  retention time,  required for  the  smoke
puff to  escape the pit.  Given the depth of  the  pit and  the  retention  time,
the upward  escape  velocity  was  computed  for each observation.   This velocity
was then compared  to coincident  meteorological  conditions  measured  both in and

out  of  the  mine  pit.   The  techniques used  to  make  these   comparisons  are

similar to  those described  previously  in  this  section — namely, comparison of
mean values, and regression analysis.


      Average  escape  velocities were computed  for all data combined,  as  well
as  for  each mine  individually,  each stability category,  and   each  wind  speed
category.  These results are presented in Table 5.6.
                                       51

-------
                                   TABLE 5.6
                     AVERAGE ESCAPE VELOCITY AS A FUNCTION
                        OF METEOROLOGICAL PARAMETERS (a)
DATA BASE
ALL
MINE= 1
MINE= 2
MINE= 3
MINE= 4
MINE= 5
ESCAPE DATA BASE
VELOCITY
53.
40.
50.
59.
49.
65.
4 STAB= 1
3 STAB= 2
6 STAB= 3
5 STAB= 4
0 STAB= 6
9
ESCAPE DATA BASE ESCAPE
VELOCITY VELOCITY
39.6 WS. CAT.= 1 34.3
48.4 WS. CAT.= 2 47.2
61.5 WS. CAT.= 3 65.3
54.2 WS. CAT.= 4 67.5
30.6 WS. CAT.= 5 74.2

    velocity expressed in cm/sec.
      From Table  5.6  it is  seen that the computed escape  velocity varies for
each mine,  stability  category,  and  wind  speed.   The average  escape velocity
was found  to be much higher for Mine 5  (Rosebud)   and Mine  3 (Caballo) than
for the other mines tested.   From  the configuration of the mines  tested it is
unclear why  the escape velocity  for  Mine  5 (Rosebud)  should be higher than for
the other mines.  The Rosebud mine pit is about as wide as it  is  deep,  as are
the Yampa  (.at Location  1),  and the Spring Creek mines (see Table 4.4).  If the
mine configuration  were the  only  factor  affecting  the escape  velocity,  then
the values  from these three  mines  should be roughly  the  same.   Evidently the
mine  configuration  is  only  one of  the   factors  which  determine  the  escape
velocity.

      The  distributions  of  escape velocities  as functions of  stability class
and  wind  speed  are  more   in  line   with expected   results.   Average  escape
velocities   were  found   to  be  highest  during   neutral   and  near   neutral
atmospheric   stabilities   (P-G   Classes  3-4),   and   lowest  under  stable
conditions.   Consistent  with these  results, higher  escape velocities occurred
during high wind speeds, than during  low wind speeds.
                                     52

-------
      It  is  likely  that  the  out-of-pit  wind  speeds  are,  in  fact,  the  most
significant  factor   in  determining  the   escape   velocity.    The  stability
categories are themselves a  function of  wind speed.  Both  the  most stable and
the  most  unstable  categories  occur  during light  wind  speeds,   while  near-
neutral  conditions  are associated  with  high wind  speeds.   Since  the  escape
velocities during  the  most  unstable  categories  (P-G Class  1) are  less  than
during neutral conditions,  even  though  unstable  atmospheres enhance vertical
movements, it is likely that  the  magnitude  of the wind speed is more important
in  determining  the  escape   velocity  than  is   the   thermal   stratification
described by the stability category.

      The  importance of mine configuration  and  meteorological conditions  in
determining  the  escape   velocity   can  be  quantified  more   readily   using
regression analysis.   It  is  apparent  that this velocity  is  controlled  by  a
variety of factors, so  the most appropriate analysis tool for evaluating these
effects  simultaneously is  linear,  multi-variate  regression  analysis.   This
sophisticated statistical approach is available on the SAS computer package.

      A  linear  model  constructed   using  a  multivariate  regression  approach
takes the form:

      y = b  + b,X1 + b0X0 +  b,X. ... b X
           o    11     2. 2.    JJ      nn

where  y  is  the  dependent  variable,  b   is  the intercept  of  the  best  fit
"line",  and  b,   through   b    are  regression  parameters,  equivalent  to  the
slope  of  the line  in  single  variable regressions,  for  independent variables
X^  through  X  .   Interpretation  of  the  results  of multivariate  regressions
is   similar   to   that  described   previously  for   single  variable   linear
regressions.     The   regression  parameters,   b..   through  b ,    should   be
significantly different from zero,   as  indicated  by the  Student's-T test,  for
                                     2
significant  correlation,   and  the   r   value,  a  measure  the  model's  "fit",
should approach unity for a good-performing  regression model.

      Two  basic  models were  tested using  multivariate  regression techniques.
The  first was based  on meteorological  measurements performed  outside  of the
nine  pit,  and  the  second   was  based   on  in-pit  measurements.   Table  5.7
describes  the variables used  in each of  the models, while  Table  5.8 lists the
results of the regression analysis.

                                      - 53  •

-------
                                   TABLE  5.7
               PARAMETERS  USED  IN MULTIVARIATE REGRESSION ANALYSIS
DEPENDENT
MODEL Y
1 ESCAPE
VELOCITY
2 ESCAPE
VELOCITY
INDEPENDENT
Xl
WIND SPEED
(out)
WIND SPEED
tin)
X2
TEMPERATURE
(out)
TEMPERATURE
(in)
X3 X4
TACK P-G
STABILITY
TACK P-G
STABILITY
X5
WIDTH
WIDTH
The reader is reminded that the variable TACK  represents  the  angle of the wind
direction with  respect  to the  long  axis of the mine  pit.   Angles approaching
zero  represent  winds that  are  parallel to  the long  axis of  the  pit,  while
angles approaching 90° represent crosswinds perpendicular to the long axis.
                                    TABLE  5.8
                   RESULTS  OF  MULTIVARIATE REGRESSION  ANALYSIS
NO.
MODEL SAMPLES bo PR>T
1 634 3.41 2.88
.0001
2 543 -6.27 3.94
.0001
b2
PR>T
0.48
.0001
0.47
.0001
b3
PR=-T
0.11
.0001
0.10
.0002
b4
PR^T
-3.49
.0001
-0.94
.0949
b5
PR>T
-0.01
.0154
-0.01
.2040

r2
0.32

0.32

      The  results  indicate  that  neither of  the  models  tested  can adequately
predict  the  escape  velocities  computed from  the  smoke  release experiments.
Regardless  of  whether  in-pit  or   out-of-pit  meteorological  parameters  are
employed,  only  32%  of the variability of the  escape  velocity can be explained
by the models.   This  is  not  to say  that the  independent  parameters tested are
not  correlated with  the escape  velocity.   In fact,  significant correlations
were  found between the  escape velocity  and the  measured wind speed,  both  in
and  out  of pit,  the  temperature,  both in and out of pit,  the angle of the wind
                                        54

