Ohio Highway Accident Interstate 70

1
Ohio Highway Accident - Interstate 70

• February 7, 2000
• Smoke Simulation Analysis
• Gary L. Achtemeier
• USDA Forest Service
• Athens, GA
• 11 June 2003

2
Task Description

• The following set of slides summarizes a
  PB-Piedmont analysis of smoke movement from the
  burning of a pile of Christmas trees at the
  Village of Lewisburg, Ohio, between 800 AM and
  300 PM on 7 February 2000 at which time the
  brush fire was thought to have been extinguished.
  It is alleged that residual smoke from this burn
  continued through the night and crossed
  Interstate 70 about one-half mile south of the
  burn site. This smoke, in addition to local dense
  fog, was implicated as a factor in a vehicle
  pileup on I-70 at approximately 1055 PM on 7
  February.

3
About PB-Piedmont
PB-Piedmont is an experimental model. PB-Piedmont
DOES NOT model combustion rates nor smoke
emissions. PB-Piedmont assumes that residual
smoke of some unknown level of concentration is
being discharged from a particular site.
PB-Piedmont proceeds under the assumption that
Given that smoke is emanating from a site, what
is the likely location of that smoke at a given
time within an entrapping layer of cooled air
near the ground at night as it moves over complex
terrain under varying synoptic weather
conditions.
4
About PB-Piedmont
As regards the Ohio I-70 accident on 7 February
2000, and the interpretation of the results to
follow no claim is made that smoke was being
generated at the burn site and, if smoke was
being generated at the burn site, no claim is
made regarding the emission rate of combustion or
whether resulting concentrations were sufficient
to significantly impair visibility. PB-Piedmont
tells us that, if smoke was being generated at
the burn site, where the most likely destination
will be.
5
Weather

• Weather during the burn and the post-burn smoke
  movement period was dry with temperatures near
  40F (4.4C) in early afternoon falling into the
  mid 20s (-3C) by 11 PM. Winds were blowing from
  the west at noon LST. Winds shifted to blow from
  the northwest during the afternoon, from the
  north by evening and from the north-northeast by
  1055 PM the time of the accident.

6
Modeling Assumptions

• Nocturnal boundary layer drainage layer 10 m
  deep decouples from the free atmosphere at
  sunset.
• Drainage layer cools at rate of 5K/30min after
  sunset unless cooling rate modulated by
  cloudiness.
• Maximum lapse rate of temperature within the
  drainage layer cannot exceed 5K/30 m.
• Maximum lapse rate of temperature within the
  drainage layer decreases incrementally as airmass
  becomes saturated.
• Drainage winds form as weak internal pressure
  forces drive cooled air toward lower elevation.
• Synoptic boundary conditions flow through for
  winds, synoptic temperatures, dew point
  temperatures, and pressure.
• Governing equations time-dependent Navier-Stokes.

7
Explanation of Cooling Rate

• Cooling of the shallow layer of air near the
  ground is what drives drainage winds.
  Observations of drainage wind formation in the
  South reveal that the transition from well-mixed
  conditions with relatively uniform winds over an
  area to drainage conditions with local winds
  impacting air movement occurs rapidly just after
  sunset. Increasing the empirical cooling rate
  speeds up the process while decreasing the
  cooling rate slows it down. Nevertheless, within
  about 2 hours, results are essentially the same.
  The cooling rate for dry soils of the Piedmont is
  probably less than that for snow cover. However,
  after the transition period, the differences in
  the model simulations should be minor. Therefore,
  PB-Piedmont has been run with the Piedmont
  cooling rate.

8
Modeling Information

• PB-Piedmont Version 3.0-2003
• Grid spacing 60m
• Domain size 145x135 -gt 8.7x8.1 km
• Visual domain 1.7x1.3 km
• Weather data Hourly METAR
• Elevation data USGS 30 m DEM
• Bulk drag coefficient 0.0025

9
Explanation of 60-m Grid Spacing

• Elevation data are provided to PB-Piedmont on a
  dense grid with 30-m grid spacing. PB-Piedmont
  was run with a 30-m grid and a 60-m grid.
  Comparisons of the results showed no real
  differences except that the 30-m simulations take
  hours to run. This result was expected since the
  topography of the area is smoothly varying with
  few side streams with ravine-like features.
  Therefore, the results to follow were done with
  the 60-m grid.