-------
TACK,  and   P-G  stability  category   (a  negative,   but   still  significant
correlation).  The weakest  correlation between the escape  velocity  and one of
the  independent  variables  was demonstrated  for the mine width.   When out-of-
pit  measurements are  used,  the width was  found  to  be  significantly,  but
negatively,  correlated  with the escape  velocity  at the  98%  confidence level.
However,   when   in-pit   measurements   are  used,    the   confidence   level  for
significance drops to  80%.   Given the  fact  that  all  the independent variables
tested demonstrate  some degree  of  correlation with  the escape  velocity,  and
yet  the model  only  accounts for 32% of  the  variation  of the velocity, clearly
some, as yet unaccounted for,  factors  or  variables must influence  the escape
velocity.

      In an  effort to  improve  the performance  of  the model, several iterations
of  the  multivariate   regression  techniques  were  performed.   The  above  two
models were  modified,   by  eliminating  the  variable WIDTH  from  the  analysis,
without a  meaningful improvement in  the  results.   The  same  two models  were
also  evaluated  by  restricting the data base to data  from  each mine,  and each
stability  class,  again  without improvement.   The  results of  each  of  these
investigations  is included as Volume 2 of this report.

      In  summary,   the  escape  velocity  has  been  found  to  be  positively
correlated with wind  speed,  temperature, and  wind direction, and  negatively
correlated with  stability  category  and  the  width of  the pit.  However,  these
parameters,  when used  in a  linear  regression model,  do  not  provide  very good
predictions of  the escape velocity.  Only  32%  of  the  variability  of the escape
velocity can be  explained  by  these  parameters.   Certainly other processes must
act,  in conjunction  with the  variables tested here,  to determine  the escape
velocity.

      Perhaps  it  is  not  surprising  that  a  regression model,  relying  on wind
speed,  wind  direction,  and  stability  class  data  collected  at  just  two
locations  at a  surface  mine would have little  success  in predicting airflow
parameters in  the  pit.  Atmospheric modeling  is  one  of the  most  challenging
simulations  performed  today,  and  even  very elaborate  numerical  models  which
use  complete descriptions  of  the  upwind  flow  field   (velocity,  temperature,
diffusivity, vorticity),  and  exact  representations of the  terrain  features,
have difficulty predicting characteristics of flow fields.
                                       55

-------
5.4   COMPARISON OF ESCAPE FRACTION AND METEOROLOGY
      Escape fractions were computed from four different methods in this study:


      •   SMOKE  PUFF   EXIT  VELOCITY   (SETTLING).    Escape   fractions   were
          determined  for  each of  the roughly  800 smoke  release episodes  by
          calculating  the  puff  exit  velocity  for  each  episode,  and  then
          entering  the curves shown in Figures  4.3 and  4.4 to  find  the escape
          fraction   corresponding  to the  universal and  the  EDS  particle  size
          distributions,  respectively.   This effort,  which was  computerized,
          yielded   roughly   1600   escape   fractions   (800   for  the  universal
          distribution  and  800  for  the  EDS  distribution).   These  escape
          fractions  were  written on magnetic tape for  subsequent statistical
          analysis.

      •   SOURCE DEPLETION  (DEPOSITION).   Escape  fractions  were determined for
          each  of   the  roughly  800  smoke  release episodes  by  computing  the
          escape   fraction   directly  from   the   familiar   source   depletion
          equation.  This effort, which was  computerized, yielded roughly 1600
          escape fractions  (800  for  the  universal  distribution and 800 for the
          EDS distribution).   These  escape fractions  were written on magnetic
          tape for  subsequent statistical analysis.

      •   FABRICK  EQUATION.   Values of  wind speed,  effective pit  width,  and
          deposition  (or  settling)  velocity were  entered  into the  Fabrick
          escape fraction  equation  (section  4.5.7)   for  each  smoke  release
          episode  and  for both  the  universal and the  EDS  size distributions.
          The  individual  values  of escape  fraction  determined by the Fabrick
          equation  were written on magnetic  tape for  subsequent statistical
          analysis.

      •   WINCES EQUATION.    Values  of  pit   depth,  vertical  diffusivity,  and
          deposition  or  settling   velocity were entered   into  the  Winges
          equation  (section  4.5.8)  for each  smoke  release  episode and for both
          the universal and  the  EDS  size  distributions.   The individual values
          of escape  fraction determined  by the  Winges equation were written on
          magnetic  tape for  subsequent statistical analysis.

      The  values   of   escape  fraction  inferred  from  the  smoke  puff  exit

velocity,  inferred  from  the  source  depletion equation,  calculated  from the

Fabrick  equation,  and calculated from the Winges  equation,  are shown in Table

5.9, grouped by stability class.
                                        56

-------
                                   TABLF 5.9
                       ESCAPE FRACTION BY STABILITY CLASS
DISTRIBUTION
UNIVERSAL




EDS




P-G STABILITY
1
2
3
4
6
1
2
3
4
6
EXIT VELOCITY
(SETTLING)
1.00
1.00
1.00
1.00
1.00
0.81
0.85
0.93
0.90
0.70
SOURCE DEPLETION
(TEPOSITION)
0.93
0.88
0.86
0.81
0.58
0.59
0.46
0.43
0.36
0.21
WINCES
0.99
0.98
0.96
0.92
0.58
0.90
0.84
0.73
0.59
0.20
FABRICK
0.58
0.72
0.85
0.88
0.68
0.11
0.17
0.28
0.32
0.14
From Table 5.9, it is seen that escape  fractions  for  the  0-30 micron particles
that make  up the  universal  particle size  distribution are  larger  than  those
associated with the 0-130 micron  EDS  distribution.   This  is  true  for  all four
escape  fraction  computation  methods,   and  reflects  the  fact that  deposition
velocity and  settling  velocity are parameters  that appear in the computation
of all  four  escape fractions.   Intuitively it seems  reasonable  that a greater
fraction of  the large EDS distribution  particles  would be  retained  in  the  pit
than would the smaller diameter universal distribution particles.

      The escape fractions inferred  from field  data by both  the  ex:t  velocity
(settling)  and  the  source  depletion  (deposition)  methods  clearly show that
less  particulate  escapes from the pit  during  stable  ("F"  class)  conditions
than  during  other  stability   classes.   This finding  suggests  that the  very
stable  atmosphere  suppresses  vertical motion, causing  more particulate  to  be
retained in  the  pit.   The Winges  escape fractions  agree  well with  the  escape
fraction inferred  from the source  depletion calculations, and  are also much
smaller during stable  conditions  than during unstable  and  neutral conditions,
which  reflects  the  presence   of  vertical  diffusivity  (K  )  in  the  Winges
equation.   The  Fabrick  escape fractions  do not  exhibit  the  characteristic
decrease in  magnitude  with  "F" stability.   This  is to be  expected,  since  the
Fabrick  escape  fraction  equation is  not  a  function  of  stability class  or
vertical diffusivity.
                                       57

-------
      An  evaluation of  the  variations  in  escape  fraction with  wind  speed
category,  shown  in Table  5.10,  exhibits  the  same trend  as the variation  in
escape velocity discussed earlier.  The magnitudes of  escape fraction  increase
with increasing wind speed, for all four  escape fraction techniques.   However,
                                   TABLE 5.10
                       ESCAPE  FRACTION BY WIND SPEED CLASS
DISTRIBUTION
UNIVERSAL




EDS




WS CLASS
1
2
3
4
c
1
2
3
4
5
EXIT VELOCITY
(SETTLING)
1.00
1.00
1.00
1.00
1.00
0.75
0.85
0.96
0.96
0.99
SOURCE DEPLETION
(DEPOSITION)
0.78
0.84
0.86
0.88
0.88
0.35
0.46
0.43
0.43
0.43
WINCES
0.90
0.91
0.95
0.95
0.96
0.70
0.70
0.73
0.69
0.76
FABRICK
0.60
0.81
0.87
0.92
0.93
0.10
0.20
0.28
0.38
0.43
escape  fractions  determined by  the  Fabrick equation  exhibit slightly  better
agreement with escape fractions  inferred  by  the  settling  and deposition models
than  do the escape fractions  determined  by Winges  equation.   The  reason  for
this  may  be that  the  Winges  equation  does  not  include  wind  speed  as  an
explicit  parameter,  as  does   the  Fabrick equation.    Inclusion  of  wind  speed
directly  into the  Winges equation may improve its ability  to  match the trends
observed  in the  escape fractions inferred  from the  settling  and deposition
models.

      The  differences  between escape  fractions  inferred  by the  exit velocity
(settling)  method  and   the  source  depletion   (deposition)  method  raises  an
obvious question:   "Which  method is  correct?"  Unfortunately,  no  answer  is
possible  because  the  field test was not designed  to measure escape fraction-
directly,  nor  do  the  data lend  themselves  to  straightforward  computation  of
escape  fraction.   Both  the  exit  velocity  and the source depletion methods used
to  infer  escape  fractions  incorporate  numerous assumptions  (see  Sections 4.5.5
and  4.5.6 of this  report),  the  validity of which  cannot be checked  with  the
available  data.  However, it should  be remembered that there is some reason to
                                       58

-------
suspect that  the  escape fractions inferred  from the smoke puff  exit  velocity
may be  too  large,  especially  for  small diameter  particles.   The  reasons  for
this  are  that  the  computation of  exit velocity  for some of  the  smoke  puff
episodes probably  overestimates that  velocity  (see  Sections  4.2  and  4.5.4),
and that the  exit velocity method does not  take into account  the plume-terrain
interaction that governs particle deposition (see Section 4.5.5).

      The grouping of escape  fractions by mine  is  not very instructive.  There
is no apparent trend or pattern discernible for any of the escape fractions.

-------
6.0   SUMMARY OF FINDINGS


      Data from  over  800 smoke  release  experiments were analysed  to describe
the   removal   mechanisms  and   dispersion  affecting   particulate   emissions
occurring within  surface mine pits.  An escape velocity, essentially  the  net
upward velocity within each pit, was  computed  from the observed retention time
of the tracers and  the depth  of  each  pit.   This upward velocity, when compared

to the downward  settling  and  deposition  velocity for different size particles,

was the  basis for  the calculation  of an escape fraction —  the percentage of

particulate  emissions  expected  to  escape  the  mine  pit.  Independently,  the

source depletion  equation was used, in  conjunction with wind  measurements made

in the mine pits, to compute  escape fraction.   These computed escape fractions

were  then compared to escape fractions computed  using  methodologies proposed

by Fabrick (1982) and Winges  (1981).   In addition, meteorological measurements

performed  inside  the  mine  pits  were   compared  to  simultaneous  out  of  pit

measurements.  Finally,  observed movements  of  the smoke  tracer plumes  were

categorized  and  then compared  to the  meteorological conditions occurring  at

the same time.


      The  following conclusions are  presented  from  the  findings  of  these
analyses:


      •   Computed  escape  velocities  and  escape  fractions  are  lowest  during
          night-time,  stable  atmospheres,  and  during light wind  speeds.   This
          finding  is  in  agreement with observed  flow patterns  in the  mine
          pits,  as  the  released smoke  tracer  was frequently  observed  to  have
          stagnated within the mine pits during these conditions.  Conversely,
          the greatest ventilation rates were  observed  during high  wind  speeds
          and near neutral atmospheres.