10
This is a map of the Lewisburg area showing key
reference points for the analysis I-70, SR-503,
the Water Works and Twin Creek. The Xs mark
where location information was collected for
input into PB-Piedmont.
11
This is the full 8.7x8.1 km domain for
PB-Piedmont. It is included to show the terrain
features that will impact the model simulations.
The small red square identifies the location of
the burn site.
12
Two Simulations

• PB-Piedmont was run for two simulations. The
  first simulation was done with just the USGS 30-m
  DEM data. The second simulation was run with the
  elevation data modified to account for the I-70
  embankment.

13
PB-Piedmont observation points 0800 EST 16
January 2003
This is the viewing grid for the first
simulation. Blue lines identify key features for
the analysis. Click left mouse button to see
these features.
5 Light smoke (0811)
Depot Ln
6 Smoke (0820)
Burn Site
Clay St.
7 Smoke (0825)
Wind symbol Flag 5.0 kt Barb 1.0 kt Short
barb 0.5 kt
8 Smoke boundary (0832)
9 Smoke (0837)
Twin Creek
10 Smoke boundary (0856)
11 Light smoke (0904)
Model Smoke at 500 PM
SR-503
12 Smoke (0908)
13 Smoke boundary south (0920)
14 Smoke boundary (0954)
Shading intervals are 3 m elevation
15 No smoke (0800)
I-70
Click left mouse button to view observation
points.
14
The smoke plume at 700 PM EST. Cloudiness has
retarded drainage wind formation. Nevertheless,
the smoke is confined to the valley passing
across I-70 east of the Twin Creek bridge.
15
The smoke plume at 1100 PM EST the time of the
accident on I-70. Drainage flows have
concentrated the smoke to a narrow path west of
the Twin Creek bridge. White contours identify
lines of constant temperature. Fog conditions can
occur where the temperature is less than 23F.
16
SIMULATION TWO The second simulation uses
elevation modified for the I-70 embankment via
aerial photographs of Preble County. Note the
extension of high elevation into the Twin Creek
valley. Grid points are 60 m. Smoke plume at 500
PM.
17
SIMULATION TWO Smoke tends to dam up behind the
I-70 embankment. The plume becomes more diffuse
and spread out. Smoke crosses I-70 from the Twin
Creek Bridge westward. Smoke plume at 800 PM.
18
SIMULATION TWO Wind currents in the Twin Creek
valley become light and variable as PB-Piedmont
simulates weak eddies in the flow. At 900 PM,
wind speeds are near zero at the burn site. This
allows smoke to build up in high concentrations
locally.
19
SIMULATION TWO At 1000 PM, the patch of dense
smoke is observed moving southward toward I-70.
20
SIMULATION TWO At 1100 PM, the patch of dense
smoke is crossing I-70 from the Twin Creek Bridge
westward.
Accident Site at 1055 PM
21
Summary What PB-Piedmont tells us

• Synoptic forcing with winds blowing generally
  from the north combined with drainage forcing in
  a stream valley oriented north-south to confine
  the winds to blow southward from the burn site to
  I-70.
• SIMULATION ONE produced a narrow and steady smoke
  plume that crossed I-70 west of the Twin Creek
  bridge at 1100 PM just after the reported time
  of the accident.
• SIMULATION TWO, with the obstacle of the I-70
  embankment, produced a broader, unsteady smoke
  plume that crossed I-70 from the Twin Creek
  bridge westward.

22
Summary Point of Clarification about SIMULATION
TWO

• PB-Piedmont produced patches of dense smoke as
  very weak valley flow behind the I-70 embankment
  broke up into eddy circulations.
• Although the formation of dense patches of smoke
  in weak flow may be physically realistic, the
  timing of the formation and subsequent movement
  down-valley is not predictable.
• Nothing can be claimed regarding the timing of
  arrival of the dense smoke at I-70 other than it
  is possible that dense smoke/fog (superfog) could
  have contributed to accident conditions at a
  particular time.

23
Final Comments The PB-Piedmont analysis suggests
that a number of factors combined to produce
hazardous visibility at the accident site on
I-70. These are

• Winds blowing from the west during the active
  part of the burn but shifting to blow from the
  north toward I-70 after dark.
• A north-south oriented stream valley that aided
  in channeling residual smoke from the burn site
  to I-70.
• A relatively moist ambient airmass that would
  support formation of valley fog by 1100 PM.
• Obstruction to the valley winds presented by the
  I-70 embankment which dammed up the smoke-filled
  air and weakened valley winds thus allowing for
  residual smoke to build up in higher
  concentrations.
• Doing the burn in a valley bottom which increases
  the likelihood of smoke entrapment at night.
• Extinguishing the fire with water thus creating
  smoldering and cooling the combustion to allow
  the smoke to become entrapped in the drainage
  layer near the ground.