      •   The  computed  escape velocity  was  found to  be positively correlated
          with  measured   wind  speed,   temperature,  and  wind  direction,  and
          negatively correlated  with  stability  category, and  the width  of  the
          mine  pit.  However,  when  these parameters  were  used in   linear,
          multivariate  regression  analysis,   only  32%  of   the  variation  in
          escape  velocity values  could  be accounted  for.   The linear  model
          could  not be   improved  upon through  the use  of  in-pit measurements
          rather  than out-of-pit measurements,  or by  stratifying the data  by
          mine,  by  stability  class,  or  wind  speed category.    It  is  concluded
          that  some  processes  or   variables,   not   accounted  for  in  this
          analysis,  must act in  conjunction  with  the above  meteorological
          parameters to  determine the  escape velocity.
                                       61 -

-------
In-pit winds are significantly different from  out-of-pit  winds.   The
in-pit wind direction differs  from the out-of-pit wind direction by
about  60°.   Further,   no  correlation  between  the  in-pit  versus
out-of-pit  wind   direction  was  found   using  linear   regression
techniques, hence  the   in-pit  wind direction  can not accurately be
predicted from a knowledge  of  the  out-of-pit direction.   In-pit  wind
speeds  are,  on  the  average,  25%  smaller  than  out-of-pit  wind
speeds.   Linear  regression  analysis  did  identify  a  significant
positive correlation between in-pit and out-of-pit wind speeds.
                              62

-------
7.0   RECOMMENDATIONS FOR FUTURE WORK

      It appears  that  there are  two  data needs that must  be  addressed  in the
future to gain a  better  understanding  of  the pit retention phenomenon.  First,
there is clearly  a  need  to  quantify the magnitude of pit retention with direct
measurements of concentration,  or particle flux, so that a data base suitable
for evaluating the  performance of pit  retention  algorithms will be available.
This  direct  measurement would  also provide  an estimate of  the range  of  pit
retention escape  fractions, and should  give  EPA hard data with which to answer
policy questions:

      -   Does the  magnitude  of  pit retention  warrant  corrections  to existing
          models?
          Will pit  retention affect PM-10  concentrations?
          Is  the   fluctuation  in  escape  fractions  within  the  error  band
          ("noise") of particulate emission factors?
      At the  same time,  there is a need  to  continue to  identify  and quantify
the parameters that influence  pit dispersion.  Without  an understanding  of the
dispersion,  transport,  and  removal mechanisms  that  affect  surface  mine pits,
there is little hope of  simulating them.

      With these  needs  in mind,  the authors  offer  a series of recommendations
for future study:

      •  MODEL COMPARISON.  A  very  simple and inexpensive investigation  can be
performed  to  determine  a  "ballpark"  magnitude  of  pit  retention.   Using
existing hi-vol data and meteorological data already collected in the vicinity
of  surface mines, a comparison can be  made  of  actual  measured  concentrations
just  downwind of  a  pit  (cmeasured),  and  modeled concentrations  determined
from  the  ISCST   model  (cmodeled)> which idealizes  the terrain  as  flat  and
unaffected by the presence  of  the pit.   Emission rates  would be estimated from
AP-42,  Supplement  14  fugitive dust  factors, and a  representative background
concentration (perhaps from an upwind hi-vol) would be subtracted from the
                                       63

-------
measured   concentrations.    Any  departure   in  the   value  of   (C       ,/
                                                                      measured
C   , ,  ,)  from 1.0  would be  due  to  errors in  the emission  factors, or  to
errors  in the  model.    If  a  long  time  period is  considered —  perhaps  by
examining annual average concentrations — then random  errors  in  the  model and
emission  factors  will  cancel  out.   Differences  in the  value of  (C        /
                                                                      measured
C   , ,  ,)  from  unity  would   be   due  to   systematic   errors,   such  as  pit
retention  or  plume  perturbation  caused by   the  pit.   In  the  absence  of
systematic errors in the  emission factors or in idealizing the dust plume, the
ratio  of  (cmeasured/C  modeled^ would  be Just equal  to  the escape  fraction
for  the particle size  distribution collected  by  the  hi-vols.   This  approach
would  be  a  "first-cut"  at  estimating  the magnitude  of  pit retention.   Of
course,  it  would offer no insight  into  the  physical  mechanisms  that  control
dispersion  from the pit.   A  study  of  this sort,  using existing  data,  would
cost from £10,000 to $20,000.

      o   PARTICULATE MEASUREMENT PROGEAM.   A  logical  extension  of  the  MODFL
COMPARISON  investigation  just  described  would  be  to  measure  particulate
concentrations  at  the  downwind  edge  of a  mine  pit,  while measurements  of
meteorological  variables are being made.  As  before,  the ratio of measured and
modeled  concentrations  would provide a  measure  of  pit  retention.   However, the
availability of  detailed  meteorological data, especially stability class,  wil]
allow   an  examination   of   relationships  between   pit  escape  fraction  and
meteorology.

      The  length  of each test, or  episode,  must be  at  least 15 minutes,  which
is  the  minimum  time needed  to  approach  Guassian distributions  and  therefore
derive  pit  retention algorithms for the existing  Gaussian models.  The maximum
time  duration  for  each  test will be dictated  by  how much  particulate matter
must  be collected by the samplers  to  provide  reliable  concentrations.  Use of
filter  samplers  suspended from a  tethered balloon,  and  use of a quartz crystal
microbalance,  as suggested by  Air  Sciences  (Hittman and  Air  Sciences, 1983),
should  be considered.   The cost for these tests would  be  between $5100,000 and
$150,000.
                                        64

-------
      •  PARTICULATE TRACER  RELEASE.   Both of the studies  suggested  above are
unable to differentiate  between  systematic errors in emission  factors  and the
effect of  pit  retention.   If,  for  example,  the  ratio of cn,easured/CII,odeled
is  found  to  be  0.80,   it   is  not  certain  whether  the   escape  fraction  of
particulate is  .80, or whether the  emission factors  are consistently low by 20
percent.   This  problem  can  be  overcome  by controlling the  emission  rate  by
using a tracer.

      The portion  of  the  mine dust particulate matter  that is  retained in the
pit can be determined by  employing a particulate tracer material  that  matches
the settling  velocities  of  the  mine  particulate.   This tracer  material  would
be  emitted  continuously  for  a  short  time,  on  the  order  of  an  hour,   at  a
controlled  rate  from  a  simulated source  in  the  pit.   Measurements  of  the
particle  budget outside  the  pit would  in  turn  establish  the fraction  that
escaped from  the  pit to  the ambient  atmosphere.   The difference  between the
escaped fraction and the  emitted material would then represent   the  portion
retained,  or the pit retention fraction.

      The mass  flux of  particulate tracer  material  downwind of the  pit  would
be  determined  by  measuring   the  concentration  field.   The  methodology  to
measure this  field of concentrations  includes  the  use of  several  towers  tall
enough to encompass the vertical extent  of the  tracer plume,  a broad  enough
array of  towers to encompass the plume laterally, and  with enough measurement
points to  enable  rigorous  reconstruction of  vertical  profiles of  the  tracer
plume.  With  these concentration  measurement points  in a  crosswind  vertical
array  a   specific  cross  sectional area  is  represented  by  each  measurement
point.  A budget  of  particles observed  can then be  determined from  a  cross
wind vertical integration of all these cross sectional areas.

      Requirements:
      •   Tracer  material  in  size range  to match  dust particulate  settling
          rate.
      •   Assay  of bulk tracer  material  to determine  mean mass diameter and
          particle  size  distribution  by  percent   of   particles  in   size
          categories.
      •   Assay   of  observed   concentration   times   with   particle   size
          distribution by percent  in  particle  size  categories at each  sampler
          location.
                                       65

-------
                               FIGURE  7.1

                     PARTICLE  LIDAR EXPERIMENT
Ambient
Wind
                                  66

-------
      The following observations and measurements would be made:
      PHYSICAL FACTORS
DETERMINE
Pit parameters
Tracer Source
location
MEASURE

Length
Width
Depth
Orientation
Location
Dissemination rate
ENVIRONMENTAL FACTORS

DETERMINE

Atmospheric Stability
out of pit
Average wind speed

Stability out of pit
                                         Stability in pit
                                         Vertical profile of
                                         Wind speed

                                         Concentration field
                                          (tracer)
  MEASURE

Cloud cover
Time of day
Wind speed
(out of pit)
Statistics of
wind speed
variability
Wind direction
(out of pit)
Statistics of
wind direction
variability
Wind speed
(in pit)
Statistics of
wind speed
variability
Wind direction
(in pit)
Statistics of
wind direction
variability
Wind speed at
several height
on one tower
Concentration
at several
heights on each
tower
Time on-off
each sampler
      The  tracer  material  to  be  employed   is  a  broad-band  particle  size
fluorescent  particle material.   This  material can  readily be  distinguished
from  dust when  assayed  under ultraviolet  radiation.   The bulk  fluorescent
particle  material  (FP)  would have to be  obtained in  a size range and width of

size  ranges  to duplicate  a typical  range  of  dust  settling rates.   That  is,_
since the FP  material is  normally of  specific  density in a different range
                                       67

-------
than  dust,   it  would  be  necessary  to match  the dust  and  FP  by  equivalent
settling rates.   Dust from mines can be expected  to be in  the specific  density
range near  2.0  while the most  commonly used FP  is  specific density 4.0.   If
the dust size range  can  be  expected to vary from 2  to 100 microns  then  FP in
the range from 1  to  50 microns  could be employed so  long as the  particle  shape
factor  is  roughly equivalent.   There  is  however, some  anticipation  that  FP
material cannot be obtained  in  a size  range extending as  large  as  50 microns.
A size  range  up  to  20 urn is more  likely.   Alternately fluorescent  dyed  glass
beads of  sizes 50  to 100  microns  with  a specific  density of  2.0 could  be
combined with the FP to fill out the full size  range  of concern.

      The specific information  that would be  derived from such an  experiment
would include the following:

      •   Escape  fraction and retained  fraction  in each  particle size category
          by  in-pit  and  out-of-pit  stability  class, wind  speed  class,  pit
          parameter.
      •   Lateral and  vertical  plume  dimension  at the  location of  exit  from
          pit,  categorized  by  in-pit  and out-of-pit  stability  class,   wind
          speed class, and pit parameters.
      The relative cost  of  performing particulate tracer experiments would be
considerably  higher  than  the   previous  experiment   discussed.   Particulate
tracer  work  is  quite  labor  intensive, in addition   to  the  field  measurement
area.  Due to the high cost of  the  manual  optical assay  techniques  the  overall
cost of such a program probably would exceed £300,000.

      •  PARTICULATE FLUX MEASUREMENT

      The final  level of complexity  and  rigor suggested here involves making
Lidar measurements of the actual dust  source  in  the  pit  and companion  I.idar
measurements  of   the  dust  plume  shortly  after   it  has  exited  from the  pit.
Several types of Lidar devices  are  sensitive to  dust  particles.   That is,  they
can  detect  the  presence of dust  particles and in turn  can determine relative"
concentrations  of  dust  particles.   The  CO-  Lidar  is  probably  the   most
sensitive to dust and has the added  advantage  of  being "Eye Safe"  beyond  a few
feet range.  The  Lidar devices have been developed to the point they are range
                                       68

-------
gated  and  thus  can  give mean  concentrations for  increments  of  range.   This
means  that one can obtain  an along-path concentration profile for each firing
of  the Lidar  beam.   The range  gating yields  a concentration  profile,  along
path, integrated based on mean concentrations for every  3  meters of Lidar beam
range.

      The  approach  to  be  employed with  this device  would be  to  locate  one
Lidar  in the pit to  measure  a crosswind vertical profile  of  the concentration
of  dust  generated  by the activities  of the mining operation.   A  second  Lidar
would  be located to  the side of  the  pit  so it  could  do a cross  wind  profile
just  downwind  of the pit.   This  second Lidar would  be  located  so  as  to scan
directly cross wind  at  a  constant azimuth angle  starting  with  a horizontal
position  and  make  successive   firings at  successively  increased  elevation
angles.  This  would  yield  a crosswind  vertical  profile  of dust  concentration.
Concurrent measurements  of  wind profiles  both  in-pit and out-of-pit would  be
needed to  enable calculations of out-of-pit dust particle  budgets which  would
be  compared  to the  in-pit budgets to  determine  pit retention.   A schematic  of
the experimental configuration is shown in Figure 7.1
                                       69

-------
      The following observations and measurements  would be  made:
      PHYSICAL FACTORS
DETERMINE
Pit parameters
Dust Sources
MEASURE

Length
Width
Depth
Orientation
Locations
Activity level
ENVIRONMENTAL FACTORS

DETERMINE

Atmospheric Stability
out of pit
Average wind speed

Stability out of pit
                                         Stability in pit
                                         Vertical profile of
                                         Wind speed (out  of
                                         pit)
                                         Dust profile (in pit)
                                         Dust profile (out of
                                         pit)
    MEASURE

    Cloud Cover
Time of day
Wind speed
(out of pit)
Statistics of
wind speed
variability
Wind direction
(out of pit)
Statistics of
wind direction
variability
Wind speed
(in pit)
Statistics of
wind speed
variability
Wind direction
(in pit)
Statistics of
wind direction
variability
Wind speed at
several height
on one tower
Lidar profile
Lidar profile
The cost  of  conducting  a Lidar field  measurement  program would be  relatively
high, most probably on  the  same order  as  the  tracer experiment,  that is,  about
£300,000.


      The advantage in  employing Lidar measurements  is  that  they would  provide
a direct measurement of  the dust behavior itself,  rather  than  a measurement  of

a  simulant  as a tracer  study  would.   The  disadvantage is that there would  be
no  information  generated  about   particle  size  distributions  unless  extra
observations were incorporated for that specific purpose.
                                       70

-------
      •  FEDEEAL HIGHWAY ADMINISTEATION STUDY

      As  explained  in  Section  2.0,  BACKGROUND  AND  LITERATURE  SURVEY,  the
Federal Highway Administration will fund  a  20 month long wind  tunnel  study of
airflow and  dispersion  in  street  canyons and deep  cuts.   One purpose  of  the
study will be to develop algorithms that  can  be  used  to improve predictions of
CO concentrations  in existing Gaussian models.   EPA may choose to  monitor this
study since  it  will likely offer  some insight  into  mine  pit  flows.  Or,  EPA
may  even  consider  funding  an   expansion   of   the   FHWA  study   to  include
investigation of surface mine pits.
                                       71

-------
                                   REFERENCES

AMS, 1977, "AMS Workshop on Stability Classification Schemes and Sigma Curves
      —   Summary   of   Recommendations",   appearing   in   Bulletin   American
      Meteorological Society, vol 58, no. 12, December 1977, pp.1305-1309

Bowne, N.E., et al, 1S82, "Overview, Results, and Conclusions for the EPRI
      Plume  Model   Validation  Project:  Plains  Site",   for Electric  Power
      Research Institute, Palo Alto, CA., Project 1616, November 1982.

Cole, C.F., and A.J. Fabrick, 1984 "Surface Mine Pit Retention", Journal Air
      Pollution Control Association, vol  34, no. 6, June 1984, pp. 674-675.

Cole, C.F., 1984, personal conversation with Rodger Steen, Air Sciences, Inc.,
      July 20, 1984.

Draxler, R.F., 1977, "A Mesoscale Transport and Diffusion Model", NCAA Tech.
      Memo, ERL-ARL-64, p. 31.

EPA,  1980, "Workbook for Estimating Visibility  Impairment",  OAQPS, contract
      68-02-3337, July 1980.

Fabrick, A.J., 1982, "Technical Note:  Calculation of the Effective Emissions
      from Mine  Pit Operations by Incorporating Particulate Deposition in the
      Excavated Pit", MEF Environmental,  Del Mar, CA, 1982.

Gresho, P.M., et al, 1976, "Modeling the  Plantetary Boundary Layer Using the
      Galerkin   Finite  Element  Model",   presented   at  Third  Symposium  on
      Atmospheric  Turbulence,  Diffusion,  and  Air  Quality, Raleigh,  NC,  Oct.
      26-29,  1976.

Hanna, S.R.,  G.A. Briggs, and R.P. Hosker,  1982, Handbook of Atmospheric
      Diffusion, DOE/TIC-11223, Technical Information Center.

Herwehe,  J.A., 1984, "Numerical Modeling  of  Turbulent Diffusion of Fugitive
      Dust  from  an  Idealized  Open Pit Mine",  masters  thesis,  Iowa   State
      University, Ames,  Iowa. 1984.

Herwehe,  J.A., 1984a,  private communication x/ith Clifford Cole, July 20, 1984.

Hittman and Air Sciences, 1983,  "Studies  Related to Retention of Airborne
      Particulates  in  Coal  Mine  Pits—Data  Collection Phase",  prepared for
      U.S. EPA, IEFL,  Cincinnati, Ohio,  contract #68-03-3037, August 1983.

Hogstrom, A.S., and U. Hogstrom,  1978,  "A Practical Method  for Determining
      Wind  Frequency  Distributions  for  the  Lowest  200  m.  from  Routine
      Meteorological  Data",  Journal  Applied Meteorology,  vol  17,  July   1978,
      pp. 942-953.

Irwin, oT.S.,  1980,  "Dispersion Estimate  Suggestion #8", metro from J.S. Irwin
      to  Regional Meteorologists, July 31,  1980.
                                    73

-------
Lapple, C.E. 1961, "Characteristics of Particles and Particle Dispersoids",
      laboratory  wall   chart   reprinted  from  Stanford   Fesearch  Institute
      Journal, Third Quarter, 1961, SPI International,  Menlo Park, CA.

Lavery, T.F., et al, 1982, "EPA Complex Terrain Model Development First Mile-
      stone   Report",   for  Environmental  Protection   Agency,   Environmental
      Research Laboratory, Fesearch Triangle Park,  NC,  PB82-231713, April 1982.

Martin, D.O., 1976, "Comment on the Change of Concentration Standard
      Deviations  with  Distance,"  Journal  Air  Pollution  Control Association,
      vol. 26, No. 2 (February 1976), pp. 145-146.

Mitchell, A.E., and K.O. Timbre, 1979, "Atmospheric Stability Class from
      Horizontal   Wind   Fluctuation,"   presented   at   72nd   meeting   APCA,
      Cincinnati, Ohio, June 24-29, paper no.  79-29.2,  pp. 16.

Nelli, J.P., et al 1983, "Analysis and Modeling of Air Quality at Street
      Intersections,"  Journal  Air Pollution Control Association,  vol  33,  no.
      B, August 1983, pp. 760-764.

Pasquill, F., 1974, Atmospheric Diffusion, 2nd ed., John Wiley and Sons, New
      York.

PEDCo & TRC, 1982, "Characterization of PM-10 and TSP Air Quality Around
      Western Surface  Coal Mines,"  prepared for EPA, Air Management Technology
      Branch, contract #68-02-3512, June 1982.

Shearer, D.L., et al, 1981,  "Coal Mining Emission Factor Development Study,"
      prepared by TEC  Environmental Consultants, Inc., 0908-D10-15, Englewood,
      CO July 1981.

Turner, D.B., 1970, "Workbook of Atmospheric Dispersion Estimates," EPA,
      OAQPS, AP-26, 1970.

Van der Hoven, I., 1968,  "Deposition of Particles and Gases," appearing in
      Meteorology  and  Atomic   Energy  1968,  ed.   D.H.  Slade,  Technical
      Information Center, U.S. DCE, TID-24190, July 1968.

Wedding, J. B., et al, 1977, "A Wind Tunnel Study of Gaseous Pollutants in
      City  Street Canyons,"  Journal Air  Pollution Control Association,  vol.
      27, p. 557.

Winges, K.D., 1981, "Description of the EFTEC Mining Air Quality Model," ERTEC
      Northwest,  Inc., Seattle, WA  1981.

Zamars, J., and R. Piracci,  1982,  "Modeling of Carbon Monoxide Hot  Spots,"
      Journal  Air Pollution Control Association,  vol.  32, no.  9,  Sept. 1982,
       pp. 947-953.
                                     74

-------
APPENDIX 1

-------
                                  BIBLIOGPAPHY

Chan, S.T., "Numerical Simulations of Atmospheric Releases of Heavy Gases over
      Variable  Terrain",   Lawrence  Livermore  Laboratory,  UCPL-87256,  August
      1982.

Chan, S.T., "FEM3—A Finite Element Model for the Simulation of Heavy Gas Dis-
      persion  and  Incompressible  Flow  User's  Manual",  Lawrence  Livermore
      National Laboratory, UCPL-53397, February 1983.

Cole, C.F., and A.J. Fabrick, "Surface Mine Pit Petention", Journal Air
      Pollution Control Association, vol 34, no. 6, June 1984s pp. 674-675.

Fabrick, A.J., "Technical Note: Calculation of the Effective Emissions from
      Mine  Pit  Operations  by  Incorporating   Particulate reposition  in  the
      Excavated Pit", MEF Environmental, Del Mar, CA, 1982.

Gresho,  P.M., et al, "Modeling the Plantetary Boundary Layer Using the Galerkin
      Finite  Element  Model",   presented   at  Third  Symposium  on  Atmospheric
      Turbulence, Diffusion, and Air Quality, Raleigh, NC, Oct. 26-29, 1976.

Herwehe, J.A., "Numerical Modeling of Turbulent Diffusion of Fugitive Dust
      from an Idealized Open Pit  Mine", masters thesis,  Iowa State University,
      Ames, Iowa.  1984.

Hoydysh W. G. and Ogawa "Characteristics of Wind Turbulence and Pollutant
      Concentration  in and  Above a  Model City."   Report No.  EEPL NYU 110,
      Environmental  Engineering  Research  Laboratories,  New  York  University,
      Bronx, New York.

Hoydysh W.G. and Ogawa "A Two-Dimensional Study of Dispersion of Automotive
      Pollution  in  Street  Canyons",  Report  No.  EEPL  NYU 111,  Environmental
      Engineering Research Laboratories, New York University, Bronx, New York.

Johnson, W.B., et al, "Field Study for Initial Evaluation of an Urban
      Diffusion  Model  for Carbon  Monoxide",  Stanford Research  Institute,  SPI
      Project 8563, June 1971.

Lee, R.L., et al, "A Modified Finite Element Model For Application to Terrain-
      Induced  Mesoscale   Flows",   Lawrence  Livermore   National  Laboratory,
      UCRL-88033, November 1982.

Lee, R.L. and J.M. Leone, Jr.,  "Numerical Calculations of Stratified Ekman
      Layer Flow Over  Ridges with a Finite  Element  Model",  Lawrence Livermore
      National Laboratory, UCRL-90668, June 1984.

Ludwig, F.L., and W.F. Dabberdt, "Evaluation of the APPAC-1A Urban Diffusion'
      Model for Carbon Monoxide", SRI Project 8563, February 1972.

Mehta, U.B., and Z. Lavan, "Flow in a Two-Dimensional Channel with a Rectangu-
      lar Cavity", Transactions of the ASME, Dec. 1969, pp.897-901.
                                      A-l

-------
Nelli, J.P., et al, "Analysis and Modeling of Air Quality at Street Inter-
      sections",  Journal  Air  Pollution Control  Association,  vol  2?,  no.  8,
      August 1983, pp. 760-764.

Pan, F.,  and Acrivos,  A.,  "Steady Flows in Rectangular Cavities", Journal
      of Fluid Mechanics,  Vol. 28, 1967, pp. 643-655.

Wedding,  J.B., et al "A Wind-Tunnel Study of Gasecus Pollutants in City Street
      Canyons", Journal Air Pollution Control Association, vol. 27, p. 557.

Winges, K.D., "Description of the ERTEC Mining Air Quality Model", EETEC
      Northwest, Inc., Seattle, WA, 1981.

Zarcars, J.,  and R. Piracci, "Modeling of Carbon Monoxide Hot Spots", Journal
      Air  Pollution  Control Association,  vol 32,  no.  9, September  1982, pp.
      947-953.
                                        A-2

-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-450/4-85-001
4. TITLE AND SUBTITLE
Dispersion of Airborne Particulates In Surface Coal
Mines, Data Analysis
7. AUTHOR(S)
9. PERFORMING ORGANIZATION NAME AND ADDRESS
TRC Environmental Consultants, Inc.
7002 South Revere Parkway, Suite 6
Englewood, Colorado 80112
12. SPONSORING AGENCY NAME AND ADDRESS
Monitoring and Data Analysis Division
Office of Air Quality Planning and Standards
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
January 1985
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT l>
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3514
13. TYPE OF REPORT AND PERIOD COVERE
14. SPONSORING AGENCY CODE
EPA/200/04
15. SUPPLEMENTARY NOTES
16. ABSTRACT

       This  report  summarizes  the  results of an effort to better understand the
  dispersion  and  transport  of particulate matter released within surface coal mine pits
  Data previously collected at  four surface coal mines were used in this investigation.
  This report describes  the analysis and interpretation of those data, examines the
  relationship between meteorology  and smoke puff behavior, and compares mine pit escap
  fraction  (that  portion of the dust emitted in the pit that leaves the pit) with those
  predicted by existing  equations.

       Two independent  techniques  were used in conjunction with assumed particle size
  distributions and  the  onsite  data, to infer values of escape fraction.  These values
  were then used  to  determine the predictive ability of two widely used model algorithm
  The report  contains numerous  tabulations and discusses the relative merits of each
  method.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Air Pollution
Coal Mining Emissions
Particulates - Escape Fraction
Meteorology
18 DISTRIBUTION STATEMENT
Unlimited
b. IDENTIFIERS/OPEN ENDED TERMS

19. SECURITY CLASS (This Report)
Unclassified
20 SECURITY CLASS /This page)
Unclassified
c. COSATI F;ield/Group

21. NO. OF PAGES
81
22. PRICE
SPA Form 2220-1 (Rev. 4-77)
                     PREVIOUS EDITION ' s OBSOLETE:

-------
                                                        INSTRUCTIONS

   1.   REPORT NUMBER
       Insert the EPA report number as it appears on the cover of the publication.

   2.   LEAVE BLANK

   3.   RECIPIENTS ACCESSION NUMBER
       Reserved for use by each report recipient.

       TITLE AND SUBTITLE
       "itle should indicate  clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller
         ,ie or otherwise subordinate it to main title. When a report is prepared in more than one volume, repeat the primary title, add volume
         mber and include subtitle for the specific title.

   5.   REPORT DATE
       Each report shall carry a date indicating at least  month and year.  Indicate the basis on which it was selected (e.g., date of issue, date of
       .:pi  oval, date of preparation, etc.).
   •
   6.   PERFORMING ORGANIZATION CODE
       Leave blank.

   7.   AUTHOR(S)
       Give name(s) in conventional  order (John R. Doe, J. Robert Doe, etc.).  List author's affiliation if it differs from the performing organ!
       zation.

   8.   PERFORMING ORGANIZATION REPORT NUMBER
       Insert if performing organization wishes to assign this number.

   9.   PERFORMING ORGANIZATION NAME AND ADDRESS
       Give name, street, city, state,  and ZIP code.  List no more  than two levels of an organizational hirearchy.

   10.  PROGRAM ELEMENT NUMBER
       Use the program element number under which the report was prepared.  Subordinate numbers may be included in parentheses.

   11.  CONTRACT/GRANT NUMBER
       Insert contract or grant number under which report was prepared.

   12.  SPONSORING AGENCY NAME AND ADDRESS
       Include ZIP code.

   13.  TYPE OF REPORT AND PERIOD COVERED
       Indicate interim final, etc., and if applicable, dates covered.

   14.  SPONSORING AGENCY CODE
       Insert appropriate code.

   15.  SUPPLEMENTARY NOTES
       Enter information not included elsewhere but useful,  such as:  Prepared  in cooperation with, Translation of, Presented'at conference of,
       To be published in, Supersedes, Supplements, etc.

   16.  ABSTRACT
       Include a brief (200 words or less) factual summary of the most significant information contained in the report.  If the report Contains a
       significant bibliography or literature survey, mention it here.

   17.  KEY WORDS AND DOCUMENT ANALYSIS
       (a) DESCRIPTORS -  Select from the Thesaurus  of Engineering and Scientific Terms the proper authorized terms that identify the major
       concept of the research and are sufficiently specific and precise to be used as index entries for cataloging.

       (b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, equipment designators,  etc. Use open-
       ended terms written in descriptor form for those subjects for which no descriptor exists.

       (c) COSATI FIELD GROUP - Field and group assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
       jority of documents are multidisciplinary in nature, the Primary Field/Group assignment(s) will be specific discipline, area of human
       endeavor, or type of physical  object.  The application(s) will be cross-referenced with secondary Field/Group assignments that will follow
       the primary posting(s).

   18.  DISTRIBUTION STATEMENT
       Denote releasability to the public or limitation for reasons other than  security for example "Release Unlimited." Cite any availability to
       the public, with address and price.

   19. & 20. SECURITY CLASSIFICATION
       DO NOT submit classified reports to  the National Technical Information service.

   21.  NUMBER OF PAGES
       Insert the total number of pages, including this one and  unnumbered pages, but exclude distribution list, if any.

   22.  PRICE
       Insert the price set by the National Technical Information  Service or the Government Printing Office, if known.
EPA Form 2220-1  (Rev. 4-77) (Reverse)

-------